WHICH VEHICLES HAVE A HIGHER PROBABILITY OF FAILING A CALIFORNIA SMOG CHECK INSPECTION PRIMARILY CONSISTING OF A DIAGNOSTIC SCAN OF THE VEHICLE’S ON-BOARD COMPUTER SYSTEM? A Thesis Presented to the faculty of the Department of Public Policy and Administration California State University, Sacramento Submitted in partial satisfaction of the requirements for the degree of MASTER OF PUBLIC POLICY AND ADMINISTRATION by William Dean Thomas SPRING 2014 © 2014 William Dean Thomas ALL RIGHTS RESERVED ii WHICH VEHICLES HAVE A HIGHER PROBABILITY OF FAILING A CALIFORNIA SMOG CHECK INSPECTION PRIMARILY CONSISTING OF A DIAGNOSTIC SCAN OF THE VEHICLE’S ON-BOARD COMPUTER SYSTEM? A Thesis by William Dean Thomas Approved by: __________________________________, Committee Chair Su Jin Jez, Ph.D. __________________________________, Second Reader Robert Wassmer, Ph.D. ____________________________ Date iii Student: William Dean Thomas I certify that this student has met the requirements for format contained in the University format manual, and that this thesis is suitable for shelving in the Library and credit is to be awarded for the thesis. __________________________, Department Chair Robert Wassmer, Ph.D. Department of Public Policy and Administration iv __________________ Date Abstract of WHICH VEHICLES HAVE A HIGHER PROBABILITY OF FAILING A CALIFORNIA SMOG CHECK INSPECTION PRIMARILY CONSISTING OF A DIAGNOSTIC SCAN OF THE VEHICLE’S ON-BOARD COMPUTER SYSTEM? by William Dean Thomas An important component of California’s smog check program is the policy of directing the highest polluting vehicles to specialized smog check stations for testing. The State Bureau of Automotive Repair identifies these ‘directed vehicles’ through the use of a regression model that identifies the potentially highest polluting vehicles based upon past tailpipe emissions readings of the same type of vehicles. However, beginning in 2014, the testing procedure for a large portion of the vehicles in California will no longer include a tailpipe emissions measurement. The revised testing procedure will rely upon a scan of the on-board computer diagnostic and control system that controls and constantly evaluates the function of the engine and emissions control systems present on most vehicles manufactured since 1996. The revised procedure will also include a visual inspection of the emissions control devices present on the vehicle. Consequently, there is a need for a regression model capable of identifying the vehicles with the highest likelihood of failure based upon the results of the scan of the diagnostic system and the visual inspection. In this thesis, I developed a binomial logistic regression model that predicts which vehicles are highly likely to fail the vehicle computer diagnostic scan or visual inspection procedure comprising the new inspection procedure. The regression analyses described herein accurately v identified a group of approximately 40% of the vehicles subject to smog check inspections that have a higher likelihood of failure than the remaining vehicles subject to testing. Implementation of the regression models described in this thesis, or similar models, will enable the Bureau to continue to identify approximately 30% of the fleet of vehicles as directed vehicles. _______________________, Committee Chair Su Jin Jez, Ph.D. _______________________ Date vi ACKNOWLEDGEMENTS I would like to thank Dr. Jez and Dr. Wassmer for their guidance and support in this endeavor. Their always-accurate suggestions were immensely important in keeping this project moving in the right direction and helping to achieve final product. Additionally, I cannot write an acknowledgements page without recognizing the support provided by my family. Numerous times over the last months, I have been physically present at family functions without mental presence because of my focus on which regression approach works best or how to best clean 8 million observations. Moving forward, they will no longer hear “I can’t go on that camping trip because of the thesis” or other missed functions. Finally, I must acknowledge the ever-present love and support of my wife, Suzanne. She too has taken a back seat to this thesis for several months, an action I will happily spend the next several months making up for. Thank You All! vii TABLE OF CONTENTS Page Acknowledgements ................................................................................................................ vii List of Tables ............................................................................................................................ x Chapter 1. INTRODUCTION .............................................................................................................. 1 Origins of the Smog Check Inspection Program ......................................................... 1 The Creation of Test-Only Stations ............................................................................ 4 Directed Vehicles ......................................................................................................... 5 Smog Check Inspection Procedure .............................................................................. 7 Further Program Analysis ............................................................................................ 8 Recommendations for Program Improvement ............................................................. 9 Modern Vehicle Technology ..................................................................................... 11 Research Question ..................................................................................................... 12 Organization of Remainder of Paper ......................................................................... 12 2. LITERATURE REVIEW .................................................................................................. 14 Smog Check Overview ............................................................................................... 14 Modeling to Predict Emissions Failures ..................................................................... 16 On-Board Diagnostics, Generation Two (OBD II) Systems ....................................... 19 Empirical Studies of OBD II Systems ........................................................................ 21 Conclusion .................................................................................................................. 25 3. METHODOLOGY ............................................................................................................ 26 Smog Check Test Data ............................................................................................... 27 viii Regression Models: OBD II Test Failure ................................................................... 28 Regression Models: Visual Inspection Failure ........................................................... 34 4. ANALYSIS ........................................................................................................................ 36 Predicting OBD II Test Failures ................................................................................. 37 Predicting Visual Inspection Failures ......................................................................... 43 Reliability and Validity ............................................................................................... 48 5. CONCLUSION .................................................................................................................. 53 Predicting Likelihood of Failure of the OBD II Functional Test ................................ 54 Predicting Likelihood of Failure of the Visual Inspection .......................................... 56 Significance of Findings ............................................................................................. 57 Recommendations for Identifying Directed Vehicles ................................................. 58 Suggestions for Improved Statistical Modeling .......................................................... 60 Conclusion .................................................................................................................. 60 Appendix A: Test Volume for Pre-Model-Year 2000 Vehicles .............................................. 62 Appendix B: Distribution of Vehicle Look-Up Table Identification Numbers. ...................... 63 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test, with Basic Vehicle Descriptions ....................................................................................... 102 Appendix D: VLT ID Numbers Demonstrating Zero Failures of the Visual Inspection, with Basic Vehicle Descriptions. ............................................................................ 138 Appendix E: VLT ID Numbers Most Likely to Fail the Visual Inspection, with Basic Vehicle Descriptions ............................................................................. 144 References ............................................................................................................................. 149 ix LIST OF TABLES Tables Page 1. Attrition rate of older vehicles .........................................…………………………..…….7 2. Distribution of Failing Vehicles in Current Inspection Procedure . ... ………………………26 3. Distribution of Vehicle Manufacturers within the dataset. ...... ……………………………33 4. Distribution of Vehicle Model-Years within the dataset ........... …………………………. 34 5. List of Vehicles Demonstrating Zero OBD II Functional Test Failures…………………….38 6. Results of Regression Model with Vehicle Make as the Explanatory Variables …………….41 7. Odds Ratios for Vehicle Model-Years ..................................... ………………………….43 8. Results of Regression Model with Vehicle Make as the Explanatory Variables…………….46 9. Results of Regression Model with Model-Year as the Explanatory Variables……………....48 10. Detail of Chi-Squared Values for Regression Model ….……… ………………………….51 x 1 Chapter 1 INTRODUCTION Nearly thirty years ago, the State of California began a program of biennial smog check inspections on most motor vehicles registered and operated in the state. Since that time, we have become accustomed to having to get a smog check every other year when we renew our registration or when we transfer ownership of our vehicle to another Californian in a private party transaction. What many people may not be aware of are the various statutes governing the implementation of the program, including designating the Bureau of Automotive Repair (hereinafter referred to as the Bureau) as the oversight and regulatory enforcement entity for the program, appropriately referred to as the “Smog Check Inspection Program”. An on-going component of the Smog Check Inspection Program is the requirement to identify vehicles highly likely to fail the inspection. Advances in automotive technology and program changes often require adjustments to the model used to identify these vehicles. This thesis proposes new methodologies for identifying these vehicles and details the analyses conducted to validate these methodologies. The following sections of this chapter include a description of the smog check inspection program, the origin of the program, general and specific requirements of the program, and past and future changes to the program. There is also a discussion of the methodologies used for identifying vehicles highly likely to fail a smog check inspection, the policies behind those methodologies, and the need for revisions to the methodologies. Chapter 1concludes with a summary of the remaining chapters in this thesis. Origins of the Smog Check Inspection Program California led the way in the automotive pollution control realm by implementing a program to identify high polluting vehicles in the mid 1960’s (Eisinger, 2010). This initial smog 2 check program consisted of nothing more than a basic visual inspection of the basic emission control components common to vehicles produced at that time. Throughout the 1970’s, while California continued to rely upon a basic visual inspection, other states began implementing what were referred to as “inspection and maintenance programs”, commonly referred to as I/M programs, which actually measured vehicle emissions rather than simply conducting a visual inspection. Eisinger (2010) describes California’s attempt at an I/M program conducted in the Los Angeles area between 1979 and 1984 as “unpopular and inconvenient” (p. 29), due to the amount of time required to conduct an actual emissions measurement compared to the mere minutes required for the visual inspection, which is what California motorists were accustomed to. Although the United States Environmental Protection Agency (EPA) pushed for expansion of the I/M program throughout the state as part of 1977 amendments to the Clean Air Act of 1970, California resisted. Elected officials and environmental policy makers were in a difficult situation; maintaining leadership in environmental policy would require inconveniencing millions of motorists. The federal government forced the issue when in December of 1980 they delayed the transfer of federal highway funds to California as a sanction for failing to comply with EPA mandates (Eisinger, 2010). In 1981 and in response to the withholding of highway funds, California State Senator Robert Presley proposed a new approach to emissions inspection programs in Senate Bill 33. Senate Bill 33 (Presley, Chapter 892, Statutes of 1982) mandated the implementation of a smog check inspection program in the State of California. This program of testing vehicles for compliance with emission control regulations by privately owned smog check stations began in March of 1984, and although changes to the program have occurred over the years, the program is still in place today. The purpose of the program is to identify high-polluting vehicles and require 3 the repair or retirement from operation in California of those vehicles. The program became necessary because of the failure of various regions within the state to meet United States Environmental Protection Agency (EPA) air quality standards (Eisinger, 2010). The program never achieved the expected results and after a 1987 study highlighted the extent of the failure of the program, the State implemented changes aimed at improving the smog check inspection program (Eisinger, 2010). However, the enhancements still failed to achieve the desired results and the Federal Clean Air Act amendments of 1990 enacted even more stringent air quality requirements, resulting in an even greater number of California regions placed in a ‘non-attainment’ status; indicating these regions fail to meet minimum air quality standards. California’s model of independently owned automotive repair facilities performing all but a very small portion of the smog check inspections in the state was inconsistent with the EPA’s desired model of governmentally contracted centralized inspection facilities (United States Environmental Protection Agency, 1992). USEPA (1992) policies demonstrated a preference for government controlled inspection facilities that are completely separate of repair functions, presumably with the intent to exert more regulatory oversight over the program and to dissuade fraud in the program. However, based upon the number of vehicles in the state and after the failure of the 1970’s centralized inspection program in the Los Angeles Area, California officials were certain the centralized inspection model would result in failure (Eisinger, 2010). Consequently, the Bureau, the Air Resources Board of the California Environmental Protection Agency (ARB), and the state legislature vigorously opposed the centralized testing model and pushed for alternative solutions. One of those solutions, eventually drafted into legislation and becoming law, expanded 4 the existing inspection program by creating a network of privately owned ‘test-only’ stations1 (California State Assembly, 1994). The Creation of Test Only Stations The test-only stations met the EPA’s goal of separation of testing and repair functions, as statute prohibits test-only stations from performing any type of repairs. As a solution to the aforementioned disagreement between the EPA and California over the model for smog check inspections, California proposed and the EPA approved the test-only station model (California State Assembly, 1994). Statute prohibits test-only stations from performing any type of automotive repairs other than testing services (Health and Safety Code, § 44014.5.(b)). In addition to limiting the types of services provided at test-only stations, statute also directs a certain portion of the vehicles required to receive a smog inspection to test-only stations. These vehicles, referred to as directed vehicles, account for approximately 30% of the vehicles tested each year and are statistically more likely to fail a smog check inspection (BAR, 2012). I discuss the method for identifying directed vehicles later in this chapter. Finally, regulations permit test-only stations to certify ‘gross polluting’ vehicles, which are vehicles exhibiting exceedingly high emissions levels. Regulations prohibit regular, non-test-only stations, referred to as “Test and Repair” stations, from certifying directed and/or gross polluting vehicles. The theory behind test-only stations was that by separating the repair and inspection functions of the smog check program, there would be no incentive for the test-only station to fraudulently certify a vehicle; thereby, ensuring the integrity of the smog check inspection program. The incentive for being a test-only station was the privilege of having a certain portion of the fleet directed to these stations (California State Assembly, 1994). Health and Safety Code section 44010 establishes the mechanism for privately owned ‘stations’ which shall be referred to as smog check stations and are authorized to certify vehicles pursuant to the Motor Vehicle Inspection Program established by the Code. Throughout this thesis, the terms station or stations are referring to smog check stations. 1 5 These test-only stations met California’s goal of continuing to allow privately owned facilities to perform practically all (a small portion, 0.45%, of inspections are performed at Bureau contracted referee stations) smog check inspections while meeting the EPA’s goal of separating inspection and repair functions (California State Assembly, 1994). The statutes authorizing privately owned test-only stations were contained in SB 521 (Presley, Chapter 29, Statutes of 1993) and became effective in March of 1994. Test-only stations first began operation in 1997 and eventually grew to comprise the significant portion of smog check stations, 34%, but performing a majority, nearly 65% of the approximate 12 million initial inspections annually (Bureau of Automotive Repair, 2012b). Directed Vehicles The practice of identifying and directing vehicles for testing at specific stations was also created by the smog check program amendments implemented on March 30, 1994, pursuant to Senate Bill 521 (Presley, Chapter 29, Statutes of 1993). As stated above, test-only stations receive the privilege of testing directed vehicles. Beginning in 1996 and continuing until 2012, the Bureau utilized a model referred to as the high-emitter profile (HEP) model (Bureau of Automotive Repair, 2012a and Choo, Shafizadeh, and Niemeier, 2007) to identify directed vehicles. Although the exact specification for the HEP model remains protected as intellectual property, it reportedly is a logistic regression model that predicts whether a vehicle is likely to generate high emissions based upon certain vehicle design variables (Choo, Shafizadeh, and Niemeier, 2007). In 2012, the Bureau reported that beginning in 2013, the primary factor for identifying directed vehicles would be the model-year of the vehicle (Bureau of Automotive Repair, 2012a). Specifically, all model-year 1976 through 1999 vehicles will receive directed vehicle status. However, these model-years are diminishing as a percentage of the total volume of tests 6 performed each year and the Bureau has indicated the volume of directed vehicles will remain consistent at the current level of approximately 30 percent of the total volume of tests each year (Bureau of Automotive Repair, 2012a). A concern with identifying directed vehicles by model-year is simple attrition. As vehicles age and owners replace them with newer vehicles, the number of older vehicles tested each year decreases. As an example, in 2011, model-year 1976 to 1999 vehicles accounted for just over 4.5 million, or 41.80% of the initial inspections performed; while in 2012, these same vehicles accounted for just over 4.0 million, or 35.54% of the initial inspections (Bureau of Automotive Repair, 2012b and 2013). Table 1 below highlights the attrition rate of older vehicles. Appendix A, “Test Volume for Pre-Model-Year 2000 Vehicles”, details the information presented in this graph. As the graph shows, it will only be a few years before the number of vehicles meeting these criteria is too small to provide a sufficient set of directed vehicles. Additionally, California Health and Safety Code section 44010.5 requires the Bureau to direct the vehicles with the highest probability of generating the highest emission levels. Although it is reasonable to assume that the oldest vehicles will generate the highest emissions, this is not a statistically valid approach as required by the code. Finally, beginning in 2014 and as a result of revisions to the smog check inspection procedure, the measurement of emission levels is longer a component of the smog check inspection. 7 Table 1 Attrition rate of older vehicles Percentage of Pre-2000 Vehicles 120% 100% 80% 60% Percentage of Pre-2000 Vehicles 40% 20% 0% 2001 02 03 04 05 06 07 08 09 10 11 12 Smog Check Inspection Procedure The current smog check inspection procedure consists of three components. First is a visual inspection wherein the technician visually inspects all emissions related to components to verify they are present on the vehicle, properly installed, and are free from any visual defects. The second portion of the inspection is the functional test. Depending on the model-year of the vehicle, one or more of the following functional tests will be conducted on the vehicle: LowPressure Fuel Evaporate Test (LPFET), which tests the fuel-evaporative control system (primarily the fuel-tank) for leaks; ignition timing test, which measures the ignition timing of the engine; fuel cap test, which is separate from the LPFET and tests the sealing integrity of the fuel cap; OnBoard Diagnostic System, Generation II (OBD II) test, which tests the vehicle’s engine and emissions control computer; and a visible smoke test, which as the name implies requires the technician to inspect for smoke emanating from the engine or exhaust of the vehicle. The final 8 portion of the inspection is the emissions measurement portion of the test. (Bureau of Automotive Repair, 2009) Two distinct methods are used for measuring vehicle emissions. The first is the twospeed idle (TSI) test which measures the emissions with the vehicle stopped and the engine running at 2500 revolutions-per-minute (rpm) and at idle speed, generally 600 to 800 rpm. The second came about as a response to the Federal Clean Air Act Amendments of 1990, and is an acceleration simulation mode test of vehicle emissions (Singer & Wenzel, 2003). The acceleration simulation mode (ASM) test places the vehicle on a treadmill-like device allowing the smog check technician to operate the vehicle under a load, simulating driving conditions; specifically vehicle acceleration at 15 miles per hour (mph) and 25 mph (Choo et al., 2007). Both tests are still utilized today, along with the visual inspection and a functional test of certain emissions control components (Bureau of Automotive Repair, 2009). The ASM test is utilized in the more populous areas of the state, referred to as “enhanced” areas (Bureau of Automotive Repair, 2009) and accounted for 76% of the tests performed in 2012 (Bureau of Automotive Repair, 2013) Further Program Analysis SB 521 also mandated that BAR perform random, roadside inspections of vehicles throughout the state to confirm compliance with vehicle emissions control laws and to obtain empirical data for smog check program analysis (Health and Safety Code §44024.5). An analysis of the roadside inspection data gathered between 2000 and 2002 revealed that in the category of 1974 to 1995 model year vehicles that had failed an initial smog check inspection, were repaired and subsequently certified as passing, 40% of those failed a roadside inspection within one year after certification (Austin, McClement, & Roeschen, 2009). 9 To confirm the accuracy of this analysis, Sierra Research conducted a study of 2003 to 2006 roadside inspection data and found that 1976 to 1995 model year vehicles were highly likely to fail a roadside inspection within one year after having failed an official smog check inspection, receiving repairs, and certified as passing; at a level of 49% (Austin, McClement, & Roeschen, 2009). The researchers hypothesized several possible reasons for these results and concluded that improper and poor smog check inspection procedures on the part of technicians is the primary cause of the high roadside inspection failure rates. Further research led to the conclusion that the test-only stations are the predominate source of improper testing procedures. Although it is unclear as to why this is the case, one hypothesis is that the expertise needed to only perform smog check inspections is far less than what is required to diagnose and repair vehicles, leading to poor performance among test-only technicians. Recommendations for Program Improvement The Sierra Research analysis concluded with several recommendations for improvements to the smog check inspection program. The primary recommendations are as follows: The establishment of performance standards designed to evaluate the inspection practices of technicians and stations Elevated monetary penalties, the ability to issue orders of abatement, and other penalty enhancements for those found to be improperly or fraudulently inspecting vehicles Modernized testing procedures for vehicles with advanced emissions control technologies A complete revision of the model for directing specific vehicles to test-only stations for inspection, so that only those stations demonstrating the highest performance in smog check inspections receive the privilege of testing directed vehicles. The recommendations and other improvements to California’s smog check inspection program were drafted into new legislation and enacted into law as California Assembly Bill 2289 10 (Eng, Chapter 258, Statutes of 2010). AB2289 directs the Bureau to implement a program for certifying smog check inspection stations based upon inspection performance and subsequently directing vehicles most likely to be high emitters and gross polluting vehicles only to those stations meeting the aforementioned performance standards. In addition, AB2289 directed the Bureau to implement a new testing model wherein most 2000 model-year and newer vehicles would receive an inspection that consisted only of a computer scan of the vehicles’ diagnostic system and a visual inspection of the other applicable emission components. This eliminates the emissions testing portion of the smog check inspection for these vehicles. This testing model is consistent with the USEPA’s current preferred inspection model and is already utilized in 31 states and jurisdictions throughout the United States (Environmental Protection Agency, 2013). Inspection procedures for 1999 model-year and older vehicles remain unchanged. (AB2289) In developing the performance standards as required by AB2289, the Bureau developed a regression model that determines the inspection effectiveness of technicians and stations based upon certain testing behaviors and the probability of a vehicle certified by a particular technician passing a subsequent smog check inspection. These performance measures encompass the Bureau’s ‘STAR” program, which became effective on January 1, 2013. The STAR program identifies smog check stations that meet the applicable performance criteria as ‘STAR Certified’, and affords these stations the opportunity to inspect and certify directed and gross polluting vehicles. The performance criteria encompass two sets of measures: short-term and long-term measures. The short-term measures evaluate behaviors such as bypassing a portion of the test or a high number of aborted tests, among other factors. The long-term measure looks at a vehicle certified by a station in the previous test-cycle, 18 to 30 months prior to the current test, and determines the probability of that vehicle passing. A high confidence level in the vehicle passing as it should results in a higher score, on a scale of 0.00 to 1.00 and results in a higher performance 11 rating for the station. Consequently, test-only stations are no longer able to ‘self-certify’ as able to inspect these vehicles and are only able to avail themselves of the increased revenue opportunity in testing directed and gross polluting vehicles by demonstrating acceptable inspection performance levels and becoming STAR certified. (BAR, 2012c) Modern Vehicle Technology The above statutorily required modernized testing procedures for the Smog Check program include the elimination of the acceleration simulation mode test for model-year 2000 and newer vehicles equipped with the on-board diagnostics, generation II (OBD II), system (Lyons and McCarthy, 2009). The OBD II system is a microprocessor based electronic control system that monitors and controls the engine operation and the emissions control systems in most lightduty vehicles manufactured since 1996 (Supnithadnaporn, Noonan, Samoylov, and Rodgers, 2011). Light –duty vehicles are passenger cars and light trucks under 14,000 pounds gross vehicle weight rating. This testing method is scheduled for implementation in late 2013 (Bureau of Automotive Repair, 2012d). Using various sensors, the OBD II system monitors the emissions control and engine management systems to determine if the potential exists for a malfunction that may result in elevated emissions (Lyons & McCarthy, 2009). If a malfunction is identified, the system may illuminate the vehicle’s ‘malfunction indicator lamp’ (commonly referred to as the ‘check engine light’) on the dash and store pertinent information related to the malfunction (Lyons & McCarthy, 2009). Lyons and McCarthy (2009) conducted research and determined that this system is quite accurate and an inspection system that scans and reports this information, and causes the vehicle to fail an emissions inspection as a result of a detected malfunction, is equally as effective as the acceleration simulation mode test. However, a major concern with the Lyons and McCarthy study is that their sample set consisted of only 74 vehicles, which when compared to the greater than 20 12 million light duty vehicles on California’s roadways is not statistically valid. Conversely, the EPA also considers the OBD II system to be quite reliable (Environmental Protection Agency, 2013). Research Question The program analysis recommending the use of the OBD II diagnostic system for smog check inspections and redesigning of the model for identifying directed vehicles create a critical need for a regression model capable of predicting which vehicles have a higher probability of failing the revised smog check inspection based solely upon the OBD II functional test results. This leads to the research question for this thesis: Which vehicles have a higher probability of failing a California smog check inspection primarily consisting of a diagnostic scan of the vehicle’s on-board computer system? To answer this question, I conducted a regression analysis to determine which vehicles are most likely to fail the OBD II functional test of the current smog check inspection. The OBD II functional test is equivalent to the new test procedure that begins in late 2013. Although the primary component of the new inspection procedure is a scan of the on-board diagnostic system of the vehicle, a visual inspection is also part of the procedure. To insure an accurate model for identifying all vehicles likely to fail the procedure, I developed and evaluated a second set of regression models aimed at predicting which vehicles would fail this portion of the new smog check inspection procedure. Organization of Remainder of Paper In Chapter 2, I review the available literature regarding on-board diagnostic systems and their use in smog check inspection programs. Other states are already using this methodology, and the results generated from their experiences provide useful background for the current research. Chapter 3 outlines the methodology employed in the conduct of this research, describes 13 the data collection process, and defines the dataset used in the regression analyses. I also describe the functional form of the regression model selected and discuss the statistical tests performed on the data to confirm the accuracy and significance of the results. Chapter 4 contains an in-depth discussion of the results of the regression analysis and proposes an improved model for identifying directed vehicles. In Chapter 5, I draw conclusions derived from the analysis, discuss the significance of the findings, and make recommendations for future studies and or policy initiatives, as appropriate. 14 Chapter 2 LITERATURE REVIEW Chapter 1 of this thesis provides an introduction to the smog check program in California, the policies associated with the program, and future direction of the program. These details served as the impetus for this thesis. In preparation for developing the regression models employed in the current research, I reviewed numerous articles related to the smog check program and automotive on-board diagnostic systems. Chapter 2 provides a summary of the articles and a progression of facts that provide the basis for the statistical analyses described later in this thesis. The first section of this chapter outlines a summary of the smog check program as described by the available literature. The next section discusses the various research projects previously completed that present approaches to statistical models aimed at predicting emission failures. Following is a section a literature that provides an overview of automotive and light truck on-board diagnostic systems or OBD II systems in industry vernacular. Finally, I discuss the literature reporting the results of various empirical studies of OBD II systems, their functionality, and their effectiveness at identifying emissions test failures. Smog Check Overview In response to EPA mandates for the State of California to comply with the Clean Air Act Amendments of 1990, the State updated its Smog Check program in 1997 and implemented the acceleration simulation mode (ASM) test described in the introduction to this paper (Eisinger, 2010). The ASM continues as the mandated smog check inspection procedure used in California. As with the original Smog Check inspection procedure, this updated test consists of the same three parts: an emissions measurement conducted at the tailpipe of the vehicle while under load, a visual inspection of the emissions components, and a functional test of certain emissions related devices or functions such as engine timing (Choo, Shafizadeh, and Niemeier, 2007). 15 The ASM test is a variation of the federal I/M 240 test, which is a 240 second driving sequence that simulates Los Angeles area rush hour traffic, that was in use in other jurisdictions (Eisinger, 2010). It is important to note that the development of the ASM test was a compromise between the EPA, which was pressuring for a centralized testing structure consisting of I/M 240 testing and California officials who felt a centralized testing structure would be ineffective in the State and I/M 240 testing leads to vehicle failures (Eisinger, 2010). This pattern of conflict and compromise between State regulators and the EPA has exemplified the Smog Check program since its inception. In the mid-1980’s, California implemented the Smog Check program, which mandated a biennial emissions test for most vehicles registered in the State (Singer & Wenzel, 2003). This original Smog Check inspection consisted of three parts: a visual inspection of the emissions control devices, a functional test of a small portion of the devices, and an emissions measurement of the vehicle’s exhaust (Choo, Shafizadeh, and Niemeier, 2007). The emissions measurement portion of the inspection measured the level of pollutants in the vehicle’s exhaust with the vehicle stationary and the engine running at or below 1000 revolutions per minute (rpm) and at 2500 rpm (Singer & Wenzel, 2003). This emissions measurement test is the "two-speed idle" test and has expanded to the current ASM test to meet changes in EPA requirements. Another of the compromises between the EPA and the State of California in creating the current smog check program was the implementation of test only stations and the direction of a portion of the total fleet of vehicles registered in the State to these stations. The test only stations model served to address the EPA’s demands for centralized testing, which would be impractical in California because of the number of vehicles tested each year (Eisinger, 2012). The prevalent theory at the time of creation of this category of directed vehicles and the test-only stations was that there would be no incentive for these stations to game the system by improperly failing 16 vehicles to generate repair revenue or by improperly certifying the vehicles after collecting payment for repairing a vehicle when those repairs were ineffective (Eisinger, 2010). Modeling to Predict Emissions Test Failures The on-going dilemma, however, has been in determining which vehicles are appropriate for testing at test only stations. The Bureau eventually selected the high emitter profile (HEP) model developed in 1996 by Radian International corporation discussed earlier (Choo, Shafizadeh, and Niemeier, 2007). However, a later study demonstrated that this model is only slightly more accurate than random chance at identifying vehicles likely to fail a smog check inspection in California (Choo, Shafizadeh, and Niemeier, 2007). Although various research projects produced other models for predicting failures, the Bureau continued to use the HEP model until mid-2012 (Bureau of Automotive Repair, 2012a and Choo, Shafizadeh, and Niemeier, 2007). Moghadam and Livernois (2010), Bin (2003), and Washburn et al. (2001) all conducted studies designed to develop a regression model capable of accurately predicting which vehicles are most likely to fail an emissions inspection with the goal of allowing the majority of vehicle owners whose vehicles consistently pass an emissions inspection to forego the inconvenience of obtaining a required emissions inspection. The problem with the models developed in these studies is that all of these models focus on predicting which vehicles are more likely to fail an emissions test based upon emissions readings. California’s smog check program is changing, reportedly in late 2013, to a testing program consisting only of a visual inspection and a scan of the vehicle’s on-board diagnostic system, referred to as the OBD II system. This change renders invalid any failure prediction model based upon emissions readings and necessitates the development of a model for predicting which vehicles may fail a smog inspection that does not utilize emissions measurements. 17 Moghadam & Livernois (2010) determined the current emissions inspection model utilized in Toronto, Canada achieves the greatest levels of emissions reductions; however, the program is not efficient from a cost standpoint. The researchers conducted a multi-leveled analysis in order to arrive at this determination. First, they conducted a probability analysis to determine at what point in a vehicle’s lifespan it would fall out of compliance with emissions standards. The researchers adjusted this value depending upon whether the vehicle had undergone emissions related repairs. Next, the researchers applied cost estimation functions to the level of emissions reduction achieved per dollar spent. Finally, utilizing the above data, the researchers performed a calculation to determine the lifetime (vehicle life span) costs of maintaining emissions compliance. Based upon their findings, they argue that a 2% reduction in the goal for attainment of emissions reductions would result in a 10 – 13% reduction in costs of the program. However, there probably exists little political support for a program that reduces the goals for pollution reduction. This research is important to the current study because the researchers determined that the prevailing theory of focusing inspections on the oldest vehicles in the fleet is not always the most effective approach because these vehicles often do not experience as much use in their remaining years of usage. In fact, the researchers found that by focusing on median aged vehicles, society achieves far greater emissions reductions per dollar spent. Bin (2003) also conducted a study aimed at determining which vehicles are most likely to fail an emissions inspection. The cost savings to society achieved by only directing those vehicles most likely to fail an inspection for testing and excluding all other vehicles from testing requirements served as Bin’s motivating factor in his research. Bin determined there is a positive correlation between the emissions levels of a vehicle and the vehicle’s age, number of miles on the vehicle, smaller engine sizes, and certain vehicle manufacturers. The study found that age was the most significant of the variables with a one-year increase in the vehicle’s age increasing the 18 likelihood of an emissions failure by 1 to 2%. This study is relevant to this research because it confirms that vehicle age affects emissions levels, supporting the need to use year of manufacture as an explanatory variable in the analysis. Washburn, Seet, and Mannering (2001) also conducted a study to identify vehicles likely to fail an emissions inspection; however, their study found, at a statistically significant level, that increased age and mileage will cause a decrease in a vehicle’s emissions levels. This is inconsistent with most other studies that generally conclude increased vehicle age and odometer reading cause an increase in emissions levels (Bin, 2003; Wenzel and Ross, 2003; Moghadam and Livernois, 2010). Simply stated, prevailing theory dictates that as a vehicle ages in terms of both years and miles accumulated, the emissions levels will increase; yet, Washburn, Seet, and Mannering produced results stating the opposite. Additionally, Wenzel and Ross (2003) published a comment on the Washburn et al. study and suggested that in the development of their model, Washburn may have focused on the wrong vehicle parameters by utilizing emissions values at an idle and misrepresented certain principles of emissions testing by utilizing certain measured exhaust gasses as explanatory variables for other exhaust gasses. Wenzel and Ross (2003) were correct in their observation that Washburn, Seet, and Mannering’s use of emissions values at an idle was incorrect because of the fact that vehicle emissions are nearly always lower at an idle because of the relatively small load (amount of work the engine must perform) placed upon the engine. In addition, Washburn, Seet, and Mannering (2001) utilize certain emissions values as explanatory variables for other emissions values labeled as dependent variables. As an example, Washburn, Seet, and Mattering list carbon dioxide as an explanatory variable for carbon monoxide. Both gasses are products of the combustion process and not causal of one-another. This specification error is problematic in that although the five emissions values measured during an emissions inspection are interrelated 19 and will increase or decrease in expected patterns, they are all the result of vehicle combustion and not causal of one-another. The two errors discussed above likely resulted in the inconsistent results obtained from the study. The published studies present effective models at predicting which vehicles are likely to fail emissions tests consisting of measurements of vehicle emissions; however, the fact that California will no longer be utilizing tailpipe emissions to certify vehicles as emissions compliant necessitates a different model for predicting failures. The studies above are useful in demonstrating that regression analysis is the appropriate tool for identifying these vehicles, provided the researchers select the appropriate variables. This supports the research conducted in this analysis. On-Board Diagnostics, Generation Two (OBD II) Systems The current smog check inspection program consists of three parts: a visual inspection of the emissions components on the vehicle, an emissions measurement of the exhaust gasses of the vehicle, and a functional test of certain vehicle components (Bureau of Automotive Repair, 2009). One of the functional tests performed during a smog check inspection on model-year 1996 and newer vehicles is a scan of the on-board diagnostics, generation two (OBD II) system (Bureau of Automotive Repair, 2009). The OBD II system is a computer based electronic control system that controls engine, transmission, and emissions control system functions to maintain proper engine efficiency and achieve required emissions reductions (Sosnowski and Gardetto, 2001). Currently, 33 states or municipalities utilize an emissions inspection program consisting wholly or partly of an OBD II inspection (Environmental Protection Agency, 2013). By the 1990’s, state and federal regulators had begun enacting regulations requiring automobile manufacturers to utilize on-board diagnostic systems to control the function of the vehicle’s engine. California implemented such a requirement by mandating that beginning with 20 model-year 1991 vehicles, all new cars sold in the State would have on-board diagnostics, first generation, (OBD I) systems (Air Resources Board, 2009). At the federal level, the Clean Air Act Amendments of 1990 mandated improvements to on-board diagnostic systems, including standardization between manufacturers, and that these improvements be incorporated into all passenger cars and light trucks beginning with model-year 1996 (Sosnowski and Gardetto, 2001). Eventually, experts attached the label of "OBD II" systems to these improved, standardized onboard diagnostic systems (Air Resources Board, 2009 and Sosnowski and Gardetto, 2001). California followed the federal lead and also implemented a requirement for OBD II systems beginning with model-year 1996 (Air Resources Board, 2009). OBD II systems are still in use today and have undergone numerous improvements in functionality (Lyons and McCarthy, 2009). This supports the use of the system in evaluating the performance of vehicle emissions control systems. OBD II systems control the operation of the vehicle’s engine, transmission, and emissions control systems to obtain the greatest efficiency while producing the lowest possible emissions levels. The system utilizes a series of sensors to monitor numerous engine parameters. The readings from these sensors are transmitted to the electronic control unit which processes the information and determines how much fuel to supply to the engine and at what point during the engine rotation to ignite the fuel by creating a spark in the ignition system which travels through a spark plug in the engine cylinder, thereby igniting the fuel in the cylinder which generates heat and expansion, creating power. Additionally, the OBD II system determines when to activate various emissions components present on a given vehicle. (Sosnowski and Gardetto, 2001) Another function of the OBD II system is to perform self-tests of the system and related components. These self-tests are referred to as monitors and are performed automatically during vehicle operation when certain conditions are met. If a failure is detected during the completion 21 of these monitors, the OBD II system may generate a diagnostic trouble code (DTC) and may turn on the malfunction indicator lamp, commonly referred to as the ‘check engine’ light, on the vehicle dash display. By regulation, the OBD II system is required to generate a DTC and illuminate the malfunction indicator lamp if a fault is detected that may cause the vehicle to produce 150% of the specified emissions level for that vehicle.(Sosnowski and Gardetto, 2001) The OBD II functional test portion of the California smog check inspection consists of scanning the OBD II computer to determine if all of the monitors are complete. If the monitors are complete, the assumption is that the values generated during the functional test are an accurate representation of the engine and emissions system’s current state of health. In addition to determining monitor status, the functional test also requires the smog check technician to determine if the malfunction indicator lamp illuminates when the key is turned on and turns off when the engine is started. This is another indicator of the proper performance of the system. Finally, the functional test scans the OBD II computer to determine if any DTCs are stored in the system. If DTCs are stored in the computer, this is an indicator of a recent or pending malfunction. Failure of any of these three steps will result in the vehicle failing the smog check inspection. (Bureau of Automotive Repair, 2009) Empirical Studies of OBD II Systems Researchers have conducted empirical analysis of the functionality of OBD II systems. Some of these studies sought to identify a qualitative method for determining the appropriate design and components for the OBD II system. Others have sought to determine the long-term accuracy of the system at identifying emissions failures. These studies demonstrate the validity of applying statistical models to these systems. The first study carried the goal of developing a set of models for use by engineers to formulate the decision algorithms for OBD II systems to identify failures (Cascio, Console, 22 Guagliumi, Osella, Panati, Sottano, and Dupre, 1999). At the time of the research, engineers manually created the algorithms for each application, which countered the EPA’s stated goal of standardization. The researchers concluded they were able to develop specific algorithms for OBD II failure determinations. The principles proposed by the researchers are valid because although each vehicle manufacturer may use different structural components within their emissions and engine control systems, the principles of operation are the same. As an example, an ignition system failure in one cylinder of an engine will result in elevated emissions, regardless of the type of ignition system employed. The specific algorithms developed by the researchers will be useful in developing standardized models for the design of the diagnostic and control structures of the systems in place on automobiles such as the fuel control system or the emissions control systems, and identifying specific areas of fault within those systems (Cascio et al, 1999). Because of the need to use the available computing power for the operation of the system, OBD II systems in use in 1999 were only capable of identifying a specific portion of system operating outside of expected parameters or a generalized fault. Research such as Cascio et al has led to OBD II systems that are much more precise in identifying faults than earlier systems and are improved in overall functionality (Lyons and McCarthy, 2009). This is important because variations in technological advancements between automotive manufacturers will likely lead to variation between manufacturers in failure rates of the OBD II functional portion of the smog check inspection. Barone conducted additional research at streamlining the design process of OBD II systems in 2006. Barone’s stated goal was to develop a statistical method for determining which subsystems to include in the OBD II system. His purpose for the research was to assist automobile manufacturers in achieving the balance between government regulation and consumer satisfaction (2006). Barone pointed out that an OBD II system that was overly sensitive at 23 identifying potential faults would frequently illuminate the malfunction indicator lamp, creating customer satisfaction issues; while a system that was not sensitive enough would fail to meet the requirements of government regulations (2006). Barone’s research provides a fundamental statistical basis for the development of OBD II monitors, which are evaluated as a portion of the OBD II functional portion of the smog check inspection. This is related to the current research because of the fact that monitors are evaluated as part of the inspection and variations in the function of the monitors may lead to certain vehicles exhibiting a higher likelihood of failure of the inspection. In 2011, Supnithadnaporn, Noonan, Samoylov, and Rodgers published a study that detailed the results of an analysis they had conducted to determine the reliability of the emissions inspection program in Atlanta, which utilizes only the on-board diagnostic scan for 1996 and newer light duty vehicles. Their analysis compared the results of on-road emissions measurements gathered for analytical purposes to the results of official inspections to predict the probability of a vehicle passing the official inspection while in actuality the vehicle was emitting excessive emissions. Supnithadnaporn et al. utilized data from an on-going Georgia Tech Research Institute study of actual in-use vehicle emissions and compared that data to the results from the on-board diagnostic scan inspection method. The in-use vehicle emissions data are gathered through the use of remote sensing devices, which measure the emissions of vehicles while the vehicles are in operation (Supnithadnaporn, Noonan, Samoylov, and Rodgers, 2011). The study found that as vehicles age, they are 3.3% per year more likely to generate elevated vehicle emissions while in use in spite of having achieved passing results during an on-board diagnostic scan inspection (Supnithadnaporn, Noonan, Samoylov, and Rodgers, 2011). This means vehicles that successfully pass an inspection consisting of only a scan of the OBD II system may in fact still be generating excessive emissions; especially as those vehicles age. These 24 findings certainly raise concerns about the effectiveness of this methodology at identifying excessively polluting vehicles. The findings of Supnithadnaporn et al. are statistically significant at the p<0.001 level (2011), which indicates a 99.9% probability that actual test results will match the results of their findings. In spite of this statistical significance, there are concerns with the methodology of the study. The major concern is the fact that the regression model compared the on-board diagnostic scan results to the remote sensing data utilizing inspection results obtained after the collection of the remote sensing data. This creates a specification error in the model in that dependent upon the period between the events; it is possible for the vehicle to have been repaired, resulting in an omitted variable. Another concern with the Supnithadnaporn analysis is the model-year distribution of the sample. The study, although published in 2011, only included 1996 through 2002 model-year vehicles. Lyons & McCarthy (2009) indicate that the on-board diagnostic system on 1996 through 1999 model-year vehicles was not fully functional. This creates an additional error in that the regression model does not account for this lack of functionality in the system. In addition to these issues, Supnithadnaporn et al. (2011) incorrectly identify supporting information such as listing that the on-board diagnostic systems were mandated to be installed beginning in the 1994 modelyear and that the system is capable of identifying if the fuel cap has been left off. Lyons & McCarthy (2009) correctly indicate the systems were mandated beginning in 1996 and that although the system is able to identify a vapor leak in the fuel system; it is not able to pinpoint a cause, such as a missing fuel cap. Neither of these errors is as significant as the specification errors and is not likely to affect the regression model. The problems identified with the Supnithadnaporn study do not diminish its usefulness for this research. The Bureau of Automotive Repair mitigated much of this concern by specifying 25 the new smog check inspection program applies to 2000 and newer vehicles only, likely in response to the Lyons and McCarthy study. Consequently, the research conducted in this study is relevant and necessary for identifying model-year 2000 and newer vehicles that are highly likely to fail a smog check inspection consisting of only a visual inspection of the emissions components and a functional test of the OBD II system. Conclusion The literature clearly demonstrates a need for a modern smog check inspection program that takes advantage of the technology employed on modern vehicles. Additionally, the literature outlines the statutory requirement of the Bureau of Automotive Repair to identify a certain portion of the vehicles subject to smog check certification and direct those vehicles to specific facilities for testing. The combination of these two factors identifies the need for a new model of identifying and classifying vehicles as directed vehicles that accounts for a technology based testing program. The literature demonstrates that over the years, various research projects have sought to create an effective model for identifying these directed vehicles. However, previous research fails to identify a model capable of identifying vehicles likely to fail a smog check inspection that consists of only a visual inspection and scan of the OBD II diagnostic system. The functionality and reliability of the OBD II system as described in the literature supports the development of a model that evaluates the data collected from the OBD II functional test portion of the current smog check inspection procedure and operationalizing that data for the purposes of identifying vehicles highly likely to fail the imminent new smog check inspection procedure. 26 Chapter 3 METHODOLOGY To answer the research question of which vehicles are highly likely to fail the OBD II functional portion and the visual inspection portion of the current smog check inspection procedure, I used binomial logistic regression. Binomial logistic regression was the best method to answer this question because of the fact that both dependent variables are binary variables. Both OBD II functional test failure and visual inspection failure are yes (1) or no (0) possibilities. Specifically, I identified the dependent variables as follows: the dependent variable for the regression to identify vehicles highly likely to fail the OBD II functional test portion of the current smog check inspection procedure was “FailOBD” and the dependent variable for the regression model employed to identify vehicles highly likely to fail the visual inspection portion of the current smog check inspection procedure was “FailVisual”. Table 2 below details the number and percentage of vehicles failing each of these two portions of the current procedure. Table 2 Distribution of Failing Vehicles in Current Inspection Procedure Dependent Variable (Component of Current Smog Check Inspection Procedure) Number of Vehicles Failing Number of Vehicles Passing Total Observations OBD II Functional Test Visual Inspection 389,398 54,950 6,894,829 7,229,277 7,284,227 7,284,277 At first glance, the percentages of vehicles failing the OBD II functional test and visual inspection, 5.6% and 0.75% respectively, were lower than desirable for statistical validity. However, evaluating the validity of these values requires consideration of two important factors. First, the overall failure rate for the smog check inspection program in the 2012 calendar year was 27 13.1% (Bureau of Automotive Repair, 2013). As such, the values are consistent with overall program results. Second, and most importantly, the purpose of this thesis is to identify which vehicles are most likely to fail when compared to other vehicles subject to inspection. Consequently, the regression model compared each of the failure groups and identified the vehicles within each group that are most likely to fail the test or inspection. These facts supported the use of the binomial regression analysis with the current dataset to answer the primary thesis question. The remainder of this chapter discusses the data used in the analysis, the coding of the data, and the process utilized in selecting variables. Finally, I conclude with a description of the regression models used to answer the research questions. Smog Check Test Data The primary data source for this thesis is the Smog Check Test Record data. These data are required to be maintained by the California State Bureau of Automotive Repair pursuant to section 44024.5 of the California Health and Safety Code. I requested and obtained the test data for the nearly 13.3 million smog check inspections performed in the state during the 2012 calendar year. Each inspection is considered one observation and each observation consists of 152 categories of information related to vehicle characteristics such as engine size and manufacturer; inspection characteristics such as the technician performing the inspection and the date and time of the inspection; weather conditions at the time of inspection; the results of the inspection; and other information related to the inspection performed on each vehicle. Pursuant to public disclosure laws prohibiting the release of personal identifying information, the Bureau redacted the vehicle license plate numbers and vehicle identification numbers from the dataset prior to providing me with the data. This limited my ability to identify and remove duplicate tests performed on the same vehicle in the dataset. A vehicle receives 28 multiple tests in the same calendar year under two circumstances. First, the vehicle fails the initial test, is repaired, and then retested for certification. In limited instances, the vehicle may receive multiple retests. The second circumstance is change of ownership during the same calendar year the vehicle was tested for biennial registration purposes. Both of these circumstances potentially may skew the results, with the first increasing the number of test failures for vehicles in that category and the second decreasing the number of failures for vehicles in that category. However, unless only a select group of vehicles fall within these circumstances, the size of the dataset overcomes any skewed values caused by these circumstances. The 2012 Smog Check Test Record contains data for all inspections performed, including those not subjected to the new inspection procedures (model-year 1999 and older vehicles). Consequently, I deleted all vehicles not subject to the new inspection procedures from the dataset, leaving approximately 7.9 million observations relevant to only model-year 2000 and newer vehicles. From the 152 categories of information and based on the reviewed literature, I selected only those items capable of identifying those vehicles highly likely to fail the OBD II functional test portion or the visual inspection portion of the current smog check inspection procedure. Regression Models: OBD II Test Failure The first set of regression models sought to predict which vehicles or vehicle combinations of make, model-year, model, engine size, and transmission type are most likely to fail the OBD II test. Problems such as manufacturing defects, computer diagnostic strategies, and defective components can all lead to an increase in the likelihood of a vehicle in any or all of the categories above failing the OBD II functional test. Additionally, although the federal and state requirements for OBD II system functionality vary somewhat, both require that the system identify faults that may cause an increase in vehicle emissions. Consequently, and as the body of literature indicates, each vehicle manufacturer is free 29 to develop specific criteria and algorithms for the OBD II system to follow when performing selfdiagnostic tests aimed at identifying potential emissions failures. This leads to a variation between manufacturers in the functionality and reliability of the OBD II system. Often, there will even be differences in functionality and reliability between model-years of the same manufacturer, as exhibited by the OBD II reference guide the Bureau provides to industry (Bureau of Automotive Repair, 2010b). This variation in design and functionality creates the potential that certain vehicles will demonstrate a higher likelihood of failing the OBD II portion of the smog check. In order to address this issue, it was necessary for me to not only employ categorical variables identifying each vehicle manufacturer as an individual categorical variable, I also needed to include individual categorical variables that identify each vehicle in detail; specifically model-year, manufacturer (make), and engine size, among other characteristics. This led to a large number of explanatory variables; however, this was necessary to specifically identify, with statistical validity, vehicles likely to fail the OBD II functional test. Consideration of these factors led to the following functional form for the first set of regression models: Failure of the OBD II Functional Test (FailOBD) = f (Vehicle Design Characteristics, Vehicle Manufacturer, Year of Manufacture) Previously, I discussed the fact that the dataset contains 152 different categories of information gathered during the smog check inspection procedure. Of these, three were necessary to create the explanatory variables needed to operationalize the three causal categories above. These three categories were VLT REC NO, recoded to evaluate Vehicle Design Characteristics; VEH MODEL YR, recoded into categorical variables to quantify the manufacturer-designated vehicle Model-Year; and VEH MAKE, which I separated into individual categorical variables to describe Vehicle Manufacturer. I discuss the process used to create these variables below. 30 The most important of the explanatory variables from the base dataset was the “VLT REC NO”, which is used to describe Vehicle Design Characteristics. The Vehicle Look-Up Table (VLT) Record (Row) Number (“VLT ID”) identifies the manufacturer of the vehicle (vehicle make), the model-year, the vehicle model, engine size, transmission type, and other factors specific to that vehicle and provides the greatest level of detail about each vehicle.. The Vehicle Look-Up Table is a database programmed into each emissions inspection system throughout the state and is maintained and updated by the Bureau on a regular basis (Bureau of Automotive Repair, 2009). Each specific vehicle configuration, with some exceptions, receives a unique VLT ID used to specifically identify that vehicle. By using each VLT ID as a unique and separate explanatory variable, the analysis will identify specific vehicle configurations likely to fail the new smog check inspection procedures. The specific numerical VLT IDs within the table extend from 00001 to 53084, indicating that there are potentially 53084 specific vehicle configurations identified in the VLT. However, after removing incomplete and invalid observations, the current dataset only contained 2925 VLT ID numbers accounting for approximately 7.5 million observations. There are three probable reasons for only 2925 of 53084 VLT ID numbers remaining in the dataset. First, since the inception of the smog check program in the 1980s, this is the method utilized for identifying vehicles in the program. Consequently, not all vehicle configurations remain in the program. Second, the Bureau frequently updates these identification numbers to identify new vehicle configurations; however, some of the 53084 potential numbers are not in use as of yet. Finally, filtering and cleaning of the dataset likely resulted in the deletion of some VLT ID numbers. In addition to removing observations that were incomplete or invalid, I also deleted observations accounting for incomplete tests, aborted tests, or other irregularities causing a flawed record. The final number of observations for the dataset is 7,284,277. The most prevalent 31 of the vehicle specific ID numbers was VLT ID 29434, which is a 2000 Honda with a 1.6 liter, four-cylinder engine. This vehicle accounts for nearly 50,000 of the smog check inspections performed in 2012. I used this dataset for all regression models in the current study. Appendix B contains a detail of the distribution of the VLT ID numbers in the dataset. As stated above, nearly all of the VLT ID numbers apply to a specific vehicle configuration. However, there are some exceptions to this statement in that in certain situations, vehicles manufactured by a different corporation but with similar engine configurations, vehicle weight, identical model-years, and other categories were grouped together under one VLT ID number. An example of this is VLT ID 2270, which accounts for 51,434 observations. This VLT ID number encompasses model-year 2008 compact vehicles with four cylinder engines and manufactured by several different companies; including Honda, Toyota, Chevrolet, Nissan, and others. This grouping within the same VLT ID number of vehicles manufactured by different companies accounts for approximately 900,000 observations. I have accounted for this circumstance by running a second binomial logistic regression analyses aimed at determining which manufacturers within grouped VLT ID numbers are more likely than the other manufacturers in that group to fail the OBD II test portion or visual inspection portion of the current smog check inspection procedure. In the first iterations of the regression analyses I conducted, I included vehicle make, vehicle model-year, and VLT ID number as the explanatory variables. However, because vehicle make is one of the factors identified in the VLT ID number, these iterations of the regression model resulted in significant multicollinearity. Studenmund (2011) defines multicollinearity in regression formulas as two or more variables that behave identical to one-another in the formula. In essence, the variables are one in the same and are incapable of individual evaluation as to their effect on the dependent variable. This is to be expected in the current analysis because of the fact 32 that vehicle make is a component of VLT ID numbers. As an example, it is quite possible for the aforementioned VLT ID number 29434 as an explanatory variable to exert the same effect on the dependent variable as Honda as an explanatory variable. To remedy this situation and to account for grouped VLT ID numbers, I conducted three separate regression analyses on each dependent variable. The first regression analysis was a binomial regression with failure of the OBD II functional test as the dependent variable and 2925 VLT ID numbers as the explanatory variables. To account for the VLT ID numbers containing multiple manufacturers and to determine if a specific vehicle manufacturer within a combined VLT ID is more likely than others to fail the OBD II functional test; I completed a second regression analysis with individual vehicle makes as the explanatory variables. Conducting the binomial regression with vehicle manufacturer as the explanatory variable sorts the manufacturers by likelihood of failure and created the ability to accurately classify each manufacturer within combined VLT ID numbers. There were 48 individual categorical variables created from the unique identifiers within the VEH MAKE category from the primary dataset. Table 3 below details the distribution of the vehicle manufacturers within the dataset. Table 3 Distribution of Vehicle Manufacturers within the dataset Manufacturer Number of Observations Percentage of Total Observations ACURA 134,221 1.84 ASTON-MARTIN 114 0 AUDI 39,829 0.55 BENTLEY 92 0 BMW 230,074 3.16 BUICK 53,544 0.74 CADILLAC 76,206 1.05 CHEVROLET 819,246 11.25 CHRYSLER 181,412 2.49 33 DAEWOO 3,879 0.05 DODGE 363,085 4.98 FERRARI 325 0 FIAT 669 0.01 FORD 963,066 13.22 GMC 201,102 2.76 HONDA 796,354 10.93 HUMMER 12,500 0.17 HYUNDAI 122,244 1.68 INFINITI 80,954 1.11 ISUZU 9,034 0.12 JAGUAR 18,313 0.25 JEEP 134,006 1.84 KIA 81,529 1.12 LAMBORGHINI 48 0 LAND ROVER 24,374 0.33 LEXUS 197,881 2.72 LINCOLN 54,229 0.74 LOTUS 61 0 MASERATI 425 0.01 MAZDA 109,333 1.5 MERCEDES 189,849 2.61 MERCURY 34,258 0.47 MINI 23,060 0.32 MITSUBISHI 105,790 1.45 NISSAN 470,520 6.46 OLDSMOBILE 12,183 0.17 PLYMOUTH 5,633 0.08 PONTIAC 71,791 0.99 PORSCHE 20,590 0.28 SAAB 7,967 0.11 SATURN 87,776 1.21 SCION 69,453 0.95 SMART 512 0.01 SUBARU 59,938 0.82 SUZUKI 15,406 0.21 TOYOTA 1,159,923 15.92 VOLKSWAGEN 177,448 2.44 VOLVO 63,981 Total Observations 0.88 7,284,227 34 Finally, I conducted a third binomial regression analysis with the explanatory variables generated from the VEH MODEL YR category of information from the 2012 smog check dataset. Completion of this third regression analysis was necessary to provide further detail as to which vehicles will likely fail the OBD II functional test portion of the current smog check inspection procedure. The additional detail is necessary to clarify any discrepancies in the results of the primary regression model conducted using VLT ID Numbers. From the VEH MODEL YR category of information, I created 12 individual categorical variables, one for each model-year between 2000 and 2012. Table 4 below details the distribution of vehicle model-years within the primary dataset. Table 4 Distribution of Vehicle Model-Years within the Dataset Vehicle Model Year Number of Observations Percentage of Total Observations 2000 1,122,911 15.42 2001 556,844 7.64 2002 1,246,385 17.11 2003 508,263 6.98 2004 1,353,929 18.59 2005 417,171 5.73 2006 1,432,545 19.67 2007 210,009 2.88 2008 190,003 2.61 2009 63,743 0.88 2010 71,235 0.98 2011 73,767 1.01 2012 37,422 0.51 Total Observations 7,284,227 Regression Models: Visual Inspection Failure The second dependent variable was “FailVisual”, which is the overall pass or fail result for the visual inspection portion of the current smog check inspection procedure. The inclusion of 35 a regression model to identify vehicles likely to fail to the visual inspection portion of the smog check inspection procedure was aimed at determining whether certain vehicles are likely to undergo modifications or incur deterioration of components that will cause failure of this inspection. The predominant reason for a failure of the visual inspection is modification of emission components with the goal of improving the performance, often associated with ‘street racing’, the illegal modification and racing of vehicles operated on public roadways. If certain vehicles are more prone to modification, it is appropriate to direct those vehicles for specialized testing procedures. The functional form for this regression model was as follows: Visual Inspection Failure (FailVisual) = f (Vehicle Design Characteristics, Vehicle Manufacturer, Year of Manufacture) For the second regression analysis evaluating the likelihood of a vehicle failing the visual inspection portion of the current smog check inspection procedure, I utilized the identical final dataset and sets of explanatory variables (See Tables 2, 3 and Appendix B). Additionally, I performed additional binomial logistic regression analyses on the dependent variable of failure of the visual inspection portion of the current smog check inspection procedures. This was for the same reason as performing additional regressions on the failure of the OBD II functional test, to determine if specific manufactures within the VLT ID numbers that contain multiple manufacturers result in a higher likelihood of failure of the visual inspection and to further define the results of the analyses. The methodology and categorization described was the best approach for answering the research question. In the next chapter, I detail the results of the regression analyses just described. 36 Chapter 4 ANALYSIS Completion of the regression analyses as described in Chapter 3 provided significant and meaningful results. The results accurately identify a specific group of vehicles significantly more likely to fail a smog check inspection consisting of a scan of the vehicle’s OBD II diagnostic system and a visual inspection of the vehicle emission control devices. In review, I designed two separate but similar groups of regression models to identify which vehicles are highly likely to fail the revised smog check inspection procedure scheduled for implementation by the Bureau of Automotive Repair in 2014. One group of regression models focuses upon identifying vehicles likely to fail a scan of the OBD II diagnostic system present on nearly all passenger cars and light trucks produced for sale in the United States since 1996. The second group of regression models seeks to identify vehicles likely to fail the visual inspection portion of the smog check inspection. I conducted binomial logistic regressions with the results reported as “odds ratios”. This reporting method lists the coefficient for each explanatory variable as a prediction of the likelihood of an occurrence. In this reporting method, a coefficient of 1.000 is the base value, meaning the event or effect described by a variable with this value is equal to the event or effect described by the control variable. A variable with a coefficient value of 0.500 is 50% less likely to take place or exerts 50% less of an effect on the dependent variable as compared to the control variable. Finally, a variable with a coefficient value of 2.000 is 100% more likely to take place or exerts 100% greater of an effect on the dependent variable as compared to the control variable. Although this reporting method enables me to predict the likelihood of a vehicle failing the inspection as compared to a reference vehicle, the goal of this thesis is to predict which vehicles are most likely to fail. Consequently, the result tables included later in this thesis and in 37 the appendices attached display likelihood of failure expressed as odds ratios with the higher odds ratio values indicating a higher likelihood of failure. It is important to note that I originally completed a fourth regression model wherein I combined the vehicle make with the vehicle model year, 2002 Chevrolet as an example. However, this methodology resulted in substantial collinearity, resulting in the omission of a large number of variables by the statistical software and compromising the validity of such an approach. Additionally, the Vehicle Look-Up Table identification numbers previously described provide far greater detail for analysis. The next paragraphs detail the results of the regression analyses. Predicting OBD II Test Failures I. First Regression: Fail OBD = f (VLT ID Numbers) The three regressions I conducted to identify vehicles highly likely to fail the OBD II functional test portion of the current smog check inspection procedure, and subsequently the revised smog check inspection procedure consisting of a scan of the OBD II system and visual inspection, utilized failure of the OBD II functional test as the dependent variable. For the first of the three, the explanatory variables were the Vehicle Look-Up Table identification numbers. The final cleaned version of the VLT ID numbers consisted of 2925 individual identifiers (Appendix B). During previous iterations of the regression model I conducted in determining the preferred method for conducting this analysis; I discovered Stata, the statistical software used in the analysis, automatically selected the last explanatory variable as the control variable. This takes place regardless of whether I independently selected a control variable. Consequently, I allowed Stata to select the control variable, which in this regression model was VLT ID number 53079. This VLT ID number applies to a 2011 Chevrolet Silverado 2500 pick-up truck with a 6.6-liter diesel engine and either a manual or an automatic transmission. There were 657 observations 38 associated with this vehicle, accounting for 0.01% of the nearly 7.3 million observations. I compared the remaining 2924 VLT ID numbers to this VLT ID. Completion of the regression model provided meaningful results. Exactly 116 of the VLT ID numbers resulted in a lower likelihood of failing the OBD II functional test as compared to VLT ID 53079, a 2011 Chevrolet Silverado. Of the remaining group, 18 of VLT ID numbers demonstrated zero failures, meaning none of these vehicles resulted in a failure of the OBD II functional test. Finally, 2790 of the VLT ID numbers demonstrated a higher likelihood of failing the OBD II functional test as compared to a 2011 Chevrolet Silverado pick-up truck. The 18 VLT ID numbers that demonstrated no failures of the OBD II functional test are all 2005 and newer vehicles. In theory, these vehicles will likely demonstrate future failures as they age. It is interesting to note that of the vehicles demonstrating zero failures the Toyota Corporation (Toyota and Lexus) manufactures 10 of the 18 vehicles, or 56%, while at the same time these makes only account for 18.7% of the dataset. Table 5 below provides information for these vehicles. Table 5 List of Vehicles Demonstrating Zero OBD II Functional Test Failures VLT ID 2487 33036 33235 33696 33768 34025 34081 34351 34359 34366 34370 34549 Model-Year 2012 2005 2005 2006 2006 2007 2007 2007 2007 2007 2007 2007 Make Volvo Lexus Toyota Lexus Mercury Cadillac Chevrolet Land Rover Lexus Lexus Lexus Toyota Model S 60 LS 430 Sequoia 4WD SC 430 Mariner DTS K1500 Suburban Range Rover GS 350 LS 460 RX 350 4WD 4 Runner 4WD Engine Size 2.5 liter 4.3 liter 4.7 liter 4.3 liter 3.0 liter 4.6 liter 5.3 liter 4.4 liter 3.5 liter 4.6 liter 3.5 liter 4.0 liter 39 34570 34580 34583 34597 52982 53038 2007 2007 2007 2007 2010 2010 Toyota Toyota Toyota Volkswagen BMW Ford Rav 4 2WD Tacoma 4WD Tundra 2WD Rabbit X5 F350 Diesel 2.4 liter 4.0 liter 4.7 liter 2.5 liter 3.0 liter 6.4 liter Beyond the vehicles that demonstrated zero failures of the OBD II functional test, the actual odds ratios for the remaining variables, while significant and meaningful, are less important than the rank of each VLT ID number as compared to the remaining ID numbers. As stated throughout this thesis, the goal of the current research is to predict which vehicles are most likely to fail the OBD II functional test portion of the current smog check inspection procedure. Consequently, the following discussion and analysis of the regression results focuses on the distribution of the VLT ID numbers within the regression model. The mean odds ratio value for the regression analysis was 10.981. In determining which vehicles are more likely to fail the OBD II functional test, any VLT ID number with an odds ratio above this value has a greater than average likelihood of failing the test. Reviewing the results of the analysis as detailed in Appendix C shows that 1636 VLT ID numbers were below this mean value, leaving 1288 VLT ID numbers with odds ratio values above the mean value. These 1288 VLT ID numbers account for 36.92% of the total observations in the dataset. The first of the vehicles identified by this group of VLT ID numbers range is a 2004 Toyota Tundra with a 3.4 liter engine and two-wheel drive. The vehicle identified by this VLT ID number is only slightly above average in terms of likelihood of failing the OBD II functional test with an odds ratio of 10.986. The vehicle identified by the VLT ID number most likely to fail the OBD II functional test is a 2001 Mazda MPV mini-van with a 2.5-liter engine. This vehicle is very highly likely to fail the OBD II functional test with a reported odds ratio for this VLT ID number of 43.542. 40 Appendix C contains a detailed list of the VLT ID numbers and associated vehicles with odds ratio values greater than the mean. The binomial logistic regression with VLT ID numbers as the explanatory variables predicting which vehicles are highly likely to fail the OBD II functional test was very effective at identifying a group of vehicles likely to cause a failure of the test. This group of VLT ID numbers, especially those with the highest odds ratio values, identifies a very specific group of vehicles that are appropriate for identification as directed vehicles. II. Second Regression: Fail OBD = f (Vehicle Make) The second of the three regressions utilizing failure of the OBD II functional test portion of the current smog check inspection procedure encompassed vehicle make (manufacturer) as the explanatory variables. The regression model with VLT ID numbers as the explanatory variables resulted in the identification of approximately 65 VLT ID numbers that are attributable to multiple vehicle manufacturers. This made it necessary to run this second regression to determine if any manufacturer, or group of manufacturers, is more likely than the others to fail the OBD II functional test. These data enabled me to determine if specific manufacturers within a combined VLT ID number are more likely than the other manufacturers to fail. The next paragraphs discuss the results of this regression. There were 48 individual vehicle makes remaining in the filtered and cleaned dataset. Completion of the regression resulted in all tests performed on vehicles manufactured by Lamborghini passing the OBD II functional test. Every other manufacturer had at least one observation resulting in failure of the OBD II functional test. Additionally, the regression software automatically selected a comparison variable for the regression model. In this model, Stata omitted the manufacturer Volvo as the comparison variable. . In this regression, the mean odds ratio was 1.062, meaning those manufacturers with a higher value than 1.062 have a higher 41 likelihood of failure of the OBD II functional test as compared to the others. Table 6 below details the results of this regression model. Table 6 Results of Regression Model with Vehicle Make as the Explanatory Variables Vehicle Make 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 FIAT ASTON-MARTIN LOTUS SMART LEXUS MINI MASERATI PORSCHE ACURA INFINITI HONDA SCION SAAB TOYOTA BENTLEY MERCEDES SUBARU LAMBORGHINI BMW FORD NISSAN LINCOLN AUDI JAGUAR Odds Ratio 0.054 0.159 0.300 0.359 0.441 0.444 0.568 0.578 0.582 0.636 0.641 0.645 0.685 0.796 0.819 0.881 0.986 Omitted 1.019 1.027 1.039 1.044 1.089 1.096 Vehicle Make 25 CADILLAC 26 BUICK 27 MERCURY 28 JEEP 29 GMC 30 FERRARI 31 HUMMER 32 HYUNDAI 33 MAZDA 34 CHEVROLET 35 LAND ROVER 36 SATURN 37 VOLKSWAGEN 38 CHRYSLER 39 DODGE 40 PONTIAC 41 KIA 42 MITSUBISHI 43 PLYMOUTH 44 SUZUKI 45 ISUZU 46 OLDSMOBILE 47 DAEWOO Mean Odds Ratio Odds Ratio 1.103 1.121 1.123 1.155 1.168 1.181 1.222 1.235 1.265 1.295 1.326 1.331 1.351 1.394 1.430 1.521 1.556 1.631 1.762 1.795 1.847 1.872 3.329 1.062 As Table 6 shows, the results of this regression model are evenly distributed. The mean odds ratio value of 1.062 is very close to the base value of 1.000 for binomial regression. Additionally, the reported odds ratio values for all but one of the manufacturers range from 0.054, 42 which is 95% less likely than a Volvo to fail the OBD II functional test to a high of 1.872, which is 87% more likely than a Volvo to fail the OBD II functional test. The only exception to this is the manufacturer Daewoo, which is 3.3 times more likely than the Volvo to fail the OBD II functional test. This regression model is very effective at showing that, with the exception of Daewoo; the vehicles are nearly identical in terms of likelihood of failure of the OBD II functional test. Consequently, only Daewoo vehicles contained within any of the combined VLT ID numbers have a greater likelihood of failure of the OBD II functional test. The remaining manufacturers, while showing noticeable differences in the results tables, cannot be labeled as “highly likely” to fail the OBD II functional test portion of the current smog check inspection procedure. III. Third Regression: Fail OBD = f (Model-Year) Finally, the third regression with failure of the OBD II functional portion of the current smog check inspection procedure as the dependent variable encompasses model-year as the explanatory variables. As a reminder, the model-year is identified by the manufacturer and is not necessarily the year of production. As an example, many 2012 model-year vehicles were manufactured in July or August, and occasionally earlier months, of 2011. This circumstance has always existed in automobile manufacturing. Given the current study evaluated model-year 2000 and newer vehicles, the regression model consisted of 13 explanatory variables; model-year 2000 through model-year 2012. As always, Stata dropped the last explanatory variable as the control variable. In this instance, this was model-year 2012. The distribution of odds ratios details the expected results. The older a vehicle is, the more likely it is to fail the OBD II functional test. Model-years 2007 through 2011 are least likely to fail the test, while model-years 2000 through 2006 are clearly more likely to fail 43 the test. For this regression, I did not include the mean odds ratio of 4.459 as the results in Table 7 clearly demonstrate the model-years most likely to fail the OBD II functional test. Table 7 Odds Ratios for Vehicle Model-Years ModelYear MY2011 MY2010 MY2009 MY2008 MY2007 MY2006 MY2004 MY2005 MY2002 MY2003 MY2000 MY2001 Odds Ratio 0.948 1.023 1.550 1.640 1.818 2.908 4.376 4.610 7.415 7.549 7.551 12.120 It is important to note that in the first six model-years, the odd year is more likely to fail than the even year; despite the fact the even year vehicles are one year older. This is likely because the smog check is required every other year once a vehicle is six years old or when a transfer of ownership occurs. Consequently, in even calendar years, even model-year vehicles are undergoing inspection as a requirement for renewal of registration while odd model-year vehicles are undergoing an inspection primarily for change of ownership. The dataset in this analysis is for the 2012 calendar year. Consequently, it is reasonable to assume that a substantial portion of the model-year 2001 vehicles received inspections for change of ownership purposes, and as Table 7 shows, these vehicles are likely to fail the OBD II test at a significantly higher rate than other model-years. The fact that vehicles inspected for change of ownership purposes demonstrate a higher likelihood of failure is significant for the purposes of identifying directed vehicles. Predicting Visual Inspection Failures I. First Regression: Fail Visual = f (VLT ID Numbers) As stated previously, I conducted two sets of regression analyses in determining which vehicles are highly likely to fail the new smog check inspection procedure scheduled for implementation in 2014. The first set of analyses sought to predict vehicles highly likely to fail the OBD II functional test portion of the current smog check inspection procedure. The new 44 inspection procedure mirrors this test. The second set of three regression analyses seek to predict which vehicles will fail the visual inspection portion of the current smog check inspection procedure, which will mirror the new inspection procedure as well. All three of the regression models in this group utilize “Fail Visual” as the dependent variable. Each of the three models applies a set of explanatory variables against this variable to determine which of the explanatory variables identifies vehicles highly likely to result in a visual inspection failure. The first regression analysis in this group utilizes the same set of VLT ID numbers as explanatory variables as utilized in the first OBD II regression model. Stata once again selected VLT ID number 53079 as the control variable. As stated earlier, this VLT ID number applies to a 2011 Chevrolet Silverado 2500 pick-up truck with a 6.6-liter diesel engine and either a manual or an automatic transmission. However, there are little similarities beyond the same control variable for the two regression models utilizing VLT ID numbers as explanatory variables. Although the comparison of the explanatory variables to one-another is again more important than the actual odds ratio values, there is a substantial difference between the two models as far as this comparison is concerned. This regression model identified 2435 VLT ID numbers less likely to fail the visual inspection as compared to the control variable, which is nearly the exact opposite of the results with OBD II failure as the dependent variable. Additionally, there were 211 VLT ID numbers demonstrating zero failures of the visual inspection as compared to only 18 demonstrating zero failures in the OBD II functional test. Appendix D lists the VLT ID numbers and associated vehicles that demonstrated zero failures of the visual inspection portion of the smog check inspection in the 2012 calendar year. As with the regression model with OBD II failure as the dependent variable, the actual values of the odds ratios reported in this regression model are less important that the distribution 45 of the VLT ID numbers in the list. In the visual inspection failure model, the mean value of the odds ratios was 0.624. The fact that the mean value is noticeably below the base value of 1.000 for logistic regression is likely because of the fact that, as discussed in Chapter 3 of this thesis, only 0.75% of vehicles failed the visual inspection in the 2012 calendar year. This low failure rate leads to the fact that the substantial majority of vehicles are unlikely to fail the visual inspection, in turn causing the low mean value for the odds ratios of visual inspection failure. The aforementioned circumstances led me to select only the VLT ID numbers that demonstrated statistical significance in the regression results as those VLT ID numbers with a higher likelihood of failure of the visual inspection portion of the smog check inspection procedure. In the current analysis, there were 153 VLT ID numbers with statistically significant results demonstrating a higher likelihood of failure of the visual inspection. These VLT ID numbers account for 3.71% of the inspections performed in 2012. The lowest of these, but measurably higher than the mean, was VLT ID number 50022 with a reported odds ratio of 1.605, indicating this vehicle is nearly 100% more likely than the mean and 60% more likely than the control variable to fail the visual inspection. VLT ID number 50022 is a grouped VLT ID number that applies to heavy duty light trucks with diesel engines and built by domestic manufacturers (Chevrolet, Dodge, GM, and Ford). The VLT ID number demonstrating the highest likelihood of failure of the visual inspection is VLT ID 31158, with a reported odds ratio of 4.331. This VLT ID number applies to a 2002 Volkswagen Cabrio with a 2.0-liter engine and manual transmission. Appendix E provides details for the 153 VLT ID numbers demonstrating a significant likelihood of failure of the visual inspection portion of the current smog check inspection procedure. II. Second Regression: Fail Visual = f (Vehicle Make) As with the second regression model to identify vehicles highly likely to fail the OBD II functional test, the second regression model to identify vehicles highly likely to fail the visual 46 inspection portion of the smog check inspection procedure utilizes vehicle make as the explanatory variables. The set of explanatory variables consisted of the same 48 individual vehicle manufacturers as the OBD II set, and Stata again dropped Volvo as the control variable. Although there are similarities between the two regression models with different dependent variables, there are also noticeable differences in the results of the regression analysis. As with the fail OBD regression model, the manufacturer Lamborghini demonstrated zero failures of the visual inspection. However, in this model, the manufacturers Aston Martin, Bentley, Ferrari, Lotus, and Maserati also demonstrated zero failures of the visual inspection. The fact that these are all high-end luxury car manufacturers is worth noting. Another notable difference between the two models is that this regression model demonstrated a greater number of manufacturers with a higher likelihood of failure of the visual inspection. In spite of this greater number of manufacturers demonstrating a higher likelihood of failure, the manufacturer Daewoo is once again the most likely manufacturer to fail the visual inspection portion of the current smog check inspection procedure. However, the manufacturers Pontiac, Mitsubishi, Dodge, Oldsmobile, Volkswagen, and Plymouth also demonstrate higher odds ratio values, thus indicating a higher likelihood of failure. Table 8 below details the odds ratio values generated in the regression analysis. Table 8 Results of Regression Model with Vehicle Make as the Explanatory Variables 1 2 3 4 5 6 7 Vehicle Make FIAT PORSCHE SMART MERCEDES LEXUS LAND ROVER MINI Odds Ratio 0.237 0.285 0.310 0.409 0.413 0.581 0.677 25 26 27 28 29 30 31 Vehicle Make HYUNDAI GMC ISUZU SAAB SCION SATURN BUICK Odds Ratio 1.080 1.091 1.096 1.122 1.150 1.232 1.347 47 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 TOYOTA BMW HONDA INFINITI CADILLAC JAGUAR LINCOLN ACURA ASTONMARTIN BENTLEY FERARI LAMBORGHINI LOTUS MASERATI NISSAN KIA HUMMER 0.683 0.700 0.746 0.773 0.801 0.844 0.920 0.996 Omitted Omitted Omitted Omitted Omitted Omitted 1.030 1.069 1.073 32 33 34 35 36 37 38 39 SUZUKI AUDI CHEVROLET SUBARU FORD MAZDA MERCURY CHRYSLER JEEP 40 PONTIAC 41 42 MITSUBISHI DODGE 43 44 OLDSMOBILE 45 VOLKSWAGEN PLYMOUTH 46 DAEWOO 47 Mean Odds Ratio 1.360 1.369 1.438 1.487 1.561 1.580 1.637 1.790 1.855 2.201 2.325 2.351 2.606 2.651 4.041 4.714 1.305 The results of the binomial logistic regression analysis clearly identify a set of vehicle manufacturers with a higher likelihood of failing the visual inspection portion of the current smog check inspection procedure. This information will again prove useful at identifying specific vehicle makes within combined VLT ID numbers that are more likely to fail the visual inspection. III. Third Regression: Fail Visual = f (Model-Year) The final regression model in the second group encompassed vehicle model-year as the set of explanatory variables utilized to describe which vehicles are highly likely to fail the visual inspection portion of the smog check inspection. Stata again dropped model-year 2012 from the dataset as the control variable. Another similarity to the regression conducted to predict which vehicles are highly likely to fail the OBD II inspection as explained by vehicle model-years is that older vehicles are substantially more likely to fail the visual inspection portion of the smog check 48 inspection procedure. A final similarity is that over the first six model-years in the dataset, the Table 9 Results of Regression Model with ModelYear as the Explanatory Variables ModelYear MY2011 MY2010 MY2009 MY2008 MY2007 MY2006 MY2004 MY2005 MY2002 MY2003 MY2000 MY2001 Odds Ratio 1.488311 3.400279 3.643585 4.523026 6.906138 9.775916 14.06943 15.0705 23.86911 25.09562 31.61048 32.40856 odd model-year vehicles exhibit a higher likelihood of failure than the even model-year in spite of the fact that the even model-year vehicles are one year older. Again, this supports the theory that vehicles inspected for the purposes of change of ownership are more likely to fail the inspection. One difference between these regression results and those for the OBD II failure regression with model-years as the dependent variable is the odds ratio values are substantially higher in the visual inspection regression model. This indicates that vehicle model-year is quite significant in terms of predicting which vehicles are highly likely to fail the visual inspection portion of the smog check inspection procedure. Table 9 details the results of the regression analysis. Reliability and Validity In any study, the implementation of measures to insure the reliability and validity of the conclusions drawn from the analysis is an important step. In order to draw valid conclusions from a regression analysis, reliability of measurement is essential. According to Singleton and Straits (2010), reliability of a measurement means whatever is under measurement in the analysis is consistently and dependably measured. Validity means the predictions made by the analysis are a true reflection of the potential actual outcome of the event (Singleton and Strait, 2010). Reliability of measurement in regression analysis is generally a concern because most regression analyses analyze a sample of data from a much larger dataset. However, this is not a concern in this analysis because the model did not examine a sample of the smog check 49 inspection data. I specifically decided to apply the regression model to the entire dataset of smog check inspections to overcome reliability concerns. Additionally, the large number of potential explanatory variables required inclusion of all valid and applicable observations to ensure sufficient data for each variable. Finally, because of the fact that a small portion of the vehicles in the dataset received multiple inspections in the reporting period, inclusion of all possible observations was necessary to overcome any skewed values created by the multiple inspections. These factors cause me to have a high confidence level in the reliability of the dataset and subsequent regression analyses. The question asked in evaluating the validity of this model is whether it accurately identified those vehicles with the highest likelihood of failing either the OBD II functional test or the visual inspection portions of the current smog check inspection procedure. A potential cause for failure of the model to accurately identify these vehicles is the omission of an important variable or variables. In the current study, two potential omitted variables were vehicle condition and testing behaviors. A poorly maintained vehicle may cause an unexpected failure of either portion of the inspection procedure while similar vehicles are less likely to fail the inspection. At the same time, a less-than-scrupulous smog check technician or station may, for whatever reason, take steps to allow a vehicle to pass the inspection when it should not. Both of these conditions can skew the results of the analysis but were addressed by the methodology selected. In regard to the issue of an unscrupulous technician or station, the volume of observations in the dataset sufficiently minimizes any risk to validity. There were 7,962 active smog check stations in 2012 and 11,921 active technicians (Bureau of Automotive Repair, 2013). The average percentage of tests in the dataset of 7,284,277 tests performed by each station was approximately 0.11% while the average percentage of tests performed by each technician was 0.16%. Consequently, the likelihood of any technician or station performing a large number of tests on a 50 specific category of vehicle within the dataset is very minimal and constitutes an acceptable level of risk. Maintenance of vehicles is not a risk to the validity of the analysis simply because a major goal of the smog check program is to encourage the improved maintenance of vehicles, resulting in lower emissions (Eisinger, 2010). Smog check programs are often referred to as “I/M” programs, meaning “inspection and maintenance” programs. The theory is that if consumers are aware of the maintenance requirements that must be met for their vehicle to pass the inspection, they are more likely to have the necessary repairs performed. Consequently, if a vehicle is less likely to fail the inspection as a result the maintenance practices of the owner, that is precisely the type of vehicle this study seeks to identify. Again, the risk to the validity of the study created by this potential omitted variable is minimal. Another common test for validity is hypothesis testing. Although none of the models were seeking to prove or disprove a hypothesis, the principles of hypothesis testing still apply to the models. Hypothesis testing employs the use of a calculated “t-statistic” or “z-test”, which are compared to a chart containing a critical value for the t-statistic or a normal distribution chart for the z-test. If the calculated value exceeds the critical value, the researcher must accept the hypothesis proposed by the variable. For logistic regression, determination of validity of the hypothesis encompasses the z-test. Using the regression model with OBD II failure as the dependent variable and VLT ID numbers as the explanatory variables as an example, all 1288 VLT ID numbers above the mean value reported a z-value of greater than 4.30 while the normal distribution value for the z-value is 3.013 at a significance of less than .01. This means there is a 99% probability that the actual failure rate of the vehicles will be within the range specified in the analysis. In five of the six regression models conducted as part of this study, the explanatory variables demonstrating the greatest likelihood of failing the OBD II functional test or the visual 51 inspection all demonstrated a significance level of greater than 99%. The only exception was the model predicting failure of the visual inspection based upon VLT ID number. As discussed above, there were 153 explanatory variables demonstrating statistical significance in the model utilizing failure of the visual inspection as the dependent variable. Of these, 50 were significant at the 99% level, 57 were significant at the 95% level, and 46% were significant at the 90% level. This means that there is a 99%, 95%, and 90% probability, respectively, that the actual failure rates of these vehicles fall within the range specified in the analysis. Again, the low number of failures as a portion of the overall dataset leads to the low number of statistically significant explanatory variables. A final test for validity conducted on the regression models was the chi-squared test. This test assesses the validity of the entire regression model by calculating the likelihood of results generated by random chance. The lower the likelihood of random chance, the greater the validity of the regression model as a whole (Adcock, 2010). The chi-squared calculation is similar to the t-statistic and z-test in that the regression model calculates a chi-squared value to compare to a value contained in the “Critical Values of Chi-squared” table. Obtaining the critical chi-squared value requires identifying the intercept of rows labeled “degrees of freedom” and columns labeled “probability”. The degrees of freedom equates to the number of valid explanatory variables. The probability ranges from .001 (99.9% likelihood that the results are not random chance) to .10 (90% likelihood that the results are not random chance). For the current regression analyses, the calculated chi-squared values ranged from 11569.80 to 183315.20, well above the critical chisquared values and indicating a 99.9% likelihood the results of each regression model were not a result of random chance. Table 8 below details the chi-squared values for the current analysis. 52 Table 10 Detail of Chi-Squared Values for Regression Model Dependent Variable Explanatory Variables Degrees of Freedom Critical Chi-squared Value Calculated Chisquared Value Likelihood of Random Chance Fail OBD II Test Vehicle ModelVLT ID Make Year 2906 46 12 Fail Visual Inspection Vehicle ModelVLT ID Make Year 2173 41 12 149.45 86.66 32.91 149.45 80.08 32.91 183315.20 31335.43 93242.97 43818.82 11569.80 14038.44 < .001% < .001% < .001% < .001% < .001% < .001% As stated in the introduction to this section, the results of the regression models were significant and meaningful. In the next chapter, I will detail the conclusions drawn from the analysis and make recommendations for future policy actions. 53 Chapter 5 CONCLUSION The regression analyses conducted in this thesis provide insight into alternative methods of identifying vehicles highly likely to fail the California smog check inspection procedure. The required inspection procedure is undergoing substantial change, with the elimination of the emissions measurement portion of the inspection procedure for a large portion of the fleet being the most significant change. One of the reasons for this thesis is the fact that the current model for identifying vehicles, referred to as directed vehicles, highly likely to fail the smog check inspection is predominately predicated on the emissions measurements obtained during the inspection. Consequently, transition to a different inspection procedure requires transition to a different model for identifying directed vehicles. Historically, the Bureau of Automotive Repair has directed, for specialized testing, approximately 30% of the entire fleet of vehicles subject to inspection. All vehicles, regardless of model-year, were subject to evaluation for directed vehicle status. However, beginning in 2014, vehicles model-year 2000 and newer will no longer be subject to an emissions measurement as part of their smog check inspection. The inspection procedure for these vehicles will consist solely of a scan of the vehicle’s on-board diagnostic (OBD II) system and a visual inspection of the emission control devices on the vehicle. Consequently, the existing method for identifying directed vehicles will no longer apply to this group of vehicles, which comprises nearly 60% of the vehicles inspected each year, and this percentage grows each year. To say a system of identifying vehicles for specialized testing by applying the model to less than half of the vehicles subject to inspection lacks validity is an understatement at best. In 2011, 1999 and older vehicles accounted for approximately 41% of all inspections and in 2012, that number dropped to 35%. Based upon this pattern, in the very near future, the 54 number of these vehicles subject to testing will be less than 30%. Although the Bureau has indicated the intention to identify all model-year 1999 and older vehicles as directed vehicles, the progressively decreasing volume of these vehicles necessitates the development of a model capable of identifying directed vehicles within the model-year 2000 and newer group The regression models developed as part of this thesis proved highly effective at identifying vehicles in the model-year 2000 and newer group with a higher propensity for failure of the revised smog check inspection procedure scheduled for implementation in 2014. In the remainder of the chapter, I discuss the findings of the analysis, opportunities for improvement to the model, and recommendations for continued effectiveness of the practice of identifying directed vehicles. Predicting Likelihood of Failure of the OBD II Functional Test The first set of regression models completed in support of this thesis sought to identify vehicles highly likely to fail the OBD II functional test portion of the current smog check inspection procedure. This test is equivalent to the diagnostic scan portion of the new smog check inspection procedure; as such, vehicles highly likely to fail this portion of the current inspection procedure are also likely to fail the new inspection procedure. The three regression models separately utilized vehicle look-up table identification (VLT ID) numbers, vehicle make (manufacturer), and vehicle model-year as explanatory variables with failure of the OBD II (Fail OBD) as the dependent variable in each. VLT ID numbers provide the most precise detail about the vehicles subject to inspection because, with a few exceptions, each VLT ID number is assigned to vehicles of the same make, model, model-year, engine size, transmission type, and other characteristics. The exceptions are a group of VLT ID numbers that group together vehicles of a similar configuration but manufactured by different companies. 55 The model utilizing VLT ID numbers as the explanatory variables effectively identified a group of 1288 vehicles that account for approximately 37% of the inspection volume and are more likely to fail the OBD II functional test. The vehicles identified range from only slightly more likely than other vehicles to fail the inspection to substantially more likely to fail the inspection. Given the progressively decreasing volume of vehicles in the model-year 1999 and older set, labeling the vehicles in the 2000 and newer set with the highest likelihood of failure as directed vehicles creates the ability to maintain the level of directed vehicles at approximately 30%. The results of the regression with vehicle-model year as the explanatory variables yielded expected results in that older vehicles were significantly more likely to fail the OBD II functional test. This is expected because as vehicles age and are driven more miles, they develop malfunctions leading to emissions failures. A notable finding in this regression model is the observation that odd model-year (2001, 2003, etc.) vehicles are apparently more likely to fail the functional test than the next older even model-year vehicles. This is notable because vehicles that are more than six model-years old are required to undergo biennial inspection. This is based upon actual model-year, meaning that most model-year 2006 vehicles were required to undergo their first smog check inspection in 2012. This pattern would continue with 2004, 2002, 2000 and soon model-year vehicles requiring a biennial inspection in 2012. Odd model-year vehicles, for the most part, likely underwent inspections in 2012 for change of ownership purposes. This suggests that vehicles undergoing a change of ownership are significantly more likely to fail an OBD II functional test. The regression model based on vehicle make yielded expected results as well. Only one vehicle manufacturer, Daewoo, demonstrated a significantly higher likelihood of failing the OBD 56 II functional test. The other manufacturers represented in the dataset were all relatively close to one another in likelihood of failure. As stated in the introduction to this chapter, these results provide substantial foundation for identifying which vehicles, when compared to the other vehicles subject to inspection, have a higher likelihood of failing the OBD II functional test of the current inspection procedure and, subsequently, the revised smog check inspection procedure predicated upon a diagnostic scan of the OBD II system. Predicting Likelihood of Failure of the Visual Inspection The set of regression models to identify which vehicles are highly likely to fail the visual inspection portion of the current smog check inspection procedure utilized the same sets of explanatory variables as the previous set of models: VLT ID numbers, vehicle make, and vehicle model-year. The dependent variable for this set of models was failure of the visual inspection (Fail Visual) portion of the current smog check inspection procedure. The model with VLT ID numbers as the explanatory variables produced meaningful results as well. This model identified 153 VLT ID numbers with a higher likelihood of failing the visual inspection portion of the current smog check inspection procedure. The vehicles identified by this group account for nearly 4% of the volume of inspections in the dataset. The regression model for visual failures with vehicle make as the explanatory variables produced results similar to the same model for OBD II failure, with the exception that the distribution of odds ratio values for the visual inspection was greater than that of the OBD II test failure. This increased variance in the results points to variances of consistency in vehicle durability or maintenance practices. Additionally, the manufacturer Daewoo again demonstrated the highest likelihood of failure of the visual inspection. 57 Finally, the regression with vehicle model-years as the explanatory variables for visual inspection failure also produced results similar to the OBD II regression models. Older vehicles are significantly more likely to fail the visual inspection portion of the current smog check inspection procedure, indicating likely failure of the visual inspection portion of the new smog check inspection procedure. Significance of Findings It is important to point out that this thesis focused upon identifying vehicles highly likely to fail the smog check inspection as compared to all other vehicles in the fleet and subject to inspection. I did not seek to predict failure rates for individual VLT ID numbers, specific makes, or individual calendar years. As such, although each of the regression models produced variables with a significantly higher likelihood of failure, my focus was on the overall ranking as compared to the other variables. The binomial logistic regressions performed in this thesis clearly identified a valid method for ranking vehicles based upon their likelihood of failing either the OBD II functional test or visual inspection portion of the smog check inspection. Those vehicles identified by the VLT ID numbers with the highest likelihood of failure are appropriate for designation as directed vehicles to supplement the model-year 1999 and older group and to maintain a directed vehicle set of approximately 30% of the vehicles subject to inspection. There is less support in the data to employ vehicle manufacturer as a basis for identification as directed vehicles. Although some manufacturers demonstrated a higher likelihood of failure, the results are not significant enough to identify vehicles from any single manufacturer as directed vehicles. The data related to vehicle model-year clearly demonstrates that older vehicles are significantly more likely to fail either test or inspection. However, the ability to identify a 58 sufficient number of specific vehicles through the use of VLT ID numbers is far more equitable and appropriate than a blanket policy of directing all older vehicles for specialized testing. Additionally, as the body of literature indicated, older vehicles are often driven less miles, resulting in less of an emissions impact (Moghadam & Livernois, 2010). A more significant result arising from the regression models utilizing vehicle model-year as the as explanatory variables is the fact that it appears vehicles undergoing inspection for change of ownership purposes are substantially more likely to fail the smog check inspection. Recommendations for Identifying Directed Vehicles The process of directing vehicles for specialized testing based upon the likelihood of the vehicle failing the inspection is a necessary and important component of the smog check inspection program. Directing vehicles based upon broad categories such as age or manufacturer while appropriate in some circumstances, compromises the legitimacy of the program by failing to effectively scrutinize the potentially highest polluting vehicles. The binomial logistic regression analysis employed in this thesis creates a methodology for continued identification of the potentially highest polluting vehicles. This model, or a similar model, should be employed by the Bureau to continue the process of selecting only the appropriate vehicles as directed vehicles. Consideration of the following recommendations would support that endeavor. 1. Improve the identification of like vehicles by assigning VLT ID numbers through a process of decoding the vehicle identification number (VIN) to ascertain the correct make, model, model-year, engine size, transmission type, and other factors. It may be possible to eliminate VLT ID numbers entirely by creating tables based upon the first 11 characters in the VIN, which are the characters used to specifically identify a vehicle. 2. On an annual basis, conduct binomial logistic regression analyses of the initial inspections performed on all vehicles over the preceding two years. Analyzing the initial 59 inspections only will overcome any skewness caused by multiple tests on the same vehicle. Additionally, the initial inspection evaluates the vehicle based upon its most recent operating condition. 3. Failure of the OBD II functional test and failure of the visual inspection are the appropriate dependent variables for the regression analyses described in item two above. The VLT ID numbers or VIN assigned identification categories are the appropriate explanatory variables. 4. Also run binomial logistic regressions with inspection reason (change of ownership or biennial) as the explanatory variables and failure of the OBD II test and visual inspection as the dependent variables. These regressions, also performed on an annual basis, should only evaluate the preceding one year of data, than compared to the prior year’s results to identify variances in likelihood of failure of the inspection based upon inspection reason. 5. At the same time of the regression models described in item four, conduct regression models with model-year as the explanatory variables to evaluate the accuracy of the technician entries of inspection reason. As described previously, it can be reasonably assumed that odd model-year vehicles tested in even calendar years are undergoing inspection for change of ownership purposes, and vice-versa. 6. If the determination is that vehicles undergoing inspection for change of ownership have a significantly higher likelihood of failure of the inspection, those vehicles would be appropriate as directed vehicles. Implementation of this policy may require amendments to the California Health and Safety Code. 7. The data does not support the use of vehicle make (manufacturer) as a criterion for identification of directed vehicles. Although some manufacturers stood out in the regression models, only one, Daewoo, did so with any significance. That level of 60 significance is insufficient to support the use of manufacturer as a criterion. As the body of literature indicates, all manufacturers experience design flaws or failures that lead to failure of one of the tests or inspections. Suggestions for Improved Statistical Modeling In totality, the binomial logistic regression model was quite effective at ranking each set of explanatory variables in order of likelihood of failure. Additionally, performing the regression on the entire dataset rather than a sample of the data produced meaningful and valid results. However, an opportunity for improvement of the model exists. I recommend future research include measures to address the following item. An area for improvement is the assignment of VLT ID numbers. According to the Smog Check Inspection Procedures Manual (Bureau of Automotive Repair, 2009) technicians are required to verify and correct data entries such as the vehicle make, model, engine size, etc. Based upon these entries, the appropriate VLT ID number is assigned to the vehicle under inspection (Bureau of Automotive Repair, 2009). Also during the inspection procedure, the technicians are required to enter the vehicle identification number (VIN), a 17-digit number equating to the serial number of the vehicle. Numerous VIN ‘decoders’ exist on the internet and certainly the software is available for purchase. Programming the vehicle identification database to utilize VIN decoding software and assigning the VLT ID number based upon the VIN would remove the opportunity for human error in data entry. Such a process would ensure the validity of the inspection record for future analysis. Conclusion In this thesis, I sought to develop a methodology for the continued identification of directed vehicles based upon their likelihood of failure as compared to all other vehicles subject to inspection. The six regression models conducted in support of this analysis clearly support the 61 use of binomial logistic regression in the evaluation of the fleet of vehicles. Using failure of the OBD II functional test and visual inspection portion of the current smog check inspection procedure as the dependent variables and VLT ID numbers as the explanatory variables enabled me to identify a highly effective model for identifying directed vehicles without relying upon emissions measurements. Implementation of this model would enable officials to continue directing vehicles based upon the actual likelihood of failure, rather than based upon broad, general categories such as age or manufacturer. Calendar Year Model Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Up to 1976 122,494 103,658 60,561 32,728 19,417 20,887 18,047 17,738 14,427 12,692 11,383 10,359 1977 74,131 62,615 51,622 42,934 35,167 29,512 25,410 21,721 19,862 17,723 15,647 13,908 1978 87,837 73,387 60,139 49,903 40,132 34,196 28,692 24,888 22,636 20,203 17,834 15,816 1979 99,475 83,117 67,845 55,867 45,174 37,637 32,655 27,219 25,408 22,017 19,718 17,416 1980 74,269 59,088 46,495 37,610 29,335 24,067 20,111 17,265 15,342 13,330 11,871 10,422 1981 89,647 71,615 56,079 45,150 35,242 28,677 23,747 20,285 18,008 15,533 13,584 12,120 1982 112,385 89,679 70,099 55,597 42,097 34,588 28,237 23,733 20,776 18,085 15,687 13,788 1983 155,947 124,505 97,110 77,455 58,484 47,579 38,373 32,575 28,076 24,187 20,906 18,138 1984 275,873 225,153 178,454 142,733 108,973 87,417 71,236 59,519 51,447 44,241 37,761 32,610 1985 356,604 296,758 238,247 192,885 148,126 119,140 96,967 80,991 69,618 59,702 50,836 44,045 1986 454,687 385,767 316,763 261,552 205,539 168,222 138,276 115,660 99,174 84,947 72,238 63,072 1987 516,031 445,684 372,070 311,037 244,867 201,643 165,370 137,478 118,046 100,004 84,267 72,691 1988 554,593 491,953 417,575 357,337 285,785 239,788 197,940 166,109 142,785 122,332 102,690 88,396 1989 643,548 585,923 510,698 447,567 367,390 313,440 263,551 223,210 194,515 167,146 141,579 122,694 1990 626,273 586,711 525,079 475,250 403,116 352,860 304,411 262,996 233,044 205,538 175,110 153,305 1991 651,496 618,129 562,694 518,859 449,202 398,576 349,721 308,498 275,209 245,795 211,920 186,416 1992 586,843 563,485 523,753 488,960 432,451 387,325 342,621 304,411 275,520 248,567 215,275 191,201 1993 676,034 639,814 612,988 569,064 515,600 457,472 413,751 364,315 336,360 302,914 267,107 236,487 1994 629,098 769,550 606,536 687,338 539,705 561,923 450,534 455,387 380,486 386,791 312,719 311,089 1995 1,079,486 572,717 956,236 589,536 814,703 524,224 661,575 448,549 551,981 401,845 453,245 333,467 1996 399,765 996,133 452,187 871,049 464,151 714,842 421,534 591,217 386,543 513,482 341,603 426,162 1997 1,191,184 547,867 1,057,711 588,941 932,452 567,689 791,433 512,028 689,331 485,418 594,431 432,426 1998 381,038 1,208,695 554,818 1,088,934 580,712 953,315 568,786 814,094 532,383 740,376 497,446 647,217 1999 278,018 417,603 1,298,216 601,860 1,166,874 615,687 1,030,559 592,924 914,148 591,268 819,753 559,600 Pre 2000 MY Tests 10,116,756 10,019,606 9,693,975 8,590,146 7,964,694 6,920,706 6,483,537 5,622,810 5,415,125 4,844,136 4,504,610 4,012,845 Total Annual Tests 10,441,518 10,597,594 10,737,282 11,050,541 9,018,521 9,360,649 9,319,693 9,554,028 9,825,967 10,571,730 10,776,597 11,290,816 Percentage of Fleet 96.89% 94.55% 90.28% 77.74% 88.31% 73.93% 69.57% 58.85% 55.11% 45.82% 41.80% 35.54% 62 Appendix A: Test Volume for Pre-Model-Year 2000 Vehicles 63 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 1838 1840 1841 1847 1849 1856 3,242 4,693 1,209 1,115 744 1,497 0.04 0.06 0.02 0.02 0.01 0.02 1955 1956 1957 1964 1966 1967 1,828 773 1,520 995 6,218 20,887 0.03 0.01 0.02 0.01 0.09 0.29 1858 1859 1860 1865 1867 1868 6,726 14,344 2,618 5,320 4,446 4,594 0.09 0.2 0.04 0.07 0.06 0.06 1968 1973 1974 1975 1976 1977 2,224 8,422 753 7,591 8,668 392 0.03 0.12 0.01 0.1 0.12 0.01 1869 1876 1885 1886 1887 1892 1,038 2,079 1,323 6,124 729 3,179 0.01 0.03 0.02 0.08 0.01 0.04 1984 1993 1994 1995 2000 2002 3,823 811 5,252 731 4,947 3,485 0.05 0.01 0.07 0.01 0.07 0.05 1894 1895 1901 1903 1910 1912 1913 2,838 704 634 512 774 2,881 9,338 0.04 0.01 0.01 0.01 0.01 0.04 0.13 2003 2009 2011 2018 2020 2021 2022 1,066 1,035 464 430 2,413 9,062 990 0.01 0.01 0.01 0.01 0.03 0.12 0.01 1914 1919 1920 1921 1922 1930 1,675 2,634 591 2,598 3,217 1,790 0.02 0.04 0.01 0.04 0.04 0.02 2027 2029 2030 2038 2047 2048 2,565 4,582 5,763 712 477 2,955 0.04 0.06 0.08 0.01 0.01 0.04 1939 1940 1941 1946 1947 1948 546 3,065 444 7,010 1,027 5,924 0.01 0.04 0.01 0.1 0.01 0.08 2049 2054 2055 2056 2057 2063 434 9,328 959 21,681 3,948 2,602 0.01 0.13 0.01 0.3 0.05 0.04 1949 2,622 0.04 2064 1,030 0.01 64 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 2065 2072 2073 2074 2075 2076 1,160 1,359 406 4,178 27,056 1,737 0.02 0.02 0.01 0.06 0.37 0.02 2192 2198 2200 2209 2210 2211 8,405 581 3,008 397 5,573 628 0.12 0.01 0.04 0.01 0.08 0.01 2081 2083 2084 2092 2101 2102 10,180 14,302 13,783 3,396 759 6,047 0.14 0.2 0.19 0.05 0.01 0.08 2216 2218 2219 2225 2236 2237 2,409 2,990 1,270 1,261 527 3,594 0.03 0.04 0.02 0.02 0.01 0.05 2103 2108 2110 2111 2117 2119 696 4,960 4,955 1,943 763 1,271 0.01 0.07 0.07 0.03 0.01 0.02 2243 2245 2246 2254 2264 2269 829 1,851 1,689 459 786 537 0.01 0.03 0.02 0.01 0.01 0.01 2128 2129 2130 2135 2137 2138 2146 1,108 6,110 384 893 4,448 3,255 1,599 0.02 0.08 0.01 0.01 0.06 0.04 0.02 2270 2271 2272 2273 2274 2279 2281 51,464 2,932 38,394 9,503 397 4,340 1,070 0.71 0.04 0.53 0.13 0.01 0.06 0.01 2156 2162 2163 2164 2165 2171 2,207 10,361 437 16,712 4,551 4,734 0.03 0.14 0.01 0.23 0.06 0.06 2288 2289 2290 2291 2297 2299 1,878 517 5,038 12,154 8,096 21,845 0.03 0.01 0.07 0.17 0.11 0.3 2173 2180 2182 2183 2184 2189 1,230 881 2,885 18,027 683 3,398 0.02 0.01 0.04 0.25 0.01 0.05 2300 2308 2318 2324 2325 2326 9,732 5,333 2,042 23,586 1,408 12,106 0.13 0.07 0.03 0.32 0.02 0.17 2191 7,667 0.11 2327 2,396 0.03 65 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 2333 2335 2342 2344 2345 2351 1,878 389 587 996 2,303 3,164 0.03 0.01 0.01 0.01 0.03 0.04 2488 2489 2495 2506 2507 2513 4,331 969 820 708 1,600 2,152 0.06 0.01 0.01 0.01 0.02 0.03 2353 2354 2362 2372 2378 2379 6,203 1,696 1,582 504 27,491 925 0.09 0.02 0.02 0.01 0.38 0.01 2515 2516 2524 2534 29100 29101 4,977 1,505 2,165 1,333 7,962 1,739 0.07 0.02 0.03 0.02 0.11 0.02 2380 2381 2387 2396 2398 2399 9,674 1,891 1,779 507 907 2,716 0.13 0.03 0.02 0.01 0.01 0.04 29102 29117 29118 29123 29124 29137 4,504 409 405 570 393 2,484 0.06 0.01 0.01 0.01 0.01 0.03 2405 2407 2408 2416 2426 2432 2433 4,415 6,318 1,411 2,578 577 27,244 699 0.06 0.09 0.02 0.04 0.01 0.37 0.01 29138 29141 29142 29143 29144 29145 29146 6,497 1,204 2,847 4,888 415 466 1,402 0.09 0.02 0.04 0.07 0.01 0.01 0.02 2434 2435 2441 2450 2452 2453 9,235 2,061 2,017 364 1,190 2,714 0.13 0.03 0.03 0 0.02 0.04 29148 29149 29152 29155 29158 29159 869 1,505 471 574 642 752 0.01 0.02 0.01 0.01 0.01 0.01 2459 2461 2462 2470 2480 2486 5,756 8,170 2,000 3,137 1,076 13,784 0.08 0.11 0.03 0.04 0.01 0.19 29161 29162 29163 29164 29165 29166 4,070 4,807 1,251 2,340 699 3,897 0.06 0.07 0.02 0.03 0.01 0.05 2487 741 0.01 29167 1,023 0.01 66 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 29168 29169 29170 29171 29173 29174 2,886 1,633 2,474 8,535 483 4,670 0.04 0.02 0.03 0.12 0.01 0.06 29236 29238 29239 29240 29241 29243 1,989 735 674 671 634 678 0.03 0.01 0.01 0.01 0.01 0.01 29175 29179 29180 29181 29182 29184 2,666 3,752 7,197 13,520 527 1,815 0.04 0.05 0.1 0.19 0.01 0.02 29244 29245 29246 29249 29250 29252 2,191 1,935 1,766 1,029 987 756 0.03 0.03 0.02 0.01 0.01 0.01 29188 29189 29190 29191 29192 29193 1,652 1,590 1,507 4,650 499 2,137 0.02 0.02 0.02 0.06 0.01 0.03 29256 29267 29268 29269 29270 29273 509 2,735 487 10,906 806 3,170 0.01 0.04 0.01 0.15 0.01 0.04 29194 29196 29200 29201 29206 29207 29209 607 700 2,356 3,451 709 5,964 2,453 0.01 0.01 0.03 0.05 0.01 0.08 0.03 29274 29277 29281 29282 29283 29285 29286 1,728 831 3,874 2,029 3,394 2,301 2,852 0.02 0.01 0.05 0.03 0.05 0.03 0.04 29211 29212 29213 29214 29215 29217 2,264 1,167 575 814 701 6,194 0.03 0.02 0.01 0.01 0.01 0.09 29287 29289 29290 29291 29292 29294 897 5,552 664 1,608 1,495 611 0.01 0.08 0.01 0.02 0.02 0.01 29219 29220 29222 29223 29224 29229 1,258 1,408 4,124 1,938 461 520 0.02 0.02 0.06 0.03 0.01 0.01 29297 29298 29299 29305 29306 29307 570 2,117 1,003 624 917 1,089 0.01 0.03 0.01 0.01 0.01 0.01 29235 2,425 0.03 29309 1,532 0.02 67 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 29313 29314 29315 29317 29318 29321 1,082 630 524 716 596 750 0.01 0.01 0.01 0.01 0.01 0.01 29366 29367 29370 29372 29373 29375 927 2,198 443 1,097 406 1,171 0.01 0.03 0.01 0.02 0.01 0.02 29322 29323 29324 29325 29326 29327 2,675 822 9,581 14,781 2,776 12,681 0.04 0.01 0.13 0.2 0.04 0.17 29377 29378 29379 29381 29383 29390 509 10,346 820 958 7,013 882 0.01 0.14 0.01 0.01 0.1 0.01 29328 29330 29331 29332 29333 29334 2,588 507 1,702 1,272 902 1,718 0.04 0.01 0.02 0.02 0.01 0.02 29391 29392 29398 29407 29408 29411 2,490 3,998 657 1,429 1,024 1,971 0.03 0.05 0.01 0.02 0.01 0.03 29335 29336 29337 29338 29340 29341 29342 739 1,148 860 553 4,184 3,902 6,991 0.01 0.02 0.01 0.01 0.06 0.05 0.1 29417 29418 29419 29420 29421 29424 29425 1,161 882 1,204 715 2,646 1,216 396 0.02 0.01 0.02 0.01 0.04 0.02 0.01 29345 29346 29347 29349 29350 29352 971 8,659 1,724 2,917 1,333 3,515 0.01 0.12 0.02 0.04 0.02 0.05 29430 29431 29433 29434 29437 29438 37,528 18,244 843 48,573 2,841 4,261 0.52 0.25 0.01 0.67 0.04 0.06 29353 29354 29355 29356 29357 29358 826 1,863 5,470 2,380 1,113 1,184 0.01 0.03 0.08 0.03 0.02 0.02 29439 29440 29441 29442 29443 29444 15,515 1,287 566 2,035 1,165 1,500 0.21 0.02 0.01 0.03 0.02 0.02 29363 884 0.01 29445 666 0.01 68 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 29446 29447 29449 29450 29453 29454 3,292 537 422 1,438 1,567 4,755 0.05 0.01 0.01 0.02 0.02 0.07 29524 29526 29533 29543 29544 29545 2,510 1,093 423 4,945 1,644 3,588 0.03 0.02 0.01 0.07 0.02 0.05 29455 29456 29465 29466 29467 29470 476 779 1,956 679 643 1,664 0.01 0.01 0.03 0.01 0.01 0.02 29546 29547 29548 29551 29553 29555 1,261 3,442 1,182 1,232 1,011 5,550 0.02 0.05 0.02 0.02 0.01 0.08 29471 29477 29485 29487 29488 29489 1,277 730 1,750 2,341 7,033 1,605 0.02 0.01 0.02 0.03 0.1 0.02 29558 29559 29562 29563 29566 29567 598 1,410 5,276 1,661 2,662 3,386 0.01 0.02 0.07 0.02 0.04 0.05 29490 29491 29493 29494 29495 29497 29499 2,375 3,994 707 3,291 3,457 2,302 1,602 0.03 0.05 0.01 0.05 0.05 0.03 0.02 29568 29570 29573 29574 29575 29576 29580 761 753 1,545 3,340 504 1,250 2,167 0.01 0.01 0.02 0.05 0.01 0.02 0.03 29503 29504 29505 29508 29509 29510 459 6,607 3,587 696 1,458 1,616 0.01 0.09 0.05 0.01 0.02 0.02 29582 29583 29584 29585 29586 29587 637 2,267 2,795 4,704 1,002 845 0.01 0.03 0.04 0.06 0.01 0.01 29511 29512 29517 29518 29519 29520 4,951 3,804 815 1,139 2,407 3,897 0.07 0.05 0.01 0.02 0.03 0.05 29588 29591 29592 29593 29598 29599 3,268 4,246 739 1,305 10,944 7,295 0.04 0.06 0.01 0.02 0.15 0.1 29521 2,808 0.04 29600 6,120 0.08 69 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 29602 29603 29604 29605 29606 29607 1,320 13,118 1,904 1,006 6,095 5,442 0.02 0.18 0.03 0.01 0.08 0.07 29661 29662 29663 29664 29670 29671 413 480 538 3,567 452 737 0.01 0.01 0.01 0.05 0.01 0.01 29608 29610 29611 29612 29613 29614 743 5,549 3,097 1,154 1,398 478 0.01 0.08 0.04 0.02 0.02 0.01 29683 29684 29686 29687 29688 29689 4,185 8,602 5,424 12,119 35,579 12,158 0.06 0.12 0.07 0.17 0.49 0.17 29615 29616 29620 29622 29624 29627 1,717 825 2,507 878 1,253 1,146 0.02 0.01 0.03 0.01 0.02 0.02 29691 29692 29695 29696 29697 29698 2,635 3,731 6,190 25,961 6,017 2,081 0.04 0.05 0.08 0.36 0.08 0.03 29628 29629 29630 29631 29632 29633 29634 564 677 1,840 2,757 809 2,291 1,229 0.01 0.01 0.03 0.04 0.01 0.03 0.02 29699 29700 29701 29704 29705 29706 29707 720 4,211 876 21,166 9,365 2,533 6,361 0.01 0.06 0.01 0.29 0.13 0.03 0.09 29635 29637 29639 29641 29645 29646 1,189 1,251 810 580 554 449 0.02 0.02 0.01 0.01 0.01 0.01 29708 29709 29710 29711 29713 29714 997 2,678 1,557 11,120 3,862 512 0.01 0.04 0.02 0.15 0.05 0.01 29653 29654 29656 29657 29658 29659 2,958 1,913 617 1,227 6,958 733 0.04 0.03 0.01 0.02 0.1 0.01 29719 29723 29724 29725 29726 29727 1,825 409 831 8,765 3,101 1,209 0.03 0.01 0.01 0.12 0.04 0.02 29660 1,688 0.02 29728 1,178 0.02 70 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 29729 29730 29731 29732 29733 29734 3,614 1,770 2,094 735 1,449 573 0.05 0.02 0.03 0.01 0.02 0.01 29835 29836 29837 29839 29842 29843 1,719 414 1,278 1,990 650 820 0.02 0.01 0.02 0.03 0.01 0.01 29735 29736 29737 29746 29747 29749 378 432 442 2,016 621 2,702 0.01 0.01 0.01 0.03 0.01 0.04 29844 29847 29848 29852 29853 29854 1,712 1,766 969 2,389 5,358 5,178 0.02 0.02 0.01 0.03 0.07 0.07 29752 29753 29754 29758 29759 29760 1,206 1,658 806 1,389 1,054 2,617 0.02 0.02 0.01 0.02 0.01 0.04 29855 29857 29858 29859 29860 29863 3,765 5,917 876 582 2,683 1,249 0.05 0.08 0.01 0.01 0.04 0.02 29762 29764 29765 29774 29776 29780 29794 2,137 1,297 2,204 597 375 554 600 0.03 0.02 0.03 0.01 0.01 0.01 0.01 29870 29871 29878 29880 29882 29883 29884 1,711 1,588 2,277 1,961 3,612 513 3,829 0.02 0.02 0.03 0.03 0.05 0.01 0.05 29802 29803 29804 29810 29812 29814 1,054 615 4,305 1,532 2,057 1,332 0.01 0.01 0.06 0.02 0.03 0.02 29885 29886 29887 29888 29890 29891 800 663 841 1,049 2,365 644 0.01 0.01 0.01 0.01 0.03 0.01 29816 29817 29819 29820 29827 29828 1,484 603 883 1,313 1,754 1,219 0.02 0.01 0.01 0.02 0.02 0.02 29903 29904 29909 29910 29912 29914 1,301 1,012 4,027 611 394 498 0.02 0.01 0.06 0.01 0.01 0.01 29834 1,854 0.03 29915 1,237 0.02 71 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 29918 29921 29922 29926 29927 29928 457 598 429 586 3,710 690 0.01 0.01 0.01 0.01 0.05 0.01 30010 30011 30012 30013 30014 30015 475 412 1,730 536 2,018 2,948 0.01 0.01 0.02 0.01 0.03 0.04 29931 29932 29935 29937 29939 29940 1,232 802 387 2,146 1,361 675 0.02 0.01 0.01 0.03 0.02 0.01 30016 30017 30018 30019 30021 30022 1,190 3,030 5,324 915 2,017 814 0.02 0.04 0.07 0.01 0.03 0.01 29941 29946 29947 29948 29949 29951 1,890 2,741 541 2,371 1,458 902 0.03 0.04 0.01 0.03 0.02 0.01 30024 30026 30027 30028 30029 30030 1,476 458 430 2,595 1,059 1,296 0.02 0.01 0.01 0.04 0.01 0.02 29952 29965 29966 29968 29974 29979 29985 749 909 367 477 1,563 614 755 0.01 0.01 0.01 0.01 0.02 0.01 0.01 30038 30039 30045 30047 30049 30051 30052 468 854 1,041 507 6,133 449 1,735 0.01 0.01 0.01 0.01 0.08 0.01 0.02 29990 29991 29992 29993 29994 29995 3,799 1,533 411 4,239 5,399 3,726 0.05 0.02 0.01 0.06 0.07 0.05 30053 30059 30060 30063 30064 30072 1,722 1,747 1,487 1,657 952 572 0.02 0.02 0.02 0.02 0.01 0.01 29996 29998 30002 30003 30004 30005 2,209 2,224 953 558 504 667 0.03 0.03 0.01 0.01 0.01 0.01 30080 30084 30085 30086 30087 30089 670 1,172 716 819 701 760 0.01 0.02 0.01 0.01 0.01 0.01 30008 477 0.01 30092 656 0.01 72 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 30096 30097 30099 30101 30102 30103 16,771 7,187 24,924 1,323 1,624 6,203 0.23 0.1 0.34 0.02 0.02 0.09 30168 30169 30172 30173 30175 30176 454 1,806 3,165 1,538 2,127 1,259 0.01 0.02 0.04 0.02 0.03 0.02 30104 30106 30107 30108 30109 30110 662 1,327 615 534 1,447 2,676 0.01 0.02 0.01 0.01 0.02 0.04 30178 30180 30181 30182 30184 30211 513 1,008 1,816 1,204 1,222 1,532 0.01 0.01 0.02 0.02 0.02 0.02 30111 30112 30113 30114 30117 30118 467 873 446 586 704 495 0.01 0.01 0.01 0.01 0.01 0.01 30212 30213 30215 30217 30218 30220 926 741 797 1,363 1,128 1,517 0.01 0.01 0.01 0.02 0.02 0.02 30119 30121 30122 30127 30128 30136 30137 2,031 737 665 927 396 552 502 0.03 0.01 0.01 0.01 0.01 0.01 0.01 30224 30226 30229 30233 30236 30237 30239 587 487 2,690 806 2,065 581 1,200 0.01 0.01 0.04 0.01 0.03 0.01 0.02 30148 30149 30150 30151 30152 30155 1,124 2,283 607 1,253 1,770 1,425 0.02 0.03 0.01 0.02 0.02 0.02 30240 30245 30247 30250 30251 30254 1,319 443 647 552 1,129 1,179 0.02 0.01 0.01 0.01 0.02 0.02 30156 30158 30159 30160 30161 30163 526 1,534 1,583 885 419 886 0.01 0.02 0.02 0.01 0.01 0.01 30256 30258 30259 30260 30262 30263 405 422 1,448 492 794 2,879 0.01 0.01 0.02 0.01 0.01 0.04 30166 1,851 0.03 30264 876 0.01 73 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 30265 30266 30268 30271 30272 30274 449 1,425 776 1,319 415 506 0.01 0.02 0.01 0.02 0.01 0.01 30348 30355 30376 30377 30378 30379 1,475 1,396 2,888 1,576 3,013 11,511 0.02 0.02 0.04 0.02 0.04 0.16 30278 30280 30281 30282 30283 30284 5,874 1,304 1,384 1,996 432 3,568 0.08 0.02 0.02 0.03 0.01 0.05 30380 30382 30383 30386 30388 30389 2,909 1,211 994 2,449 14,314 2,538 0.04 0.02 0.01 0.03 0.2 0.03 30287 30288 30290 30291 30292 30294 1,803 1,288 2,224 4,382 1,053 2,317 0.02 0.02 0.03 0.06 0.01 0.03 30390 30391 30393 30396 30397 30398 514 1,238 711 588 2,316 977 0.01 0.02 0.01 0.01 0.03 0.01 30295 30296 30298 30299 30302 30304 30311 1,067 674 674 450 465 948 539 0.01 0.01 0.01 0.01 0.01 0.01 0.01 30399 30400 30401 30402 30403 30404 30407 1,577 1,129 5,913 3,524 1,701 2,962 1,417 0.02 0.02 0.08 0.05 0.02 0.04 0.02 30312 30315 30317 30318 30319 30320 410 1,091 1,446 717 1,037 607 0.01 0.01 0.02 0.01 0.01 0.01 30408 30409 30411 30412 30417 30418 1,519 2,938 958 603 615 498 0.02 0.04 0.01 0.01 0.01 0.01 30321 30323 30325 30341 30342 30345 775 728 369 1,193 902 1,472 0.01 0.01 0.01 0.02 0.01 0.02 30419 30421 30422 30423 30424 30427 427 1,805 5,161 1,579 615 1,376 0.01 0.02 0.07 0.02 0.01 0.02 30346 3,105 0.04 30428 1,985 0.03 74 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 30429 30430 30431 30440 30442 30445 871 1,559 1,863 1,851 1,576 529 0.01 0.02 0.03 0.03 0.02 0.01 30531 30532 30533 30534 30535 30536 3,420 880 2,103 898 1,100 4,040 0.05 0.01 0.03 0.01 0.02 0.06 30446 30447 30448 30452 30454 30455 384 451 1,056 666 7,936 1,495 0.01 0.01 0.01 0.01 0.11 0.02 30537 30538 30539 30540 30542 30543 591 1,976 5,235 1,131 1,359 1,565 0.01 0.03 0.07 0.02 0.02 0.02 30456 30460 30463 30469 30470 30471 6,679 3,916 698 947 626 1,010 0.09 0.05 0.01 0.01 0.01 0.01 30544 30547 30548 30549 30550 30554 3,534 2,021 1,636 2,286 1,049 603 0.05 0.03 0.02 0.03 0.01 0.01 30472 30475 30482 30490 30494 30495 30496 498 418 452 693 2,049 1,229 1,213 0.01 0.01 0.01 0.01 0.03 0.02 0.02 30555 30556 30557 30558 30559 30560 30561 3,571 13,595 9,196 8,950 3,884 11,174 1,278 0.05 0.19 0.13 0.12 0.05 0.15 0.02 30500 30501 30503 30507 30509 30511 1,698 1,010 2,362 3,596 931 1,885 0.02 0.01 0.03 0.05 0.01 0.03 30562 30563 30565 30566 30569 30573 2,046 6,328 1,646 1,265 567 3,279 0.03 0.09 0.02 0.02 0.01 0.05 30512 30516 30517 30518 30520 30522 1,154 945 432 461 904 1,065 0.02 0.01 0.01 0.01 0.01 0.01 30574 30578 30580 30581 30583 30584 2,596 387 612 4,300 4,368 425 0.04 0.01 0.01 0.06 0.06 0.01 30530 4,244 0.06 30585 6,873 0.09 75 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 30586 30587 30588 30589 30590 30592 5,018 807 1,513 964 993 3,318 0.07 0.01 0.02 0.01 0.01 0.05 30661 30668 30669 30671 30677 30684 415 1,331 471 781 1,422 1,100 0.01 0.02 0.01 0.01 0.02 0.02 30593 30594 30605 30606 30610 30611 693 528 6,563 3,097 2,223 1,543 0.01 0.01 0.09 0.04 0.03 0.02 30686 30689 30693 30694 30695 30696 409 1,392 389 8,351 2,756 781 0.01 0.02 0.01 0.11 0.04 0.01 30614 30617 30618 30620 30622 30624 685 12,066 2,174 377 817 2,145 0.01 0.17 0.03 0.01 0.01 0.03 30697 30698 30699 30700 30701 30703 4,710 5,714 4,320 19,103 2,784 5,310 0.06 0.08 0.06 0.26 0.04 0.07 30626 30627 30628 30635 30636 30637 30642 469 850 395 1,388 7,675 1,777 2,416 0.01 0.01 0.01 0.02 0.11 0.02 0.03 30704 30706 30707 30708 30709 30710 30711 2,211 3,014 618 450 1,538 972 685 0.03 0.04 0.01 0.01 0.02 0.01 0.01 30644 30647 30649 30650 30651 30652 1,596 656 3,242 418 1,583 631 0.02 0.01 0.04 0.01 0.02 0.01 30712 30713 30714 30715 30719 30720 615 743 886 815 1,138 934 0.01 0.01 0.01 0.01 0.02 0.01 30653 30654 30655 30657 30658 30659 2,832 722 4,916 6,164 2,317 680 0.04 0.01 0.07 0.08 0.03 0.01 30721 30722 30723 30724 30725 30726 3,756 1,141 4,584 5,317 3,088 5,201 0.05 0.02 0.06 0.07 0.04 0.07 30660 930 0.01 30727 10,074 0.14 76 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 30728 30729 30730 30731 30734 30735 2,159 468 732 643 2,701 1,018 0.03 0.01 0.01 0.01 0.04 0.01 30803 30807 30810 30811 30813 30814 933 1,157 39,144 16,337 44,043 1,632 0.01 0.02 0.54 0.22 0.6 0.02 30736 30737 30738 30739 30742 30743 718 4,789 1,468 2,642 417 1,467 0.01 0.07 0.02 0.04 0.01 0.02 30817 30818 30819 30820 30823 30825 3,263 5,017 18,001 454 1,348 998 0.04 0.07 0.25 0.01 0.02 0.01 30749 30751 30753 30754 30755 30757 2,769 566 850 612 1,231 568 0.04 0.01 0.01 0.01 0.02 0.01 30826 30827 30828 30829 30830 30831 3,045 6,263 1,102 3,091 1,081 1,408 0.04 0.09 0.02 0.04 0.01 0.02 30759 30760 30761 30762 30763 30764 30768 9,748 572 2,395 2,003 548 3,155 788 0.13 0.01 0.03 0.03 0.01 0.04 0.01 30832 30835 30836 30837 30838 30840 30841 3,209 1,166 1,985 3,738 1,020 802 721 0.04 0.02 0.03 0.05 0.01 0.01 0.01 30770 30771 30773 30774 30775 30776 3,818 3,070 1,325 3,502 2,268 3,174 0.05 0.04 0.02 0.05 0.03 0.04 30842 30846 30854 30855 30858 30860 693 1,113 691 550 1,543 1,360 0.01 0.02 0.01 0.01 0.02 0.02 30777 30793 30794 30797 30798 30799 1,899 1,406 489 2,061 3,399 917 0.03 0.02 0.01 0.03 0.05 0.01 30869 30870 30871 30872 30874 30876 3,842 1,342 1,630 2,985 3,656 3,751 0.05 0.02 0.02 0.04 0.05 0.05 30800 1,137 0.02 30880 555 0.01 77 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 30882 30883 30884 30885 30886 30887 2,639 650 872 2,823 3,092 3,551 0.04 0.01 0.01 0.04 0.04 0.05 30955 30959 30963 30967 30968 30970 1,186 3,355 964 3,894 1,389 2,694 0.02 0.05 0.01 0.05 0.02 0.04 30888 30889 30892 30893 30894 30895 1,247 472 1,228 854 476 9,743 0.02 0.01 0.02 0.01 0.01 0.13 30971 30974 30976 30979 30982 30983 2,132 435 413 549 868 1,879 0.03 0.01 0.01 0.01 0.01 0.03 30897 30898 30901 30902 30903 30904 553 2,677 5,352 3,750 770 5,260 0.01 0.04 0.07 0.05 0.01 0.07 30984 30985 30986 30987 30989 30991 1,382 662 503 767 2,712 656 0.02 0.01 0.01 0.01 0.04 0.01 30905 30908 30909 30910 30911 30912 30918 2,366 522 1,299 2,178 2,445 1,415 1,855 0.03 0.01 0.02 0.03 0.03 0.02 0.03 30992 30993 30994 30996 30997 30998 31000 1,229 1,427 377 455 5,079 949 7,070 0.02 0.02 0.01 0.01 0.07 0.01 0.1 30920 30926 30938 30939 30940 30941 559 560 404 2,565 1,210 6,658 0.01 0.01 0.01 0.04 0.02 0.09 31001 31004 31005 31006 31008 31013 496 1,173 2,154 1,107 1,368 14,679 0.01 0.02 0.03 0.02 0.02 0.2 30943 30944 30945 30948 30949 30950 2,797 3,029 5,324 2,627 676 800 0.04 0.04 0.07 0.04 0.01 0.01 31014 31015 31016 31017 31018 31019 3,707 2,659 2,397 5,266 1,143 6,086 0.05 0.04 0.03 0.07 0.02 0.08 30953 1,562 0.02 31021 2,091 0.03 78 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 31022 31024 31025 31026 31027 31028 1,553 2,979 9,104 702 717 627 0.02 0.04 0.12 0.01 0.01 0.01 31103 31104 31105 31106 31115 31116 1,729 564 794 445 1,192 774 0.02 0.01 0.01 0.01 0.02 0.01 31029 31030 31031 31032 31036 31048 5,098 1,974 678 705 394 655 0.07 0.03 0.01 0.01 0.01 0.01 31123 31124 31125 31126 31127 31128 8,003 3,522 7,149 36,482 14,215 1,973 0.11 0.05 0.1 0.5 0.2 0.03 31049 31050 31051 31052 31053 31054 435 715 1,600 2,087 1,610 1,625 0.01 0.01 0.02 0.03 0.02 0.02 31129 31131 31132 31134 31135 31136 1,874 812 3,374 19,336 3,511 3,261 0.03 0.01 0.05 0.27 0.05 0.04 31056 31057 31060 31067 31068 31073 31074 797 1,098 601 514 464 718 427 0.01 0.02 0.01 0.01 0.01 0.01 0.01 31137 31138 31139 31141 31142 31143 31144 7,882 718 1,565 665 722 8,545 1,439 0.11 0.01 0.02 0.01 0.01 0.12 0.02 31077 31082 31083 31084 31085 31087 452 3,351 1,707 438 531 2,997 0.01 0.05 0.02 0.01 0.01 0.04 31145 31146 31147 31148 31149 31150 8,620 3,914 13,966 8,938 5,887 10,219 0.12 0.05 0.19 0.12 0.08 0.14 31088 31090 31092 31093 31095 31098 8,600 1,356 638 3,081 2,096 1,213 0.12 0.02 0.01 0.04 0.03 0.02 31152 31153 31154 31155 31157 31158 822 4,044 5,221 10,701 3,093 1,042 0.01 0.06 0.07 0.15 0.04 0.01 31099 1,001 0.01 31160 798 0.01 79 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 31162 31163 31165 31166 31167 31169 1,625 1,020 674 7,661 1,041 720 0.02 0.01 0.01 0.11 0.01 0.01 31291 31293 31294 31295 31299 31300 1,411 722 1,762 536 622 1,050 0.02 0.01 0.02 0.01 0.01 0.01 31171 31172 31173 31174 31177 31185 681 3,451 689 4,050 398 1,685 0.01 0.05 0.01 0.06 0.01 0.02 31303 31304 31306 31309 31310 31311 673 388 505 1,820 5,515 3,174 0.01 0.01 0.01 0.02 0.08 0.04 31189 31190 31192 31199 31201 31205 612 1,383 499 3,917 2,382 1,076 0.01 0.02 0.01 0.05 0.03 0.01 31312 31313 31314 31317 31319 31320 2,920 2,476 3,960 2,886 609 549 0.04 0.03 0.05 0.04 0.01 0.01 31211 31213 31217 31219 31246 31248 31252 997 416 709 474 664 499 502 0.01 0.01 0.01 0.01 0.01 0.01 0.01 31336 31337 31345 31347 31349 31350 31351 2,788 1,341 1,673 1,387 2,532 2,927 519 0.04 0.02 0.02 0.02 0.03 0.04 0.01 31255 31259 31264 31265 31268 31272 1,071 1,483 803 1,276 434 539 0.01 0.02 0.01 0.02 0.01 0.01 31352 31356 31370 31372 31377 31386 726 2,190 2,838 1,068 1,280 3,578 0.01 0.03 0.04 0.01 0.02 0.05 31275 31282 31284 31285 31287 31289 717 566 2,018 1,306 694 592 0.01 0.01 0.03 0.02 0.01 0.01 31387 31394 31397 31401 31402 31411 699 408 469 595 525 3,323 0.01 0.01 0.01 0.01 0.01 0.05 31290 3,046 0.04 31412 616 0.01 80 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 31416 31418 31423 31425 31428 31430 733 471 1,585 729 796 3,190 0.01 0.01 0.02 0.01 0.01 0.04 31526 31527 31528 31529 31539 31548 1,962 606 1,540 454 763 5,253 0.03 0.01 0.02 0.01 0.01 0.07 31432 31434 31435 31450 31452 31461 1,349 444 2,459 718 394 2,451 0.02 0.01 0.03 0.01 0.01 0.03 31550 31551 31552 31553 31555 31556 400 525 2,695 455 1,349 1,078 0.01 0.01 0.04 0.01 0.02 0.01 31468 31476 31480 31481 31483 31484 579 3,087 4,738 5,518 798 2,586 0.01 0.04 0.07 0.08 0.01 0.04 31558 31559 31560 31561 31563 31565 736 1,618 979 1,493 540 440 0.01 0.02 0.01 0.02 0.01 0.01 31485 31486 31488 31489 31491 31493 31494 421 1,478 1,328 660 885 805 692 0.01 0.02 0.02 0.01 0.01 0.01 0.01 31586 31591 31592 31593 31594 31596 31600 524 883 1,634 454 478 417 624 0.01 0.01 0.02 0.01 0.01 0.01 0.01 31495 31501 31511 31513 31514 31515 631 526 1,004 1,530 1,455 1,147 0.01 0.01 0.01 0.02 0.02 0.02 31605 31606 31607 31609 31610 31613 2,144 6,428 866 14,328 558 1,033 0.03 0.09 0.01 0.2 0.01 0.01 31516 31517 31519 31520 31522 31523 1,587 2,250 4,161 763 462 1,070 0.02 0.03 0.06 0.01 0.01 0.01 31614 31615 31616 31620 31621 31624 1,973 2,659 6,238 3,703 550 1,442 0.03 0.04 0.09 0.05 0.01 0.02 31525 421 0.01 31625 3,658 0.05 81 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 31626 31627 31630 31631 31633 31639 516 1,010 728 1,113 1,222 6,350 0.01 0.01 0.01 0.02 0.02 0.09 31766 31767 31769 31773 31781 31788 826 1,146 1,779 778 524 2,654 0.01 0.02 0.02 0.01 0.01 0.04 31640 31641 31669 31688 31690 31691 917 415 751 1,370 612 658 0.01 0.01 0.01 0.02 0.01 0.01 31794 31798 31799 31800 31802 31803 1,702 470 998 445 864 677 0.02 0.01 0.01 0.01 0.01 0.01 31693 31695 31698 31702 31703 31704 1,322 1,308 1,520 1,233 1,376 695 0.02 0.02 0.02 0.02 0.02 0.01 31808 31818 31825 31830 31831 31833 1,480 835 814 1,655 429 592 0.02 0.01 0.01 0.02 0.01 0.01 31706 31711 31713 31714 31716 31719 31722 1,904 844 437 1,003 2,634 867 684 0.03 0.01 0.01 0.01 0.04 0.01 0.01 31834 31835 31837 31844 31851 31853 31854 3,306 489 3,329 667 1,181 2,658 9,319 0.05 0.01 0.05 0.01 0.02 0.04 0.13 31723 31724 31727 31728 31731 31736 1,629 968 1,058 463 563 540 0.02 0.01 0.01 0.01 0.01 0.01 31855 31856 31857 31858 31860 31861 2,153 725 714 1,675 1,704 1,018 0.03 0.01 0.01 0.02 0.02 0.01 31737 31738 31743 31744 31762 31763 2,553 1,495 922 696 1,237 491 0.04 0.02 0.01 0.01 0.02 0.01 31862 31864 31865 31868 31872 31873 712 667 406 3,321 1,665 555 0.01 0.01 0.01 0.05 0.02 0.01 31764 2,454 0.03 31874 723 0.01 82 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 31875 31888 31889 31890 31894 31896 390 521 1,375 665 404 965 0.01 0.01 0.02 0.01 0.01 0.01 32004 32005 32006 32009 32010 32011 3,175 1,353 3,339 1,225 2,309 2,978 0.04 0.02 0.05 0.02 0.03 0.04 31917 31924 31925 31926 31927 31934 444 2,298 539 1,233 679 692 0.01 0.03 0.01 0.02 0.01 0.01 32013 32018 32019 32021 32022 32023 847 860 795 754 4,557 641 0.01 0.01 0.01 0.01 0.06 0.01 31935 31936 31937 31939 31943 31956 885 651 628 1,064 686 628 0.01 0.01 0.01 0.01 0.01 0.01 32028 32029 32030 32041 32057 32058 1,716 484 2,544 643 818 8,617 0.02 0.01 0.03 0.01 0.01 0.12 31958 31976 31977 31979 31980 31981 31982 703 1,345 864 573 752 1,860 12,110 0.01 0.02 0.01 0.01 0.01 0.03 0.17 32061 32062 32063 32065 32066 32069 32070 3,615 9,544 7,234 2,300 510 1,337 694 0.05 0.13 0.1 0.03 0.01 0.02 0.01 31983 31988 31989 31990 31991 31992 2,324 1,144 15,001 1,382 1,191 2,293 0.03 0.02 0.21 0.02 0.02 0.03 32073 32079 32090 32091 32094 32099 427 524 1,563 749 587 1,682 0.01 0.01 0.02 0.01 0.01 0.02 31994 31997 31999 32000 32001 32002 880 4,177 2,268 455 2,798 982 0.01 0.06 0.03 0.01 0.04 0.01 32100 32101 32102 32103 32105 32106 2,238 2,132 2,887 1,144 734 669 0.03 0.03 0.04 0.02 0.01 0.01 32003 2,623 0.04 32107 1,767 0.02 83 Appendix B: VLT ID Number Distribution VLT ID 32108 Frequency 3,090 % of Observations 0.04 VLT ID 32174 Frequency 654 % of Observations 0.01 32109 32110 32114 32116 32119 32121 429 388 915 1,208 4,781 2,985 0.01 0.01 0.01 0.02 0.07 0.04 32175 32176 32177 32179 32181 32183 7,726 12,924 1,844 6,715 2,349 1,870 0.11 0.18 0.03 0.09 0.03 0.03 32122 32123 32127 32128 32130 32132 1,467 4,762 1,106 875 2,368 3,709 0.02 0.07 0.02 0.01 0.03 0.05 32185 32187 32188 32191 32192 32193 817 421 1,552 2,300 1,822 11,190 0.01 0.01 0.02 0.03 0.03 0.15 32133 32135 32136 32137 32138 32139 1,090 2,825 1,021 710 3,210 10,468 0.01 0.04 0.01 0.01 0.04 0.14 32194 32195 32196 32198 32199 32208 2,639 4,087 436 559 775 1,651 0.04 0.06 0.01 0.01 0.01 0.02 32140 32141 32142 32143 32144 32145 32147 12,394 6,217 5,163 1,270 741 3,641 484 0.17 0.09 0.07 0.02 0.01 0.05 0.01 32209 32211 32212 32214 32215 32216 32217 859 2,255 1,871 667 536 1,551 2,090 0.01 0.03 0.03 0.01 0.01 0.02 0.03 32148 32152 32160 32161 32163 32164 3,160 557 5,793 2,726 449 4,390 0.04 0.01 0.08 0.04 0.01 0.06 32218 32219 32220 32221 32222 32224 1,110 1,733 1,674 427 5,578 987 0.02 0.02 0.02 0.01 0.08 0.01 32165 32166 32167 32168 32169 32170 456 3,868 3,374 4,341 1,375 978 0.01 0.05 0.05 0.06 0.02 0.01 32225 32226 32228 32229 32232 32233 5,834 10,881 534 2,169 2,698 1,435 0.08 0.15 0.01 0.03 0.04 0.02 84 Appendix B: VLT ID Number Distribution VLT ID 32240 32242 32243 Frequency 2,437 995 395 % of Observations 0.03 0.01 0.01 VLT ID 32292 32295 32297 Frequency 1,148 4,512 1,523 % of Observations 0.02 0.06 0.02 32244 32246 32248 32249 32251 32252 719 4,883 3,816 7,852 3,385 9,665 0.01 0.07 0.05 0.11 0.05 0.13 32311 32314 32316 32317 32318 32319 1,936 643 3,493 7,468 1,524 12,960 0.03 0.01 0.05 0.1 0.02 0.18 32253 32254 32255 32256 32257 32258 1,359 2,763 1,108 2,310 451 3,191 0.02 0.04 0.02 0.03 0.01 0.04 32320 32321 32322 32324 32325 32326 2,387 36,796 828 3,804 1,710 15,569 0.03 0.51 0.01 0.05 0.02 0.21 32259 32260 32261 32264 32265 32267 433 10,759 13,442 611 6,718 1,007 0.01 0.15 0.18 0.01 0.09 0.01 32327 32328 32330 32333 32334 32335 14,913 1,107 3,419 5,927 1,154 1,704 0.2 0.02 0.05 0.08 0.02 0.02 32268 32270 32271 32272 32273 32274 32275 7,126 1,112 506 2,340 824 2,603 5,821 0.1 0.02 0.01 0.03 0.01 0.04 0.08 32336 32337 32338 32340 32341 32342 32343 1,818 2,168 4,655 822 1,618 1,299 1,392 0.02 0.03 0.06 0.01 0.02 0.02 0.02 32276 32277 32278 32279 32281 32283 3,261 1,959 6,009 1,415 775 6,407 0.04 0.03 0.08 0.02 0.01 0.09 32345 32346 32347 32350 32351 32362 14,473 465 1,440 486 591 503 0.2 0.01 0.02 0.01 0.01 0.01 32285 32287 32288 32289 1,222 983 2,871 4,219 0.02 0.01 0.04 0.06 32366 32367 32375 32376 1,718 398 5,340 1,817 0.02 0.01 0.07 0.02 85 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 32377 32378 32380 32382 32384 32385 2,552 2,673 3,650 2,565 4,532 1,627 0.04 0.04 0.05 0.04 0.06 0.02 32434 32436 32437 32438 32441 32446 3,531 1,253 4,487 2,861 1,380 2,035 0.05 0.02 0.06 0.04 0.02 0.03 32386 32387 32388 32389 32390 32391 1,595 1,087 3,272 2,906 1,916 831 0.02 0.01 0.04 0.04 0.03 0.01 32447 32449 32450 32451 32452 32455 1,417 5,204 780 2,200 416 2,169 0.02 0.07 0.01 0.03 0.01 0.03 32392 32393 32396 32398 32399 32400 1,110 3,284 1,514 2,154 10,135 1,723 0.02 0.05 0.02 0.03 0.14 0.02 32456 32457 32458 32461 32467 32469 472 1,230 1,209 1,484 1,780 706 0.01 0.02 0.02 0.02 0.02 0.01 32403 32404 32405 32406 32407 32408 32409 437 4,907 4,467 1,037 12,113 4,831 1,581 0.01 0.07 0.06 0.01 0.17 0.07 0.02 32470 32474 32477 32478 32479 32480 32481 586 1,189 945 718 3,637 2,979 923 0.01 0.02 0.01 0.01 0.05 0.04 0.01 32410 32411 32412 32413 32414 32415 1,625 507 1,263 998 4,952 2,395 0.02 0.01 0.02 0.01 0.07 0.03 32482 32483 32485 32486 32488 32489 3,526 1,025 348 482 1,371 5,956 0.05 0.01 0 0.01 0.02 0.08 32426 32427 32428 32429 32431 32432 1,513 4,468 1,998 2,127 799 806 0.02 0.06 0.03 0.03 0.01 0.01 32490 32491 32492 32493 32494 32495 8,238 1,364 5,728 5,282 722 8,834 0.11 0.02 0.08 0.07 0.01 0.12 32433 2,323 0.03 32496 3,543 0.05 86 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 32497 32498 32499 32501 32502 32503 1,842 1,073 394 2,879 1,153 6,180 0.03 0.01 0.01 0.04 0.02 0.08 32559 32560 32565 32566 32567 32568 1,296 838 1,388 825 819 1,731 0.02 0.01 0.02 0.01 0.01 0.02 32504 32505 32506 32507 32508 32509 13,540 1,174 7,154 2,279 680 6,071 0.19 0.02 0.1 0.03 0.01 0.08 32571 32574 32575 32576 32577 32578 1,014 6,227 1,584 2,953 1,938 4,843 0.01 0.09 0.02 0.04 0.03 0.07 32510 32511 32512 32520 32521 32522 1,213 760 545 895 3,305 4,496 0.02 0.01 0.01 0.01 0.05 0.06 32579 32580 32581 32582 32583 32584 32,402 6,946 2,188 2,445 3,070 921 0.44 0.1 0.03 0.03 0.04 0.01 32523 32525 32526 32527 32528 32530 32532 988 602 734 2,246 720 630 383 0.01 0.01 0.01 0.03 0.01 0.01 0.01 32585 32586 32588 32589 32591 32592 32593 1,179 37,032 3,587 7,725 2,857 3,335 577 0.02 0.51 0.05 0.11 0.04 0.05 0.01 32534 32535 32537 32541 32542 32543 394 1,998 1,817 7,547 485 1,022 0.01 0.03 0.02 0.1 0.01 0.01 32594 32597 32598 32599 32600 32601 8,342 6,332 1,271 7,102 2,323 31,860 0.11 0.09 0.02 0.1 0.03 0.44 32547 32548 32549 32550 32551 32553 1,700 1,875 2,554 3,844 6,654 2,608 0.02 0.03 0.04 0.05 0.09 0.04 32602 32603 32604 32605 32606 32607 1,536 10,112 3,617 11,466 626 3,731 0.02 0.14 0.05 0.16 0.01 0.05 32558 556 0.01 32608 3,520 0.05 87 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 32609 32611 32613 32614 32616 32617 11,631 3,374 1,593 873 2,751 5,561 0.16 0.05 0.02 0.01 0.04 0.08 32707 32708 32709 32717 32718 32732 927 942 562 657 752 613 0.01 0.01 0.01 0.01 0.01 0.01 32621 32622 32623 32624 32625 32633 1,604 3,807 3,643 1,076 787 633 0.02 0.05 0.05 0.01 0.01 0.01 32734 32737 32739 32740 32742 32744 1,302 392 419 473 1,035 916 0.02 0.01 0.01 0.01 0.01 0.01 32634 32635 32636 32638 32639 32640 1,310 1,270 499 1,535 867 553 0.02 0.02 0.01 0.02 0.01 0.01 32746 32749 32750 32751 32752 32753 2,148 1,120 2,059 3,714 998 540 0.03 0.02 0.03 0.05 0.01 0.01 32642 32644 32645 32646 32647 32648 32652 404 1,224 2,165 590 1,137 617 525 0.01 0.02 0.03 0.01 0.02 0.01 0.01 32754 32755 32757 32759 32760 32763 32765 1,239 571 1,730 884 1,840 1,197 2,519 0.02 0.01 0.02 0.01 0.03 0.02 0.03 32653 32654 32655 32656 32658 32661 782 1,132 2,539 2,831 1,835 694 0.01 0.02 0.03 0.04 0.03 0.01 32777 32778 32781 32783 32785 32786 2,476 660 1,475 631 1,429 2,068 0.03 0.01 0.02 0.01 0.02 0.03 32662 32663 32664 32667 32675 32696 1,331 3,284 1,552 869 412 555 0.02 0.05 0.02 0.01 0.01 0.01 32787 32792 32793 32795 32797 32799 508 1,604 2,583 1,635 1,194 636 0.01 0.02 0.04 0.02 0.02 0.01 32705 467 0.01 32801 1,106 0.02 88 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 32802 32803 32812 32813 32815 32816 1,853 1,515 575 1,062 521 4,387 0.03 0.02 0.01 0.01 0.01 0.06 32887 32888 32890 32895 32896 32898 2,962 4,271 2,011 1,355 5,271 640 0.04 0.06 0.03 0.02 0.07 0.01 32817 32818 32823 32824 32826 32827 785 910 1,383 1,174 2,615 1,394 0.01 0.01 0.02 0.02 0.04 0.02 32904 32906 32907 32908 32909 32913 1,011 3,383 1,518 630 1,552 2,613 0.01 0.05 0.02 0.01 0.02 0.04 32830 32831 32835 32836 32837 32838 732 649 505 646 402 518 0.01 0.01 0.01 0.01 0.01 0.01 32918 32919 32923 32926 32940 32946 536 1,442 530 797 531 692 0.01 0.02 0.01 0.01 0.01 0.01 32842 32843 32846 32848 32849 32850 32853 1,362 1,084 3,543 409 1,561 2,724 604 0.02 0.01 0.05 0.01 0.02 0.04 0.01 32947 32949 32950 32951 32952 32955 32956 1,645 1,209 3,747 1,049 11,811 3,050 1,528 0.02 0.02 0.05 0.01 0.16 0.04 0.02 32855 32856 32865 32866 32873 32874 1,230 674 1,881 510 791 1,803 0.02 0.01 0.03 0.01 0.01 0.02 32957 32958 32959 32961 32964 32966 4,462 3,456 582 1,415 2,657 385 0.06 0.05 0.01 0.02 0.04 0.01 32876 32878 32879 32880 32882 32884 967 1,935 868 1,525 1,026 563 0.01 0.03 0.01 0.02 0.01 0.01 32968 32969 32973 32975 32976 32978 492 939 731 404 542 4,049 0.01 0.01 0.01 0.01 0.01 0.06 32886 935 0.01 33001 1,003 0.01 89 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 33002 33003 33004 33005 33007 33009 678 554 525 737 997 1,082 0.01 0.01 0.01 0.01 0.01 0.01 33102 33116 33117 33119 33124 33125 372 394 543 790 1,836 12,237 0.01 0.01 0.01 0.01 0.03 0.17 33011 33012 33013 33015 33016 33017 1,762 640 478 1,030 1,819 833 0.02 0.01 0.01 0.01 0.02 0.01 33126 33127 33130 33131 33132 33133 1,969 1,204 1,810 496 2,671 604 0.03 0.02 0.02 0.01 0.04 0.01 33019 33026 33028 33029 33034 33036 1,387 551 705 2,474 768 947 0.02 0.01 0.01 0.03 0.01 0.01 33134 33135 33136 33138 33139 33141 932 1,395 609 1,426 5,161 2,095 0.01 0.02 0.01 0.02 0.07 0.03 33038 33039 33044 33045 33057 33058 33059 1,538 430 887 799 704 1,553 913 0.02 0.01 0.01 0.01 0.01 0.02 0.01 33142 33143 33144 33148 33150 33151 33157 584 1,399 552 742 696 1,218 434 0.01 0.02 0.01 0.01 0.01 0.02 0.01 33060 33064 33065 33067 33069 33070 463 605 871 462 2,579 492 0.01 0.01 0.01 0.01 0.04 0.01 33158 33166 33170 33175 33179 33182 842 482 476 1,795 389 621 0.01 0.01 0.01 0.02 0.01 0.01 33079 33080 33082 33084 33089 33095 747 508 1,123 601 702 661 0.01 0.01 0.02 0.01 0.01 0.01 33183 33184 33185 33186 33188 33193 514 3,879 1,346 3,301 664 407 0.01 0.05 0.02 0.05 0.01 0.01 33099 514 0.01 33196 431 0.01 90 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 33199 33202 33209 33211 33212 33214 1,189 801 1,387 781 402 1,180 0.02 0.01 0.02 0.01 0.01 0.02 33282 33284 33285 33286 33287 33288 444 6,725 1,422 4,837 12,265 6,453 0.01 0.09 0.02 0.07 0.17 0.09 33215 33216 33217 33218 33219 33222 9,144 2,080 414 444 479 13,238 0.13 0.03 0.01 0.01 0.01 0.18 33293 33295 33296 33300 33303 33305 2,036 717 2,927 1,635 509 622 0.03 0.01 0.04 0.02 0.01 0.01 33224 33225 33227 33228 33230 33232 744 2,002 645 624 2,860 1,472 0.01 0.03 0.01 0.01 0.04 0.02 33321 33322 33324 33328 33332 33333 662 2,763 696 592 886 738 0.01 0.04 0.01 0.01 0.01 0.01 33233 33234 33235 33236 33238 33240 33242 411 1,641 527 5,779 2,115 3,605 1,050 0.01 0.02 0.01 0.08 0.03 0.05 0.01 33335 33336 33337 33339 33340 33341 33342 3,383 3,197 1,271 460 661 403 2,399 0.05 0.04 0.02 0.01 0.01 0.01 0.03 33243 33244 33245 33247 33249 33250 957 3,311 825 471 541 1,239 0.01 0.05 0.01 0.01 0.01 0.02 33343 33344 33351 33357 33360 33361 1,703 856 877 1,318 2,645 580 0.02 0.01 0.01 0.02 0.04 0.01 33251 33255 33256 33264 33266 33271 2,399 1,204 600 388 789 453 0.03 0.02 0.01 0.01 0.01 0.01 33362 33365 33367 33369 33374 33375 1,656 1,053 929 573 1,896 413 0.02 0.01 0.01 0.01 0.03 0.01 33281 769 0.01 33376 3,676 0.05 91 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 33377 33378 33379 33380 33381 33382 3,348 9,072 12,400 1,324 1,406 427 0.05 0.12 0.17 0.02 0.02 0.01 33447 33448 33449 33453 33454 33455 14,877 1,479 2,378 458 4,898 2,484 0.2 0.02 0.03 0.01 0.07 0.03 33383 33385 33386 33388 33390 33391 7,842 2,960 3,354 607 2,837 583 0.11 0.04 0.05 0.01 0.04 0.01 33456 33457 33458 33459 33460 33461 831 4,422 2,750 2,837 4,791 5,440 0.01 0.06 0.04 0.04 0.07 0.07 33393 33394 33398 33406 33407 33408 4,299 1,018 1,447 2,655 3,516 6,638 0.06 0.01 0.02 0.04 0.05 0.09 33462 33463 33464 33467 33468 33469 564 1,929 1,689 629 1,298 1,216 0.01 0.03 0.02 0.01 0.02 0.02 33409 33410 33412 33413 33416 33417 33418 1,941 932 468 4,078 976 3,002 4,261 0.03 0.01 0.01 0.06 0.01 0.04 0.06 33470 33471 33472 33473 33474 33475 33478 808 585 2,411 1,867 1,391 542 1,289 0.01 0.01 0.03 0.03 0.02 0.01 0.02 33420 33421 33425 33426 33428 33429 781 490 4,457 3,596 430 2,402 0.01 0.01 0.06 0.05 0.01 0.03 33479 33480 33482 33483 33484 33485 3,307 5,992 402 1,980 2,414 578 0.05 0.08 0.01 0.03 0.03 0.01 33432 33435 33436 33437 33438 33444 2,442 1,719 2,643 8,389 762 3,324 0.03 0.02 0.04 0.12 0.01 0.05 33490 33493 33494 33500 33501 33503 2,101 2,747 568 2,045 3,820 1,438 0.03 0.04 0.01 0.03 0.05 0.02 33446 774 0.01 33504 5,177 0.07 92 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 33505 33506 33507 33508 33509 33510 2,210 5,076 1,045 2,220 949 2,013 0.03 0.07 0.01 0.03 0.01 0.03 33575 33576 33577 33578 33579 33580 2,049 655 2,451 31,069 11,786 40,408 0.03 0.01 0.03 0.43 0.16 0.55 33511 33512 33513 33514 33515 33520 3,573 7,177 13,129 566 5,707 2,597 0.05 0.1 0.18 0.01 0.08 0.04 33581 33585 33586 33587 33588 33589 2,035 14,865 5,601 23,144 11,636 6,621 0.03 0.2 0.08 0.32 0.16 0.09 33522 33524 33526 33528 33529 33530 8,534 965 401 1,042 462 1,642 0.12 0.01 0.01 0.01 0.01 0.02 33590 33591 33592 33593 33594 33595 8,582 1,192 5,068 1,565 1,545 5,876 0.12 0.02 0.07 0.02 0.02 0.08 33531 33532 33533 33535 33536 33537 33538 425 3,260 3,966 8,113 6,531 2,334 3,608 0.01 0.04 0.05 0.11 0.09 0.03 0.05 33597 33598 33601 33602 33603 33604 33605 614 421 2,712 6,349 558 865 770 0.01 0.01 0.04 0.09 0.01 0.01 0.01 33539 33541 33545 33547 33548 33549 696 404 5,142 1,149 2,568 5,436 0.01 0.01 0.07 0.02 0.04 0.07 33606 33608 33609 33610 33612 33613 1,534 434 1,328 637 9,549 3,673 0.02 0.01 0.02 0.01 0.13 0.05 33552 33553 33557 33558 33559 33560 836 1,102 1,976 669 640 462 0.01 0.02 0.03 0.01 0.01 0.01 33614 33616 33617 33624 33627 33629 1,545 739 486 607 430 573 0.02 0.01 0.01 0.01 0.01 0.01 33570 1,880 0.03 33638 1,534 0.02 93 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 33639 33640 33641 33642 33643 33644 1,391 418 489 909 1,304 2,140 0.02 0.01 0.01 0.01 0.02 0.03 33695 33696 33697 33698 33700 33701 1,689 1,321 553 1,376 2,763 1,935 0.02 0.02 0.01 0.02 0.04 0.03 33645 33646 33647 33648 33649 33651 849 500 960 1,002 897 3,599 0.01 0.01 0.01 0.01 0.01 0.05 33711 33712 33713 33714 33715 33719 3,936 5,386 1,428 2,527 1,781 1,487 0.05 0.07 0.02 0.03 0.02 0.02 33653 33655 33656 33657 33658 33659 2,756 4,520 381 1,025 875 2,812 0.04 0.06 0.01 0.01 0.01 0.04 33720 33722 33723 33724 33728 33729 1,420 739 1,115 464 8,515 1,462 0.02 0.01 0.02 0.01 0.12 0.02 33660 33661 33663 33664 33665 33670 33671 2,192 2,484 4,711 477 1,426 524 1,786 0.03 0.03 0.06 0.01 0.02 0.01 0.02 33730 33736 33737 33739 33740 33741 33743 738 2,743 979 3,238 478 10,742 863 0.01 0.04 0.01 0.04 0.01 0.15 0.01 33674 33680 33681 33682 33684 33686 4,835 647 4,676 7,768 531 1,799 0.07 0.01 0.06 0.11 0.01 0.02 33748 33749 33750 33751 33752 33753 4,650 1,450 1,597 658 616 1,051 0.06 0.02 0.02 0.01 0.01 0.01 33687 33688 33689 33690 33691 33692 3,535 6,185 408 4,664 2,826 952 0.05 0.08 0.01 0.06 0.04 0.01 33754 33758 33762 33763 33766 33768 1,111 883 1,138 954 952 452 0.02 0.01 0.02 0.01 0.01 0.01 33694 3,551 0.05 33771 465 0.01 94 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 33772 33777 33784 33785 33786 33787 802 321 3,551 4,442 1,377 1,054 0.01 0 0.05 0.06 0.02 0.01 33832 33834 33837 33839 33840 33841 972 438 477 512 1,467 1,028 0.01 0.01 0.01 0.01 0.02 0.01 33788 33790 33792 33794 33796 33799 653 1,382 2,506 1,580 698 4,980 0.01 0.02 0.03 0.02 0.01 0.07 33844 33845 33852 33853 33856 33857 1,121 483 5,825 773 377 1,347 0.02 0.01 0.08 0.01 0.01 0.02 33800 33801 33802 33803 33804 33805 28,476 3,300 5,339 2,974 7,103 1,219 0.39 0.05 0.07 0.04 0.1 0.02 33858 33859 33861 33862 33863 33865 2,721 3,141 15,222 8,134 15,721 2,560 0.04 0.04 0.21 0.11 0.22 0.04 33806 33807 33808 33809 33810 33811 33812 5,252 4,439 5,718 1,238 3,077 15,749 852 0.07 0.06 0.08 0.02 0.04 0.22 0.01 33866 33867 33868 33872 33873 33875 33876 2,478 1,208 645 3,238 543 1,338 618 0.03 0.02 0.01 0.04 0.01 0.02 0.01 33813 33814 33815 33816 33817 33818 7,460 1,774 4,774 1,575 973 4,031 0.1 0.02 0.07 0.02 0.01 0.06 33877 33878 33881 33884 33885 33886 2,131 971 402 6,663 783 3,043 0.03 0.01 0.01 0.09 0.01 0.04 33819 33820 33822 33826 33828 33829 783 2,951 1,361 1,264 1,156 2,071 0.01 0.04 0.02 0.02 0.02 0.03 33887 33888 33889 33890 33891 33892 1,253 9,614 20,230 3,940 1,007 2,227 0.02 0.13 0.28 0.05 0.01 0.03 33831 590 0.01 33893 2,007 0.03 95 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 33894 33895 33897 33898 33900 33901 2,378 47,651 2,889 7,157 2,365 453 0.03 0.65 0.04 0.1 0.03 0.01 33955 33956 33958 33959 33960 33965 1,014 465 718 429 574 924 0.01 0.01 0.01 0.01 0.01 0.01 33902 33903 33904 33905 33906 33907 9,048 6,164 2,016 2,194 1,312 4,733 0.12 0.08 0.03 0.03 0.02 0.06 33969 33989 34005 34025 34027 34039 409 605 700 386 1,448 789 0.01 0.01 0.01 0.01 0.02 0.01 33908 33909 33910 33911 33913 33914 1,750 27,986 11,854 20,365 5,201 6,784 0.02 0.38 0.16 0.28 0.07 0.09 34043 34044 34045 34047 34050 34052 895 551 1,714 616 1,530 464 0.01 0.01 0.02 0.01 0.02 0.01 33915 33916 33921 33923 33924 33926 33927 16,352 4,228 1,278 1,061 7,395 4,206 4,683 0.22 0.06 0.02 0.01 0.1 0.06 0.06 34053 34057 34059 34070 34075 34081 34083 414 606 527 1,667 805 431 806 0.01 0.01 0.01 0.02 0.01 0.01 0.01 33928 33932 33933 33936 33937 33941 684 518 793 2,455 528 1,769 0.01 0.01 0.01 0.03 0.01 0.02 34084 34091 34092 34093 34099 34104 444 602 3,540 387 516 415 0.01 0.01 0.05 0.01 0.01 0.01 33942 33944 33950 33951 33952 33953 524 416 513 477 1,335 822 0.01 0.01 0.01 0.01 0.02 0.01 34116 34117 34123 34132 34133 34134 1,779 772 476 663 416 1,048 0.02 0.01 0.01 0.01 0.01 0.01 33954 1,486 0.02 34135 871 0.01 96 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 34136 34154 34159 34160 34166 34168 442 792 487 784 649 1,048 0.01 0.01 0.01 0.01 0.01 0.01 34279 34280 34290 34291 34318 34326 615 420 1,653 385 625 391 0.01 0.01 0.02 0.01 0.01 0.01 34173 34174 34183 34184 34185 34187 703 494 600 1,097 1,395 852 0.01 0.01 0.01 0.02 0.02 0.01 34330 34331 34341 34351 34358 34359 522 1,202 1,028 409 1,981 625 0.01 0.02 0.01 0.01 0.03 0.01 34188 34190 34199 34200 34202 34203 399 2,650 944 535 1,572 975 0.01 0.04 0.01 0.01 0.02 0.01 34363 34365 34366 34369 34370 34374 1,688 408 848 1,070 493 380 0.02 0.01 0.01 0.01 0.01 0.01 34205 34206 34211 34218 34254 34258 34259 518 644 788 523 1,339 1,067 5,515 0.01 0.01 0.01 0.01 0.02 0.01 0.08 34375 34385 34386 34387 34389 34399 34403 478 870 715 1,018 553 1,590 401 0.01 0.01 0.01 0.01 0.01 0.02 0.01 34260 34261 34262 34263 34264 34265 2,317 6,439 945 3,480 700 1,223 0.03 0.09 0.01 0.05 0.01 0.02 34408 34413 34414 34420 34445 34452 1,359 697 432 1,147 431 371 0.02 0.01 0.01 0.02 0.01 0.01 34266 34267 34268 34269 34271 34272 2,919 1,259 664 639 480 455 0.04 0.02 0.01 0.01 0.01 0.01 34455 34456 34457 34459 34461 34462 3,253 541 489 815 929 725 0.04 0.01 0.01 0.01 0.01 0.01 34274 924 0.01 34463 485 0.01 97 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 34465 34466 34467 34469 34470 34475 776 2,014 628 1,647 531 681 0.01 0.03 0.01 0.02 0.01 0.01 34737 34738 34739 34741 34742 34743 2,095 970 5,880 656 1,557 2,429 0.03 0.01 0.08 0.01 0.02 0.03 34478 34509 34512 34523 34525 34531 522 554 985 2,765 532 401 0.01 0.01 0.01 0.04 0.01 0.01 34744 34745 34746 34747 34748 34749 904 761 893 1,082 434 1,705 0.01 0.01 0.01 0.01 0.01 0.02 34547 34549 34551 34553 34554 34558 722 415 925 7,769 2,490 6,355 0.01 0.01 0.01 0.11 0.03 0.09 34751 34752 34753 34756 34758 34759 471 1,346 828 2,453 3,462 1,690 0.01 0.02 0.01 0.03 0.05 0.02 34559 34560 34563 34568 34570 34571 34573 747 1,400 490 884 1,218 480 388 0.01 0.02 0.01 0.01 0.02 0.01 0.01 34760 34761 34762 34763 34764 34765 34766 777 486 692 1,823 457 565 2,645 0.01 0.01 0.01 0.03 0.01 0.01 0.04 34574 34576 34577 34578 34580 34583 454 1,494 1,370 2,156 682 541 0.01 0.02 0.02 0.03 0.01 0.01 34767 34768 34769 34770 34772 34775 769 1,434 2,287 1,077 1,530 387 0.01 0.02 0.03 0.01 0.02 0.01 34585 34588 34590 34591 34592 34597 3,441 483 1,932 630 495 646 0.05 0.01 0.03 0.01 0.01 0.01 34776 34778 34779 34780 34781 34782 1,331 624 1,060 627 9,287 668 0.02 0.01 0.01 0.01 0.13 0.01 34736 3,220 0.04 34783 1,403 0.02 98 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 34784 34785 34789 34790 34791 34792 550 1,621 702 407 940 540 0.01 0.02 0.01 0.01 0.01 0.01 34848 34863 34865 34869 34877 34878 449 1,108 427 498 648 760 0.01 0.02 0.01 0.01 0.01 0.01 34793 34794 34795 34796 34799 34800 562 4,483 20,995 10,261 6,084 2,003 0.01 0.06 0.29 0.14 0.08 0.03 34888 34889 34894 34908 34915 34916 685 580 475 489 2,233 5,511 0.01 0.01 0.01 0.01 0.03 0.08 34801 34803 34807 34808 34810 34811 1,478 539 1,597 918 643 1,685 0.02 0.01 0.02 0.01 0.01 0.02 34922 34924 34926 34928 34931 34932 456 1,201 941 686 913 1,431 0.01 0.02 0.01 0.01 0.01 0.02 34815 34816 34817 34818 34819 34820 34824 3,031 2,493 432 1,451 1,043 395 478 0.04 0.03 0.01 0.02 0.01 0.01 0.01 34941 34942 34944 34958 34959 34962 34972 1,267 1,066 452 779 727 439 2,624 0.02 0.01 0.01 0.01 0.01 0.01 0.04 34825 34828 34829 34830 34834 34837 389 1,083 1,306 538 677 605 0.01 0.01 0.02 0.01 0.01 0.01 34976 34977 34978 34989 34991 34992 390 613 400 463 671 505 0.01 0.01 0.01 0.01 0.01 0.01 34838 34839 34842 34843 34845 34846 419 512 1,101 1,209 715 638 0.01 0.01 0.02 0.02 0.01 0.01 34993 34994 35006 35011 35023 35039 623 3,333 762 805 486 459 0.01 0.05 0.01 0.01 0.01 0.01 34847 2,356 0.03 35041 390 0.01 99 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 35045 35053 35055 35058 35059 35061 548 660 417 1,017 2,336 849 0.01 0.01 0.01 0.01 0.03 0.01 35191 35194 35197 35198 35199 35201 584 1,607 4,293 6,982 2,447 979 0.01 0.02 0.06 0.1 0.03 0.01 35062 35063 35065 35068 35072 35074 975 1,184 897 483 2,561 1,236 0.01 0.02 0.01 0.01 0.04 0.02 35203 35204 35206 35209 35211 35218 1,358 1,654 3,469 888 417 1,049 0.02 0.02 0.05 0.01 0.01 0.01 35077 35078 35079 35080 35087 35090 669 1,392 677 474 971 674 0.01 0.02 0.01 0.01 0.01 0.01 35221 35222 35226 35228 35229 35233 503 1,333 1,059 2,434 390 1,284 0.01 0.02 0.01 0.03 0.01 0.02 35092 35101 35103 35104 35105 35113 35126 447 549 1,346 1,109 1,095 1,086 666 0.01 0.01 0.02 0.02 0.02 0.01 0.01 35237 35243 35245 35248 35249 35251 35257 443 452 439 815 548 1,296 436 0.01 0.01 0.01 0.01 0.01 0.02 0.01 35134 35142 35157 35158 35162 35168 476 441 4,438 1,797 545 692 0.01 0.01 0.06 0.02 0.01 0.01 35276 35277 35278 35279 35290 35291 464 601 431 400 2,413 428 0.01 0.01 0.01 0.01 0.03 0.01 35169 35177 35178 35179 35181 35183 717 631 492 9,674 735 6,711 0.01 0.01 0.01 0.13 0.01 0.09 35292 35293 35294 35295 35296 35297 2,879 916 439 577 546 515 0.04 0.01 0.01 0.01 0.01 0.01 35187 2,704 0.04 35298 1,096 0.02 100 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 35299 35301 50009 50010 50013 50014 639 1,213 394 2,509 400 3,212 0.01 0.02 0.01 0.03 0.01 0.04 51899 51943 51959 51967 51985 52045 1,083 558 3,521 2,415 597 693 0.01 0.01 0.05 0.03 0.01 0.01 50017 50018 50021 50022 50025 50026 577 3,782 690 4,962 659 6,106 0.01 0.05 0.01 0.07 0.01 0.08 52059 52063 52087 52095 52111 52135 2,295 449 1,536 718 398 487 0.03 0.01 0.02 0.01 0.01 0.01 50029 50030 50033 50034 50035 50036 780 5,318 896 8,261 556 4,531 0.01 0.07 0.01 0.11 0.01 0.06 52171 52175 52197 52199 52209 52255 2,390 1,369 1,736 853 1,001 940 0.03 0.02 0.02 0.01 0.01 0.01 50038 50039 50040 50041 50042 50043 50044 3,901 457 606 899 545 4,139 3,308 0.05 0.01 0.01 0.01 0.01 0.06 0.05 52265 52269 52293 52305 52341 52361 52369 2,756 608 2,514 929 570 4,645 2,739 0.04 0.01 0.03 0.01 0.01 0.06 0.04 50045 50046 51497 51549 51565 51581 940 1,397 981 509 3,183 2,358 0.01 0.02 0.01 0.01 0.04 0.03 52387 52461 52469 52493 52505 52517 1,065 3,251 663 2,078 925 487 0.01 0.04 0.01 0.03 0.01 0.01 51687 51703 51715 51751 51767 51775 550 2,407 576 538 2,722 2,208 0.01 0.03 0.01 0.01 0.04 0.03 52545 52561 52565 52585 52645 52649 660 3,697 1,648 1,091 3,571 834 0.01 0.05 0.02 0.01 0.05 0.01 51859 1,695 0.02 52669 3,682 0.05 101 Appendix B: VLT ID Number Distribution VLT ID Frequency % of Observations VLT ID Frequency % of Observations 52677 52685 52693 52701 52705 52719 1,381 682 576 5,065 3,728 1,339 0.02 0.01 0.01 0.07 0.05 0.02 52891 52895 52961 52982 52998 53016 2,235 648 524 389 795 587 0.03 0.01 0.01 0.01 0.01 0.01 52759 52765 52767 52783 52791 52807 641 2,757 613 1,308 839 1,603 0.01 0.04 0.01 0.02 0.01 0.02 53022 53024 53031 53037 53038 53060 1,191 592 599 1,130 540 2,765 0.02 0.01 0.01 0.02 0.01 0.04 52809 52817 52853 52855 52871 52875 1,220 938 1,644 508 1,407 673 0.02 0.01 0.02 0.01 0.02 0.01 53061 53062 53064 53067 53069 53079 1,447 548 1,901 970 856 657 0.02 0.01 0.03 0.01 0.01 0.01 52887 3,191 0.04 Total Observations 7,284,227 102 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 32608 35126 30700 30047 32750 32330 30739 33150 32521 32243 31704 31856 30797 31523 32380 31494 30940 30583 32827 31476 29701 29660 35296 Odds Ratio 10.986 10.988 10.997 10.998 11.002 11.002 11.011 11.017 11.020 11.030 11.034 11.060 11.069 11.079 11.081 11.085 11.095 11.096 11.125 11.128 11.149 11.159 11.182 Model Year 2004 2005 2002 2001 2005 2004 2002 2005 2004 2004 2003 2003 2002 2003 2004 2003 2002 2002 2005 2003 2000 2000 2004 50017 11.186 2002 Multiple Light Truck Makes and Models 35059 30617 33303 29421 51715 33417 11.187 11.187 11.200 11.204 11.206 11.209 2004 2002 2006 2000 2001 2006 50030 11.218 2005 32812 11.227 2005 BMW 325CI COUPE CHRYSLER PT CRUISER (Auto) AUDI A6 GMC SAFARI 2WD PASSENGER DODGE RAM 3500 DIESEL CHEVROLET MALIBU Multiple Heavy Duty Light Truck Makes and Models CHRYSLER CROSSFIRE 1957 11.235 2002 Multiple Station Wagon Makes and Models 30941 30560 30341 11.238 11.242 11.254 2002 2002 2001 Make Model (Transmission Type) Engine Size TOYOTA CHEVROLET FORD FORD CHEVROLET HYUNDAI FORD PONTIAC PONTIAC FORD KIA NISSAN GMC FORD JEEP FORD MAZDA CHEVROLET DODGE FORD TOYOTA SUBARU BMW TUNDRA 2WD SILVERADO 2500 2WD EXPLORER 4DR RANGER SUPER CAB 4DR SILVERADO 1500 2WD ACCENT MUSTANG COUPE GRAND AM GRAND AM E150 SORENTO 2WD FRONTIER 2WD K1500 YUKON 4WD MUSTANG CONVERTIBLE LIBERTY 2WD F150 2WD MX-5 MIATA K1500 SUBURBAN 4WD CARAVAN 2WD ESCAPE RAV4 4WD FORESTER AWD 645CI 003.4 006.0 004.0 004.0 004.8 001.6 004.6 003.4 003.4 005.4 003.5 002.4 005.3 003.8 003.7 004.6 001.8 005.3 003.8 003.0 002.0 002.5 004.4 8 Cylinder Diesel 002.5 002.4 003.1 004.3 005.9 002.2 8 Cylinder Diesel 003.2 6 Cylinder Gasoline 002.0 005.3 002.2 MAZDA CHEVROLET SATURN PROTEGE/PROTEGE 5 C1500 TAHOE 2WD L100/200 103 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29327 32148 29445 30539 32543 30581 Odds Ratio 11.254 11.266 11.268 11.268 11.270 11.275 Model Year 2000 2004 2000 2002 2004 2002 2003 11.276 2003 29273 30831 33453 29736 32389 11.276 11.280 11.285 11.286 11.292 2000 2002 2006 2000 2004 50033 11.299 2006 32824 31143 34924 32787 29612 29285 33382 31515 33810 32259 30557 32367 31303 30698 29294 29707 29317 31319 32199 30584 33119 30103 11.300 11.337 11.339 11.342 11.347 11.374 11.399 11.400 11.405 11.408 11.409 11.410 11.420 11.431 11.436 11.451 11.469 11.476 11.484 11.485 11.487 11.492 2005 2002 2002 2005 2000 2000 2006 2003 2006 2004 2002 2004 2003 2002 2000 2000 2000 2003 2004 2002 2005 2001 CHRYSLER TOYOTA CHEVROLET CHEVROLET OLDSMOBILE DODGE CHEVROLET FORD NISSAN FORD CHEVROLET JAGUAR CHEVROLET FORD DODGE TOYOTA FORD CHEVROLET CHRYSLER CHEVROLET MITSUBISHI HONDA TOWN & COUNTRY 2WD RAV4 2WD SILVERADO 2500 2WD MONTE CARLO ALERO DURANGO 4WD COBALT F150 SUPER CREWCAB QUEST F150 2WD SILVERADO 1500 2WD XJR AVALANCHE 1500 2WD EXPEDITION RAM 1500 4WD TACOMA 2WD E250 ECONOLINE CORVETTE (Auto) TOWN & COUNTRY K1500 TAHOE 4WD LANCER ODYSSEY 30798 11.510 2002 GMC K1500 YUKON DENALI AWD Make Model (Transmission Type) FORD EXPLORER 4DR CHEVROLET CORVETTE HYUNDAI ACCENT (Manual) CADILLAC ESCALADE AWD SATURN L300 CHEVROLET SILVERADO 1500 4WD Multiple Full-Size Passenger Car Makes and Models DODGE DAKOTA 2WD HYUNDAI SONATA CHRYSLER SRT-8 VOLKSWAGEN PASSAT WAGON KIA SEDONA Multiple Light Truck Makes and Models Engine Size 004.0 005.7 001.5 006.0 003.0 005.3 8 Cylinder Gasoline 003.9 002.4 006.1 001.8 003.5 8 Cylinder Diesel 003.8 002.0 006.0 003.4 002.4 005.9 002.0 004.6 003.5 004.6 005.3 004.2 005.3 005.4 005.9 003.4 005.4 005.7 003.8 004.8 002.0 003.5 006.0 104 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 32101 30537 31144 30793 35078 34742 33804 31084 Odds Ratio 11.532 11.534 11.539 11.545 11.553 11.564 11.566 11.575 Model Year 2004 2002 2002 2002 2004 2000 2006 2002 1947 11.576 2002 30618 31014 29281 32209 32398 31005 29729 11.580 11.593 11.599 11.603 11.609 11.609 11.609 2002 2002 2000 2004 2004 2002 2000 BMW X3 CADILLAC ELDORADO TOYOTA RAV4 AWD GMC SIERRA 1500 4WD FORD F250 SUPER DUTY FORD RANGER REG CAB SHORT NISSAN FRONTIER 2WD SATURN LW200 Multiple European Compact Passenger Car Makes and Models CHRYSLER PT CRUISER (Manual) NISSAN ALTIMA DODGE DURANGO 2WD DODGE CARAVAN 2WD LAND ROVER RANGE ROVER MITSUBISHI MONTERO SPORT 2WD VOLKSWAGEN NEW BEETLE (Auto) 1976 11.619 2002 Multiple Full-Size SUV Makes and Models 30901 29699 29219 31594 34926 11.628 11.661 11.671 11.687 11.687 2002 2000 2000 2003 2002 2092 11.690 2004 31528 30710 29270 30389 30874 11.693 11.699 11.707 11.720 11.722 2003 2002 2000 2001 2002 2101 11.726 2004 Multiple Mid-Size Van Makes and Models 6 Cylinder Gasoline 1995 11.729 2002 Multiple Ford Vans and Light Truck Models 006.8 31854 32803 29706 31312 30742 11.732 11.734 11.735 11.742 11.751 2003 2005 2000 2003 2002 Make LEXUS TOYOTA CHEVROLET GMC CHEVROLET Model (Transmission Type) IS 300 MR2 METRO K1500 YUKON XL 4WD SILVERADO 2500 4WD Multiple Mini-Van Makes and Models FORD FORD DODGE TOYOTA JEEP NISSAN CHRYSLER TOYOTA CHEVROLET FORD MUSTANG COUPE F150 REG CAB LONG CARAVAN 2WD ECHO LIBERTY 2WD ALTIMA 300C TACOMA 2WD C1500 SUBURBAN 2WD RANGER PICKUP 4WD Engine Size 003.0 004.6 002.0 005.3 005.4 003.0 004.0 002.2 4 Cylinder Gasoline 002.4 003.5 005.2 003.8 004.4 003.0 002.0 8 Cylinder Gasoline 003.0 001.8 001.3 005.3 006.0 6 Cylinder Gasoline 004.6 004.6 003.8 001.5 003.7 002.5 005.7 002.7 005.3 004.0 105 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29349 30518 51687 31693 33820 34959 31844 29370 34915 32549 32121 30968 32377 29671 30920 30926 Odds Ratio 11.759 11.769 11.774 11.784 11.796 11.798 11.811 11.830 11.840 11.847 11.849 11.850 11.871 11.881 11.884 11.884 Model Year 2000 2002 2001 2003 2006 2002 2003 2000 2002 2004 2004 2002 2004 2000 2002 2002 1867 11.891 2000 33247 30011 32165 31155 31794 29292 32152 33300 31830 11.900 11.904 11.908 11.912 11.933 11.945 11.953 11.955 11.963 2005 2001 2004 2002 2003 2000 2004 2006 2003 VOLKSWAGEN FORD CHEVROLET TOYOTA MERCEDES DODGE CHEVROLET AUDI MITSUBISHI GTI F150 SUPER CAB LONG SILVERADO 1500 AWD TUNDRA 2WD E500 RAM 1500 2WD EXPRESS 1500 A4 QUATTRO ECLIPSE 001.8 005.4 006.0 004.7 005.0 005.9 004.3 002.0 002.4 30775 11.970 2002 GMC C1500 YUKON XL 2WD 005.3 29188 33256 34807 32338 29277 29670 30647 31190 31556 31019 11.971 11.974 11.996 12.005 12.007 12.021 12.023 12.032 12.032 12.035 2000 2005 2005 2004 2000 2000 2002 2002 2003 2002 CHEVROLET VOLKSWAGEN CADILLAC HYUNDAI DODGE SUZUKI DODGE VOLVO GMC NISSAN C2500 SILVERADO 2WD PASSAT CTS SONATA DAKOTA 4WD GRAND VITARA 2WD DAKOTA 4WD S80/S80 EXECUTIVE SIERRA 1500 2WD MAXIMA 006.0 001.8 003.6 002.7 004.7 002.5 004.7 002.9 005.3 003.5 Make Model (Transmission Type) Engine Size FORD BMW CHEVROLET JEEP PONTIAC FORD MITSUBISHI FORD ACURA SATURN CADILLAC MERCEDES JEEP SUZUKI MAZDA MAZDA FOCUS ZX3 3DR M5 SILVERADO 2500 DIESEL LIBERTY 2WD GRAND PRIX E350 ECONOLINE MONTERO SPORT 2WD RANGER SUPER CAB 2DR RSX VUE FWD DEVILLE ML500 GRAND CHEROKEE 4WD GRAND VITARA 4WD 626 B2300 REG CAB SHORT 002.0 004.9 006.6 003.7 003.8 005.4 003.0 004.0 002.0 003.5 004.6 005.0 004.0 002.5 002.5 002.3 Multiple Compact SUV Makes and Models 6 Cylinder Gasoline 106 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30097 29663 30774 34846 29335 Odds Ratio 12.044 12.056 12.061 12.072 12.084 Model Year 2001 2000 2002 2000 2000 2129 12.088 2005 Multiple Full-Size Light Truck Makes and Models 32090 32918 32163 30696 33730 32273 31606 29309 29749 12.101 12.105 12.107 12.126 12.127 12.132 12.134 12.135 12.138 2004 2005 2004 2002 2006 2004 2003 2000 2000 BMW GMC CHEVROLET FORD MERCEDES FORD HONDA FORD VOLVO 30753 12.146 2002 FORD 30538 30064 32843 32823 32099 12.155 12.161 12.176 12.177 12.181 2002 2001 2005 2005 2004 CADILLAC GMC DODGE CHRYSLER BMW 1919 12.189 2001 31304 33264 31495 29656 32138 29149 34093 33255 12.210 12.210 12.216 12.230 12.242 12.244 12.244 12.244 2003 2005 2003 2000 2004 2000 2007 2005 1859 12.252 2000 33117 33815 29487 31630 12.260 12.281 12.292 12.298 2005 2006 2000 2003 Make Model (Transmission Type) Engine Size HONDA SUBARU GMC FORD FORD ACCORD LEGACY AWD C1500 YUKON 2WD EXCURSION F150 REG CAB SHORT 003.0 002.5 005.3 005.4 004.6 8 Cylinder Gasoline 002.5 004.8 004.8 002.0 003.5 004.2 003.0 004.6 002.4 3-SERIES SIERRA 1500 2WD SILVERADO 1500 4WD ESCORT ZX2 C350 FREESTAR WAGON FWD ACCORD CROWN VICTORIA POLICE S70 RANGER SUPER CAB 2DR SH ESCALADE 2WD C1500 YUKON XL 2WD MAGNUM TOWN & COUNTRY 2WD M3 Multiple Compact SUV Makes and Models CHEVROLET VOLKSWAGEN FORD SATURN CHEVROLET BMW CHEVROLET VOLKSWAGEN AVALANCHE 1500 4WD TOUAREG F150 2WD LW SILVERADO 1500 2WD 740IL TRAILBLAZER 2WD NEW BEETLE Multiple Full-Size Light Truck Makes and Models MITSUBISHI NISSAN JEEP HYUNDAI GALANT XTERRA 2WD CHEROKEE 4WD SONATA 003.0 005.3 005.3 005.7 003.3 003.2 4 Cylinder Gasoline 005.3 003.2 005.4 003.0 004.3 004.4 004.2 002.0 8 Cylinder Gasoline 002.4 004.0 004.0 002.4 107 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 32130 31162 32634 31738 29313 30548 31486 29645 32585 31555 30532 29719 29447 34944 29664 30530 30803 Odds Ratio 12.301 12.317 12.321 12.332 12.333 12.343 12.348 12.363 12.363 12.367 12.373 12.402 12.407 12.409 12.410 12.418 12.427 Model Year 2004 2002 2004 2003 2000 2002 2003 2000 2004 2003 2002 2000 2000 2002 2000 2002 2002 2156 12.430 2005 32469 33592 31099 30108 29998 33653 12.443 12.446 12.463 12.482 12.485 12.485 2004 2006 2002 2001 2001 2006 50009 12.489 2000 32853 12.491 2005 DODGE 2002 12.492 2003 Multiple Mid-Size Passenger Car Makes and Models 51775 31989 33447 30096 31526 29420 29446 30558 12.496 12.497 12.519 12.521 12.537 12.539 12.545 12.547 2001 2003 2006 2001 2003 2000 2000 2002 Make Model (Transmission Type) Engine Size CHEVROLET VOLKSWAGEN VOLKSWAGEN LINCOLN FORD CHEVROLET FORD SAAB TOYOTA GMC BUICK VOLKSWAGEN HYUNDAI FORD SUBARU BUICK GMC ASTRO 2WD GOLF TOUAREG TOWN CAR E150 ECONOLINE AVALANCHE 1500 4WD EXPLORER 4DR 9-3 CELICA SIERRA 1500 2WD PARK AVENUE GOLF ELANTRA WAGON F350 SUPER DUTY LEGACY WAGON AWD CENTURY SAFARI 2WD PASSENGER 004.3 002.0 003.2 004.6 004.2 005.3 004.0 002.0 001.8 004.8 003.8 002.0 002.0 006.8 002.5 003.1 004.3 Multiple Full-Size Van Makes and Models MERCURY HUMMER SUBARU HYUNDAI FORD JEEP MONTEREY H3 IMPREZA WAGON AWD ACCENT EXPLORER SPORT TRAC LIBERTY 4WD Multiple Heavy Duty Light Truck Makes and Models FORD TOYOTA CHRYSLER HONDA FORD GMC HYUNDAI CHEVROLET RAM 1500 4WD F350 DIESEL COROLLA PT CRUISER ACCORD MUSTANG COUPE (Auto) SAFARI 2WD CARGO ELANTRA SEDAN C1500 SUBURBAN 2WD 8 Cylinder Gasoline 004.2 003.5 002.5 001.5 004.0 003.7 8 Cylinder Diesel 005.7 6 Cylinder Gasoline 007.3 001.8 002.4 002.3 003.8 004.3 002.0 005.3 108 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 31268 31835 Odds Ratio 12.558 12.558 Model Year 2003 2003 1856 12.558 2000 29697 32826 29138 32314 32170 32019 33459 35058 30403 29439 31028 31386 32137 12.562 12.568 12.574 12.579 12.585 12.592 12.609 12.611 12.613 12.614 12.622 12.633 12.634 2000 2005 2000 2004 2004 2003 2006 2004 2001 2000 2002 2003 2004 2018 12.652 2003 Multiple Compact Light Truck Makes and Models 29754 30169 31293 30659 31831 31265 29759 29683 31021 30714 30533 31698 12.658 12.663 12.670 12.677 12.684 12.685 12.686 12.697 12.707 12.710 12.719 12.727 2000 2001 2003 2002 2003 2003 2000 2000 2002 2002 2002 2003 VOLVO LEXUS CADILLAC DODGE MITSUBISHI BMW VOLVO TOYOTA NISSAN FORD BUICK JEEP 2009 12.727 2003 Multiple Sub-Compact Passenger Car Makes and Models 29753 31026 29375 31432 29358 12.737 12.770 12.777 12.778 12.786 2000 2002 2000 2003 2000 Make Model (Transmission Type) Engine Size BMW MITSUBISHI M3 GALANT 003.2 003.0 Multiple Compact Light Truck Makes and Models TOYOTA DODGE BMW GMC CHEVROLET VOLKSWAGEN DODGE BMW TOYOTA HONDA NISSAN CHRYSLER CHEVROLET VOLVO NISSAN FORD DODGE FORD ECHO CARAVAN 2WD 323I SAFARI MONTE CARLO GTI CHARGER 325CI CONVERTIBLE TACOMA 2WD ODYSSEY XTERRA 2WD PT CRUISER (Auto) BLAZER 4WD V40 GS 300 ESCALADE 2WD RAM 1500 4WD ECLIPSE 745LI V70 AWD 4RUNNER 2WD PATHFINDER 2WD F150 REG CAB SHORT REGAL WRANGLER 4WD S80 SENTRA (Auto) RANGER SUPER CAB 2DR RAM 1500 2WD MUSTANG COUPE (Manual) 4 Cylinder Gasoline 001.5 003.3 002.5 004.3 003.8 001.8 002.7 002.5 002.7 003.5 002.4 002.4 004.3 4 Cylinder Gasoline 001.9 003.0 005.3 004.7 003.0 004.4 002.4 002.7 003.5 005.4 003.8 004.0 4 Cylinder Gasoline 002.9 002.5 004.0 005.7 004.7 109 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID Odds Ratio Model Year 1966 12.795 2002 Multiple Compact Light Truck Makes and Models 30846 32490 32831 30469 29236 33552 34808 35191 31008 12.813 12.821 12.825 12.829 12.837 12.849 12.853 12.857 12.864 2002 2004 2005 2002 2000 2006 2005 2006 2002 ISUZU NISSAN DODGE AUDI CHRYSLER GMC CHEVROLET CHEVROLET MITSUBISHI 50029 12.870 2005 32433 35142 12.884 12.888 2004 2005 2146 12.888 2005 29580 29615 29124 29758 29282 30807 30580 31394 31189 29411 29305 29819 29346 29290 31252 31764 31310 31246 29164 32777 29246 12.927 12.929 12.935 12.938 12.938 12.944 12.956 12.956 12.956 12.963 12.970 12.972 12.982 12.986 12.990 13.000 13.005 13.007 13.015 13.023 13.048 2000 2000 2000 2000 2000 2002 2002 2003 2002 2000 2000 2001 2000 2000 2003 2003 2003 2003 2000 2005 2000 Make Model (Transmission Type) RODEO 2WD ALTIMA DAKOTA 2WD A4 QUATTRO (Auto) 300M CANYON 2WD EXPRESS 2500 COBALT MONTERO SPORT 4WD Multiple Light Truck Makes and Models MAZDA FORD MPV E350 Multiple Mini-Van Makes and Models MERCURY OLDSMOBILE AUDI VOLVO DODGE GMC CHEVROLET CHRYSLER VOLVO GMC FORD BMW FORD DODGE BMW MAZDA CHEVROLET BMW BUICK CHEVROLET CHRYSLER SABLE INTRIGUE A6 QUATTRO V70 DURANGO 2WD SONOMA 2WD SILVERADO 1500 4WD SEBRING CONVERTIBLE S60 K1500 SIERRA 4WD CONTOUR 740I FOCUS 4-DR SEDAN RAM 1500 2WD 330CI PROTEGE/PROTEGE 5 SILVERADO 1500 2WD 325CI REGAL IMPALA TOWN & COUNTRY 2WD Engine Size 6 Cylinder Gasoline 003.2 002.5 004.7 001.8 003.5 002.8 004.8 002.4 003.5 8 Cylinder Diesel 003.0 005.4 6 Cylinder Gasoline 003.0 003.5 002.8 002.4 005.9 004.3 004.8 002.7 002.4 005.3 002.0 004.4 002.0 3.9 003.0 002.0 004.8 002.5 003.8 003.4 003.8 110 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 34817 31006 30404 32432 32849 30994 33803 29700 Odds Ratio 13.060 13.060 13.093 13.095 13.096 13.097 13.100 13.109 Model Year 2006 2002 2001 2004 2005 2002 2006 2000 1948 13.121 2002 29863 32161 13.133 13.135 2001 2004 CHEVROLET EXPRESS 2500 MITSUBISHI MONTERO SPORT 2WD TOYOTA TACOMA 2WD MAZDA MIATA DODGE RAM 1500 2WD MITSUBISHI ECLIPSE NISSAN FRONTIER 2WD TOYOTA RAV4 2WD Multiple Mid-Size Passenger Car Makes and Models CHEVROLET CORVETTE CHEVROLET IMPALA 1876 13.138 2000 Multiple Mini-Van Makes and Models 2180 13.140 2006 Multiple Compact Light Truck Makes and Models 33729 31001 29724 29684 29817 33881 29442 31142 13.154 13.160 13.162 13.165 13.165 13.165 13.178 13.197 2006 2002 2000 2000 2001 2006 2000 2002 2102 13.206 2004 30578 30993 29662 34758 29622 29488 30348 52095 31627 30882 30689 29533 33448 13.224 13.233 13.236 13.245 13.269 13.274 13.275 13.276 13.284 13.309 13.319 13.327 13.336 2002 2002 2000 2002 2000 2000 2001 2003 2003 2002 2002 2000 2006 Make MERCEDES MITSUBISHI VOLKSWAGEN TOYOTA BMW SUZUKI HONDA TOYOTA Model (Transmission Type) C280 MIRAGE JETTA 4RUNNER 2WD 540I RENO PRELUDE MR2 Multiple Full-Size Van Makes and Models CHEVROLET MITSUBISHI SUBARU BMW PLYMOUTH JEEP SUBARU DODGE HYUNDAI JEEP FORD MAZDA CHRYSLER K1500 AVALANCHE 4WD ECLIPSE IMPREZA AWD X5 VOYAGER 2WD GRAND CHEROKEE 2WD FORESTER AWD RAM 3500 DIESEL SANTA FE 2WD WRANGLER 4WD E250 ECONOLINE B2500 REG CAB SHORT SEBRING Engine Size 006.0 003.5 003.4 001.8 004.7 003.0 002.5 002.0 6 Cylinder Gasoline 005.7 003.8 6 Cylinder Gasoline 4 Cylinder Gasoline 003.0 001.5 001.8 003.4 004.4 002.0 002.2 001.8 8 Cylinder Gasoline 005.3 002.4 002.5 003.0 002.4 004.0 002.5 005.9 002.7 004.0 005.4 002.5 002.4 111 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29241 Odds Ratio 13.349 Model Year 2000 2119 13.352 2005 31172 30380 29355 33182 34747 33866 35065 29449 32185 32749 29453 30989 32297 30411 31211 33728 52059 13.353 13.356 13.362 13.367 13.387 13.390 13.391 13.395 13.406 13.407 13.416 13.420 13.459 13.466 13.471 13.472 13.476 2002 2001 2000 2005 2000 2006 2004 2000 2004 2005 2000 2002 2004 2001 2003 2006 2003 VOLKSWAGEN TOYOTA FORD SATURN GMC SUBARU BMW HYUNDAI CHRYSLER CHEVROLET INFINITI MERCURY GMC TOYOTA AUDI MERCEDES CHEVROLET 29730 13.479 2000 32850 34782 33591 30858 31800 30383 32233 31434 33484 33131 31022 34756 30566 30118 32638 32567 30540 13.491 13.494 13.495 13.499 13.505 13.515 13.532 13.538 13.543 13.545 13.547 13.550 13.558 13.574 13.585 13.586 13.604 2005 2003 2006 2002 2003 2001 2004 2003 2006 2005 2002 2002 2002 2001 2004 2004 2002 Make Model (Transmission Type) Engine Size CHRYSLER LHS NEW BEETLE (Auto) CAMRY MUSTANG COUPE (Auto) VUE FWD SONOMA 2WD IMPREZA 745LI SONATA 300M SILVERADO 1500 2WD G20 SABLE ENVOY 4WD TUNDRA 4WD A4 C230 SILVERADO 2500 DIESEL 003.5 6 Cylinder Gasoline 002.0 003.0 003.8 002.2 002.2 002.5 004.4 002.4 003.5 004.3 002.0 003.0 004.2 004.7 001.8 002.5 006.6 VOLKSWAGEN NEW BEETLE (Manual) 002.0 DODGE MITSUBISHI HONDA JAGUAR MERCEDES TOYOTA DODGE DODGE DODGE NISSAN NISSAN BMW CHEVROLET INFINITI VOLVO SUZUKI CADILLAC RAM 1500 2WD ECLIPSE S2000 X-TYPE ML500 CAMRY SOLARA STRATUS RAM 1500 2WD STRATUS FRONTIER 4WD PATHFINDER 4WD 525I CORVETTE G20 S40 AERIO ESCALADE EXT AWD 005.7 003.0 002.2 002.5 005.0 003.0 002.7 003.7 002.4 004.0 003.5 002.5 005.7 002.0 001.9 002.3 006.0 Multiple Station Wagon Makes and Models 112 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 34789 29340 29245 29274 Odds Ratio 13.604 13.608 13.612 13.613 Model Year 2004 2000 2000 2000 1868 13.623 2000 Multiple Full-Size SUV Makes and Models 32961 31205 29171 13.625 13.632 13.653 2005 2003 2000 1840 13.680 2000 31763 13.694 2003 HYUNDAI ACCENT ACURA RSX CHEVROLET ASTRO 2WD PASSENGER Multiple Mid-Size Passenger Car Makes and Models MAZDA MX-5 MIATA 1886 13.696 2000 Multiple Full-Size Van Makes and Models 29614 30799 30482 30445 29163 30943 29726 29291 13.697 13.701 13.702 13.716 13.722 13.731 13.747 13.751 2000 2002 2002 2001 2000 2002 2000 2000 2030 13.753 2003 29169 30708 29544 29688 29323 32267 34803 30229 29518 29267 30017 30855 31134 13.758 13.768 13.775 13.778 13.784 13.793 13.796 13.812 13.837 13.841 13.843 13.846 13.846 2000 2002 2000 2000 2000 2004 2005 2001 2000 2000 2001 2002 2002 1955 13.854 2002 Make Model (Transmission Type) Engine Size CHEVROLET FORD CHRYSLER DODGE EXPRESS 2500 F150 SUPER CAB SHORT TOWN & COUNTRY 2WD DAKOTA 2WD 004.8 004.2 003.3 004.7 OLDSMOBILE GMC AUDI VOLVO BUICK MAZDA VOLKSWAGEN DODGE BRAVADA AWD K1500 YUKON DENALI XL ALLROAD S80/S80 EXECUTIVE PARK AVENUE TRIBUTE JETTA RAM 1500 2WD Multiple Full-Size SUV Makes and Models CADILLAC FORD MAZDA TOYOTA FORD FORD BMW MERCEDES LINCOLN DODGE FORD JAGUAR TOYOTA SEVILLE F150 REG CAB HD LONG MX-5 MIATA CAMRY ESCORT 4DR FOCUS 525I E320 LS CARAVAN 2WD F150 SUPER CREWCAB S-TYPE 4.0 LITRE COROLLA Multiple Station Wagon Makes and Models 8 Cylinder Gasoline 001.6 002.0 004.3 6 Cylinder Gasoline 001.8 8 Cylinder Gasoline 004.3 006.0 002.7 002.9 003.8 003.0 002.8 005.2 8 Cylinder Gasoline 004.6 005.4 001.8 002.2 002.0 002.0 002.5 003.2 003.0 002.4 005.4 004.0 001.8 4 and 6 Cylinder Gas 113 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30686 34958 31320 32160 30460 29155 30694 30590 35293 32559 32622 30463 29283 52209 33553 30112 29695 31015 Odds Ratio 13.857 13.869 13.873 13.877 13.887 13.887 13.890 13.901 13.927 13.946 13.957 13.964 13.982 13.988 13.993 13.993 13.995 13.999 Model Year 2002 2002 2003 2004 2002 2000 2002 2002 2004 2004 2004 2002 2000 2003 2006 2001 2000 2002 1967 13.999 2002 Multiple Full-Size Light Truck Makes and Models 30152 29635 32504 29606 29363 30501 31894 30110 32136 29331 32696 30233 14.013 14.014 14.018 14.023 14.039 14.043 14.043 14.046 14.052 14.055 14.057 14.062 2001 2000 2004 2000 2000 2002 2003 2001 2004 2000 2005 2001 JEEP PONTIAC NISSAN NISSAN FORD BMW PONTIAC HYUNDAI CHEVROLET FORD BMW MERCEDES 29372 14.062 2000 29467 29181 30620 32023 33485 14.064 14.071 14.073 14.088 14.089 2000 2000 2002 2003 2006 Make Model (Transmission Type) Engine Size FORD FORD CHEVROLET CHEVROLET ACURA BMW FORD CHEVROLET BMW SUBARU VOLKSWAGEN AUDI DODGE GMC GMC HYUNDAI TOYOTA NISSAN E150 ECONOLINE EXCURSION CORVETTE (Manual) IMPALA RSX X5 ESCAPE C2500 2WD 545I IMPREZA WRX NEW BEETLE A4 DURANGO 4WD SIERRA 2500 DIESEL CANYON 2WD SANTA FE CELICA FRONTIER 2WD GRAND CHEROKEE 4WD SUNFIRE SENTRA QUEST RANGER REG CAB LONG 330CI CONVERTIBLE SUNFIRE ELANTRA BLAZER 2WD F150 REG CAB LONG 3-SERIES E430 004.6 006.8 005.7 003.4 002.0 004.4 003.0 006.0 004.4 002.0 002.0 001.8 004.7 006.6 003.5 002.7 001.8 002.4 8 Cylinder Gasoline 004.7 002.2 001.8 003.3 002.5 003.0 002.2 002.0 004.3 004.2 002.5 004.3 FORD RANGER SUPER CAB 2DR 002.5 ISUZU CHEVROLET CHRYSLER VOLKSWAGEN DODGE TROOPER SILVERADO 1500 2WD SEBRING COUPE JETTA STRATUS 003.5 005.3 003.0 002.8 002.7 114 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 34976 31027 35011 29298 29182 33593 35006 29547 29470 32387 29392 Odds Ratio 14.097 14.099 14.100 14.119 14.137 14.148 14.165 14.167 14.172 14.203 14.203 Model Year 2002 2002 2003 2000 2000 2006 2003 2000 2000 2004 2000 Make Model (Transmission Type) Engine Size MAZDA NISSAN FORD DODGE CHEVROLET HYUNDAI CHEVROLET MERCEDES JAGUAR KIA GMC PROTEGE 5 SENTRA (Manual) F250 SUPER DUTY STRATUS C1500 SUBURBAN 2WD ACCENT SILVERADO 2500 2WD C230 KOMPRESSOR S-TYPE OPTIMA SIERRA 1500 2WD 002.0 002.5 005.4 002.4 005.7 001.6 006.0 002.3 003.0 002.7 005.3 29725 14.212 2000 VOLKSWAGEN JETTA 002.0 29585 33293 29256 29828 32188 32029 30860 29820 14.220 14.270 14.271 14.272 14.300 14.307 14.320 14.325 2000 2006 2000 2001 2004 2003 2002 2001 MITSUBISHI AUDI DODGE BMW CHRYSLER VOLKSWAGEN JAGUAR BMW GALANT A3 B1500 VAN X5 (Manual) CROSSFIRE NEW BEETLE (Manual) X-TYPE 740IL 002.4 002.0 005.2 004.4 003.2 002.0 003.0 004.4 32375 14.330 2004 JEEP GRAND CHEROKEE 2WD 004.0 32616 30997 32842 29586 34770 31264 29367 30517 29996 32648 31561 33559 31255 14.330 14.336 14.336 14.342 14.346 14.378 14.385 14.392 14.395 14.396 14.397 14.413 14.419 2004 2002 2005 2000 2003 2003 2000 2002 2001 2004 2003 2006 2003 VOLKSWAGEN MITSUBISHI DODGE MITSUBISHI BMW BMW FORD BMW FORD VOLVO GMC GMC BMW JETTA GALANT MAGNUM GALANT 525I 745I RANGER REG CAB SHORT M3 CONVERTIBLE EXPLORER 4DR V40 ENVOY 2WD ENVOY 4WD 330I 2020 14.420 2003 29593 35299 14.432 14.437 2000 2004 001.8 002.4 003.5 003.0 002.5 004.4 002.5 003.2 004.0 001.9 004.2 004.2 003.0 6 Cylinder Gasoline 003.0 004.4 Multiple Mid-Size Light Truck Makes and Models MITSUBISHI BMW MONTERO SPORT 4WD 745I 115 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30325 31563 33595 29338 30512 31372 29269 30251 29587 31874 34739 Odds Ratio 14.447 14.453 14.460 14.461 14.477 14.478 14.480 14.483 14.497 14.506 14.506 Model Year 2001 2003 2006 2000 2002 2003 2000 2001 2000 2003 2000 2198 14.511 2006 30910 32295 29326 33651 30500 30407 34791 31074 29444 32193 30727 30840 30496 31004 29727 30872 31284 35039 32437 29239 30355 31798 30495 31737 14.529 14.529 14.538 14.542 14.555 14.556 14.566 14.576 14.582 14.598 14.637 14.639 14.641 14.703 14.720 14.727 14.729 14.735 14.758 14.769 14.771 14.772 14.775 14.778 2002 2004 2000 2006 2002 2001 2004 2002 2000 2004 2002 2002 2002 2002 2000 2002 2003 2003 2004 2000 2001 2003 2002 2003 LINCOLN GMC FORD JEEP BMW TOYOTA CHEVROLET SAAB HYUNDAI CHRYSLER FORD INFINITI BMW MITSUBISHI VOLKSWAGEN JEEP BUICK HONDA MERCEDES CHRYSLER SUBARU MERCEDES BMW LINCOLN 29352 14.786 2000 FORD Make Model (Transmission Type) Engine Size PORSCHE GMC HYUNDAI FORD BMW CHEVROLET DODGE MERCURY MITSUBISHI OLDSMOBILE CHEVROLET BOXSTER ENVOY 4WD ELANTRA F150 SUPER CAB LONG 745LI TRAILBLAZER 4WD CARAVAN 2WD GRAND MARQUIS MIRAGE ALERO S10 PICKUP 2WD 002.7 004.2 002.0 004.6 004.4 004.2 003.3 004.6 001.5 002.2 002.2 4 Cylinder Gasoline 003.9 004.2 004.0 003.7 003.0 003.4 006.0 002.0 001.5 002.4 002.0 003.5 002.5 003.5 001.8 004.7 003.1 003.0 001.8 002.7 002.5 003.2 002.5 005.4 Multiple Mini-Van Makes and Models LS ENVOY 2WD EXPLORER 2DR LIBERTY 2WD 330CI TACOMA 4WD EXPRESS 3500 9-3 CONVERTIBLE ACCENT (Auto) PT CRUISER FOCUS 4DR SEDAN QX4 2WD 325I MONTERO NEW BEETLE GRAND CHEROKEE 4WD CENTURY ACCORD COUPE C230 KOMPRESSOR CONCORDE LEGACY WAGON AWD ML320 325CI CONVERTIBLE NAVIGATOR MUSTANG CONVERTIBLE (Auto) 003.8 116 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID Odds Ratio Model Year Make Model (Transmission Type) Engine Size 34767 14.799 2002 VOLKSWAGEN NEW BEETLE 001.8 30592 14.802 2002 CHEVROLET S10 PICKUP 2WD 004.3 31525 14.802 2003 FORD MUSTANG CONVERTIBLE 004.6 31548 32802 32778 31626 29620 29324 32179 29223 30547 31345 29613 29583 29627 30376 31370 30652 31596 31024 14.810 14.815 14.816 14.841 14.841 14.846 14.853 14.883 14.889 14.909 14.910 14.919 14.926 14.931 14.932 14.943 14.957 14.965 2003 2005 2005 2003 2000 2000 2004 2000 2002 2003 2000 2000 2000 2001 2003 2002 2003 2002 FORD CHRYSLER CHEVROLET HYUNDAI PLYMOUTH FORD CHEVROLET CHEVROLET CHEVROLET CHEVROLET OLDSMOBILE MITSUBISHI PONTIAC TOYOTA CHEVROLET DODGE GMC NISSAN TAURUS 300 IMPALA SANTA FE 2WD NEON EXPEDITION TRAILBLAZER 2WD S10 PICKUP 2WD FFV AVALANCHE 2WD SILVERADO 1500 4WD ALERO ECLIPSE BONNEVILLE 4RUNNER 2WD TRAILBLAZER 2WD DURANGO 4WD SAFARI 2WD PASSENGER QUEST 003.0 003.5 003.8 002.4 002.0 004.6 004.2 002.2 005.3 005.3 003.4 002.4 003.8 003.4 004.2 005.9 004.3 003.3 1946 14.968 2002 31174 14.981 2002 50021 14.982 2003 30239 30982 32181 30661 30172 32169 29334 32481 52063 30569 30470 14.989 14.990 15.027 15.036 15.041 15.041 15.048 15.069 15.094 15.098 15.100 2001 2002 2004 2002 2001 2004 2000 2004 2003 2002 2002 Multiple Mid-Size Passenger Car Makes and Models VOLKSWAGEN PASSAT Multiple Light Truck Makes and Models MERCEDES MERCURY CHEVROLET DODGE LEXUS CHEVROLET FORD MITSUBISHI CHEVROLET CHEVROLET AUDI S430 COUGAR TRAILBLAZER 4WD RAM 2500 IS 300 MONTE CARLO F150 REG CAB SHORT GALANT SILVERADO 3500 DIESEL G1500/2500 EXPRESS A4 QUATTRO (Manual) 4 Cylinder Gasoline 001.8 8 Cylinder Diesel 004.3 002.5 004.2 005.9 003.0 003.4 004.2 003.8 006.6 005.7 001.8 117 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID Odds Ratio Model Year 2021 15.111 2003 31085 29211 29322 29190 31073 15.116 15.120 15.126 15.131 15.134 2002 2000 2000 2000 2002 SATURN CHEVROLET FORD CHEVROLET SAAB LW300 K1500 TAHOE 4WD ESCORT CAMARO 9-3 003.0 005.3 002.0 005.7 002.0 30944 15.135 2002 MERCEDES C230 KOMPRESSOR 002.3 31435 29408 30754 29289 31287 29418 32224 30842 29148 15.157 15.158 15.159 15.162 15.168 15.172 15.186 15.192 15.196 2003 2000 2002 2000 2003 2000 2004 2002 2000 DODGE GMC FORD DODGE BUICK GMC DODGE ISUZU BMW RAM 1500 2WD JIMMY 4WD RANGER SUPER CAB 2DR NEON REGAL K1500 YUKON 4WD RAM 1500 2WD AXIOM 2WD 740I/740I SPORT 004.7 004.3 002.3 002.0 003.8 005.7 003.7 003.5 004.4 33386 15.208 2006 CHEVROLET COLORADO 2WD 003.5 31025 33193 31185 15.229 15.237 15.255 2002 2005 2002 NISSAN SUBARU VOLVO SENTRA IMPREZA STI S40 001.8 002.5 001.9 34738 15.276 2000 CHEVROLET C2500 SILVERADO 2WD 005.7 33130 31248 31586 30494 29630 30911 34839 15.289 15.294 15.305 15.332 15.335 15.339 15.348 2005 2003 2003 2002 2000 2002 2000 NISSAN BMW GMC BMW PONTIAC LINCOLN CHRYSLER 004.0 002.5 005.3 002.5 002.4 005.4 003.3 29353 15.352 2000 FORD 32759 32742 33547 34752 29217 29485 30611 30554 15.355 15.359 15.378 15.383 15.397 15.397 15.399 15.407 2005 2005 2006 2001 2000 2000 2002 2002 CHEVROLET CHEVROLET GMC FORD CHEVROLET JEEP CHRYSLER CHEVROLET FRONTIER 2WD 325I SIERRA 1500 4WD 325CI GRAND AM NAVIGATOR VOYAGER MUSTANG CONVERTABLE (manual) COLORADO 2WD ASTRO 2WD SIERRA 1500 2WD RANGER SUPER CAB 2DR MALIBU CHEROKEE 2WD 300M C1500 AVALANCHE 2WD Make Model (Transmission Type) Multiple Light Truck Makes and Models Engine Size 8 Cylinder 003.8 002.8 004.3 004.3 003.0 003.1 004.0 003.5 005.3 118 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30826 30155 32558 33446 30736 30240 32815 30292 31116 30015 30871 29207 29794 32480 32334 33139 29497 Odds Ratio 15.422 15.422 15.425 15.427 15.429 15.432 15.433 15.443 15.449 15.458 15.459 15.459 15.463 15.473 15.489 15.518 15.518 Model Year 2002 2001 2004 2006 2002 2001 2005 2001 2002 2001 2002 2000 2001 2004 2004 2005 2000 2000 15.534 2003 29215 31173 33385 30150 30658 31387 15.535 15.548 15.582 15.590 15.592 15.609 2000 2002 2006 2001 2002 2003 CHEVROLET VOLKSWAGEN CHEVROLET JEEP DODGE CHRYSLER LUMINA/MONTECARLO NEW BEETLE (Manual) COLORADO 2WD GRAND CHEROKEE 2WD RAM 1500 2WD PT CRUISER (Manual) 003.8 002.0 002.8 004.7 005.9 002.4 32969 15.620 2005 HYUNDAI SONATA 002.7 33016 30614 29222 29608 32926 30991 33138 29166 34994 15.637 15.647 15.649 15.651 15.676 15.676 15.685 15.696 15.698 2005 2002 2000 2000 2005 2002 2005 2000 2003 KIA CHRYSLER CHEVROLET NISSAN GMC MERCURY NISSAN CADILLAC BMW SEDONA CONCORDE S10 PICKUP 2WD SENTRA ENVOY 2WD VILLAGER QUEST DEVILLE 325I 003.5 003.5 004.3 002.0 004.2 003.3 003.5 004.6 002.5 1838 15.710 2000 Make Model (Transmission Type) Engine Size HYUNDAI JEEP SUBARU CHRYSLER FORD MERCEDES CHRYSLER NISSAN SUZUKI FORD JEEP CHEVROLET AUDI MITSUBISHI HYUNDAI NISSAN KIA ACCENT WRANGLER 4WD IMPREZA STI PACIFICA AWD MUSTANG CONVERTIBLE S500 PACIFICA AWD SENTRA GRAND VITARA XL7 4WD F150 SUPER CAB SHORT GRAND CHEROKEE 4WD SILVERADO 1500 4WD TT COUPE QUATTRO GALANT SANTA FE SENTRA SPORTAGE 001.6 004.4 002.5 003.5 004.6 005.0 003.5 002.0 002.7 005.4 004.0 005.3 001.8 002.4 002.4 001.8 002.0 Multiple Compact Passenger Car Makes and Models Multiple Mid-Size Passenger Car Makes and Models 4 Cylinder Gasoline 4 Cylinder Gasoline 119 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID Odds Ratio Model Year 1903 15.731 2001 Multiple Station Wagon Makes and Models 29424 30278 15.737 15.743 2000 2001 GMC NISSAN 2128 15.747 2005 30759 29441 29691 31631 30854 30885 31128 31163 29696 30544 30880 35023 15.752 15.758 15.761 15.762 15.782 15.786 15.790 15.798 15.800 15.808 15.809 15.810 2002 2000 2000 2003 2002 2002 2002 2002 2000 2002 2002 2003 FORD HONDA TOYOTA HYUNDAI JAGUAR KIA TOYOTA VOLKSWAGEN TOYOTA CHEVROLET JEEP FORD TAURUS PASSPORT 4WD CAMRY SOLARA SONATA S-TYPE 3.0 LITRE RIO CAMRY SOLARA GTI COROLLA ASTRO 2WD PASSENGER WRANGLER E350 ECONOLINE 003.0 003.2 002.2 002.7 003.0 001.5 002.4 001.8 001.8 004.3 004.0 005.4 29214 15.820 2000 CHEVROLET LUMINA/MONTECARLO 003.4 30045 31825 31166 30447 29629 31834 33377 29764 29734 31731 29624 32147 29206 15.826 15.841 15.844 15.850 15.875 15.875 15.881 15.883 15.889 15.911 15.913 15.918 15.921 2001 2003 2002 2001 2000 2003 2006 2001 2000 2003 2000 2004 2000 FORD MERCURY VOLKSWAGEN VOLVO PONTIAC MITSUBISHI CHEVROLET ACURA VOLKSWAGEN LINCOLN PLYMOUTH CHEVROLET CHEVROLET RANGER SUPER CAB 2DR SABLE JETTA V70 FIREBIRD/TRANS AM GALANT SILVERADO 1500 2WD INTEGRA PASSAT (Manual) AVIATOR 2WD VOYAGER 2WD COLORADO 4WD SILVERADO 1500 4WD 2048 15.931 2003 32522 15.938 2004 PONTIAC GRAND PRIX 004.0 003.0 002.0 002.3 005.7 002.4 004.3 001.8 002.8 004.6 003.3 003.5 004.8 8 Cylinder Gasoline 003.8 32482 15.959 2004 MITSUBISHI LANCER 002.0 30534 32028 15.966 15.980 2002 2003 BUICK VOLKSWAGEN RENDEZVOUS AWD NEW BEETLE (Auto) 003.4 002.0 Make Model (Transmission Type) SONOMA 2WD ALTIMA Multiple Compact Light Truck Makes and Models Multiple Full-Size Van Makes and Models Engine Size 6 Cylinder Gasoline 004.3 002.4 6 Cylinder Gasoline 120 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30262 31167 31934 30377 31480 32548 30106 30651 34778 32613 Odds Ratio 16.009 16.015 16.040 16.041 16.045 16.057 16.082 16.087 16.095 16.100 Model Year 2001 2002 2003 2001 2003 2004 2001 2002 2003 2004 1901 16.127 2001 30624 32192 29659 32617 30503 30697 32135 30507 29737 32287 33877 33144 30998 31714 29175 33143 29174 16.133 16.138 16.154 16.160 16.181 16.188 16.198 16.215 16.244 16.252 16.258 16.260 16.268 16.289 16.298 16.299 16.340 2002 2004 2000 2004 2002 2002 2004 2002 2000 2004 2006 2005 2002 2003 2000 2005 2000 CHRYSLER CHRYSLER SATURN VOLKSWAGEN BMW FORD CHEVROLET BMW VOLKSWAGEN GMC SUZUKI NISSAN MITSUBISHI LAND ROVER CHEVROLET NISSAN CHEVROLET SEBRING CONVERTIBLE PACIFICA AWD SW JETTA 330I EXPEDITION AVEO 530I PASSAT WAGON SIERRA 1500 2WD FORENZA XTERRA 4WD GALANT RANGE ROVER BLAZER 4WD XTERRA 2WD BLAZER 2WD 002.7 003.5 001.9 002.0 003.0 004.6 001.6 003.0 002.8 004.3 002.0 004.0 003.0 004.4 004.3 004.0 004.3 32272 16.348 2004 FORD FREESTAR WAGON FWD 003.9 30424 30669 29391 34737 31289 31132 29880 31154 16.354 16.363 16.376 16.376 16.386 16.389 16.389 16.414 2001 2002 2000 2000 2003 2002 2001 2002 VOLKSWAGEN DODGE GMC CHEVROLET BUICK TOYOTA CHEVROLET TOYOTA JETTA (Manual) STRATUS SIERRA 1500 2WD C1500 TAHOE 2WD RENDEZVOUS FWD CELICA K1500 SUBURBAN 4WD TUNDRA 2WD 002.8 003.0 004.8 005.3 003.4 001.8 005.3 003.4 Make Model (Transmission Type) Engine Size MITSUBISHI VOLKSWAGEN SATURN TOYOTA FORD SATURN HONDA DODGE FORD VOLKSWAGEN ECLIPSE SPYDER JETTA VUE AWD 4RUNNER 4WD EXPEDITION VUE FWD PRELUDE DURANGO 4WD E250 ECONOLINE GOLF 003.0 002.8 003.0 003.4 004.6 002.2 002.2 004.7 005.4 002.0 Multiple Station Wagon Makes and Models 4 Cylinder Gasoline 121 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30586 32021 32030 Odds Ratio 16.432 16.444 16.459 Model Year 2002 2003 2003 Make Model (Transmission Type) Engine Size CHEVROLET VOLKSWAGEN VOLKSWAGEN MALIBU JETTA PASSAT 003.1 001.8 001.8 29584 16.460 2000 MITSUBISHI ECLIPSE 003.0 31136 30212 33151 31350 30224 34765 31095 32801 29201 31290 34869 29440 31013 30660 31171 16.474 16.500 16.502 16.515 16.539 16.548 16.548 16.550 16.559 16.591 16.614 16.618 16.625 16.634 16.642 2002 2001 2005 2003 2001 2002 2002 2005 2000 2003 2001 2000 2002 2002 2002 TOYOTA MAZDA PONTIAC CHEVROLET MERCEDES MITSUBISHI SUBARU CHRYSLER CHEVROLET CADILLAC CHEVROLET HONDA NISSAN DODGE VOLKSWAGEN HIGHLANDER 2WD MX-5 MIATA GRAND PRIX MALIBU CLK320 ECLIPSE IMPREZA AWD 300 IMPALA CTS SILVERADO 2500 2WD PASSPORT 2WD ALTIMA RAM 1500 4WD NEW BEETLE 1994 16.643 2002 29212 30828 31311 30086 29297 29170 34863 16.649 16.652 16.654 16.699 16.703 16.725 16.727 2000 2002 2003 2001 2000 2000 2000 CHEVROLET HYUNDAI CHEVROLET GMC DODGE CHEVROLET MERCURY K1500 TAHOE 4WD SANTA FE 2WD SILVERADO 1500 2WD K1500 YUKON XL 4WD STRATUS ASTRO 2WD CARGO VILLAGER 002.4 001.8 003.8 003.1 003.2 003.0 002.0 002.7 003.8 003.2 006.0 003.2 002.5 005.9 001.8 8 Cylinder Gasoline 005.7 002.4 005.3 005.3 002.0 004.3 003.3 31177 16.732 2002 VOLKSWAGEN PASSAT 4MOTION 002.8 29417 32018 31087 30516 31896 29224 29526 32191 34751 16.744 16.744 16.757 16.763 16.791 16.794 16.803 16.818 16.822 2000 2003 2002 2002 2003 2000 2000 2004 2001 GMC VOLKSWAGEN SATURN BMW PONTIAC CHEVROLET MAZDA CHRYSLER FORD K1500 YUKON 4WD GOLF SC M3 VIBE S10 PICKUP 4WD 626 PACIFICA 2WD RANGER REG CAB SHORT 005.3 002.0 001.9 003.2 001.8 004.3 002.5 003.5 003.0 Multiple Full-Size Van Makes and Models 122 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29356 33250 32392 31083 29350 30026 29995 31104 29268 30087 31192 31539 30657 Odds Ratio 16.862 16.863 16.877 16.884 16.893 16.915 16.920 16.932 16.951 16.968 16.975 16.986 16.989 Model Year 2000 2005 2004 2002 2000 2001 2001 2002 2000 2001 2002 2003 2002 Make Model (Transmission Type) Engine Size FORD VOLKSWAGEN KIA SATURN FORD FORD FORD SUBARU DODGE GMC VOLVO FORD DODGE MUSTANG COUPE (Manual) JETTA SPECTRA L300 MUSTANG MUSTANG CONV. (Auto) EXPLORER 2DR LEGACY WAGON AWD CARAVAN 2WD K1500 YUKON XL 4WD V40 RANGER REG CAB SHORT RAM 1500 2WD 003.8 002.0 001.8 003.0 003.8 004.6 004.0 002.5 003.0 006.0 001.9 002.3 004.7 31300 16.994 2003 CHEVROLET ASTRO 2WD PASSENGER 004.3 33019 29407 29752 29455 34784 32485 29383 30237 34928 30737 29378 30677 32003 31165 17.027 17.031 17.043 17.045 17.046 17.048 17.060 17.070 17.087 17.097 17.101 17.111 17.123 17.128 2005 2000 2000 2000 2003 2004 2000 2001 2002 2002 2000 2002 2003 2002 KIA GMC VOLVO INFINITI VOLKSWAGEN MITSUBISHI FORD MERCEDES CHEVROLET FORD FORD FORD TOYOTA VOLKSWAGEN SPECTRA JIMMY 2WD S80 Q45 NEW BEETLE MONTERO WINDSTAR 4DR WAGON ML430 EXPRESS 3500 MUSTANG COUPE (Auto) TAURUS CROWN VICTORIA SIENNA JETTA 1921 17.151 2001 29654 29173 17.164 17.184 2000 2000 SATURN CHEVROLET LS ASTRO AWD PASSENGER 002.0 004.3 002.8 004.1 001.8 003.8 003.8 004.3 005.7 003.8 003.0 004.6 003.0 001.8 6 Cylinder Gasoline 003.0 004.3 33135 17.191 2005 NISSAN PATHFINDER 2WD 004.0 30321 34783 33557 34837 31423 17.233 17.238 17.247 17.247 17.304 2001 2003 2006 2000 2003 PONTIAC VOLKSWAGEN GMC CHEVROLET DODGE SUNFIRE JETTA ENVOY 2WD EXPRESS 3500 DURANGO 2WD 002.2 001.8 004.2 005.7 004.7 Multiple Mid-Size SUV Makes and Models 123 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29220 Odds Ratio 17.312 Model Year 2000 1912 17.326 2001 32813 30749 31213 34916 29286 31669 17.327 17.342 17.367 17.392 17.414 17.423 2005 2002 2003 2002 2000 2003 1964 17.433 2002 Multiple Compact Light Truck Makes and Models 33398 29200 30003 31610 32393 17.473 17.475 17.491 17.491 17.497 2006 2000 2001 2003 2004 CHEVROLET CHEVROLET FORD HONDA KIA 33069 17.524 2005 30245 30511 31691 29196 31402 30397 31309 33421 31868 29582 29209 17.549 17.553 17.560 17.565 17.565 17.566 17.576 17.581 17.582 17.603 17.607 32187 32145 32333 30442 31418 30021 30543 31050 30215 30760 17.614 17.620 17.641 17.655 17.669 17.670 17.677 17.690 17.710 17.717 Make Model (Transmission Type) Engine Size CHEVROLET PRIZM EXPRESS 1500 IMPALA F150 REG CAB LONG CIVIC SPECTRA 001.8 6 Cylinder Gasoline 003.5 2.3 003.0 002.5 002.7 002.5 4 Cylinder Gasoline 004.3 003.4 004.6 002.0 002.0 MERCEDES C230 KOMPRESSOR 001.8 2001 2002 2003 2000 2003 2001 2003 2006 2003 2000 2000 MERCEDES BMW JEEP CHEVROLET CHRYSLER TOYOTA CHEVROLET CHEVROLET NISSAN MITSUBISHI CHEVROLET SLK230 KOMPRESSOR 745I GRAND CHEROKEE 4WD G1500/2500 EXPRESS VOYAGER/TOWN/CTRY RAV4 2WD SILVERADO 1500 2WD MONTE CARLO SENTRA DIAMANTE SEDAN K1500 SUBURBAN 4WD 002.3 004.4 004.7 005.7 003.8 002.0 004.3 003.9 001.8 003.5 005.3 2004 2004 2004 2001 2003 2001 2002 2002 2001 2002 CHRYSLER CHEVROLET HYUNDAI VOLVO DODGE FORD CHEVROLET PONTIAC MAZDA FORD CONCORDE COLORADO 2WD ELANTRA S60 DAKOTA 2WD FOCUS ZX3 3DR ASTRO 2WD CARGO FIREBIRD PROTEGE/PROTEGE MPS TAURUS WAGON 003.5 003.5 002.0 002.4 004.7 002.0 004.3 005.7 002.0 003.0 Multiple Mid-Size Light Truck Makes and Models CHRYSLER FORD AUDI BMW DODGE JAGUAR PACIFICA 2WD RANGER REG CAB SHORT A4 325I INTREPID X-TYPE 124 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 34775 29598 30886 Odds Ratio 17.724 17.731 17.736 Model Year 2003 2000 2002 Make Model (Transmission Type) Engine Size CHEVROLET NISSAN KIA G3500 EXPRESS ALTIMA SEDONA 006.0 002.4 003.5 29180 17.753 2000 CHEVROLET SILVERADO 1500 2WD 004.8 52045 33429 35053 30016 30398 33659 30236 30654 30019 17.762 17.771 17.774 17.784 17.789 17.794 17.798 17.805 17.809 2002 2006 2003 2001 2001 2006 2001 2002 2001 VOLKSWAGEN CHEVROLET VOLKSWAGEN FORD TOYOTA KIA MERCEDES DODGE FORD JETTA TDI TRAILBLAZER 4WD PASSAT F150 SUPER CREWCAB RAV4 4WD RIO ML320 INTREPID FOCUS WAGON 001.9 004.2 002.8 004.6 002.0 001.6 003.2 003.5 002.0 33426 17.815 2006 CHEVROLET TRAILBLAZER 2WD 004.2 30448 32795 29827 29425 33136 29631 30738 32964 29495 31527 29213 34810 31048 29855 29238 17.815 17.831 17.831 17.834 17.842 17.869 17.892 17.906 17.918 17.940 17.964 17.983 17.985 18.000 18.002 2001 2005 2001 2000 2005 2000 2002 2005 2000 2003 2000 2005 2002 2001 2000 VOLVO CHEVROLET BMW GMC NISSAN PONTIAC FORD HYUNDAI KIA FORD CHEVROLET CHEVROLET PONTIAC CHEVROLET CHRYSLER V70 TRAILBLAZER 2WD X5 (Auto) SONOMA 2WD FFV PATHFINDER 4WD GRAND AM MUSTANG COUPE (Manual) ELANTRA SEPHIA/SPECTRA MUSTANG COUPE (Manual) LUMINA/MONTECARLO EXPRESS 3500 BONNEVILLE C1500 SUBURBAN 2WD CIRRUS 002.4 004.2 004.4 002.2 004.0 003.4 003.8 002.0 001.8 003.8 003.1 006.0 003.8 005.3 002.5 50026 18.020 2004 Multiple Domestic Light Truck Makes and Models 29633 30342 29714 29746 29117 29337 29390 18.042 18.050 18.060 18.069 18.090 18.092 18.093 2000 2001 2000 2000 2000 2000 2000 PONTIAC SATURN VOLKSWAGEN VOLVO AUDI FORD GMC GRAND PRIX L300 CABRIO S40 A4 QUATTRO F150 SUPER CAB LONG SIERRA 1500 2WD 8 Cylinder Diesel 003.8 003.0 002.0 001.9 001.8 004.2 004.3 125 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 31259 32144 29379 30027 31425 29466 31147 34768 29588 30734 Odds Ratio 18.098 18.112 18.114 18.139 18.167 18.169 18.171 18.181 18.188 18.206 Model Year 2003 2004 2000 2001 2003 2000 2002 2002 2000 2002 Make Model (Transmission Type) Engine Size BMW CHEVROLET FORD FORD DODGE ISUZU TOYOTA VOLKSWAGEN MITSUBISHI FORD 530I COLORADO 2WD TAURUS WAGON MUSTANG CONV. (Manual) DURANGO 4WD RODEO 4WD SIENNA PASSAT MIRAGE MUSTANG CONVERTIBLE 003.0 002.8 003.0 004.6 004.7 003.2 003.0 002.8 001.8 003.8 32797 18.223 2005 CHEVROLET TRAILBLAZER 4WD 004.2 30247 30650 29168 30085 32221 29454 30644 30535 30010 30588 30649 29885 30628 18.232 18.235 18.237 18.252 18.281 18.283 18.303 18.322 18.351 18.366 18.369 18.391 18.394 2001 2002 2000 2001 2004 2000 2002 2002 2001 2002 2002 2001 2002 MERCEDES DODGE CADILLAC GMC DODGE INFINITI DODGE BUICK FORD CHEVROLET DODGE CHEVROLET CHRYSLER SLK320 DURANGO 2WD ESCALADE 4WD K1500 YUKON 4WD INTREPID I30 DAKOTA 2WD RENDEZVOUS FWD F150 SUPER CAB LONG MONTE CARLO DURANGO 2WD METRO VOYAGER 1885 18.398 2000 31299 29419 30136 29710 30768 30137 31158 31461 31997 32022 30122 30594 18.398 18.407 18.431 18.437 18.446 18.461 18.488 18.526 18.535 18.542 18.595 18.598 2003 2000 2001 2000 2002 2001 2002 2003 2003 2003 2001 2002 003.2 005.9 005.7 006.0 003.5 003.0 004.7 003.4 004.6 003.8 004.7 001.3 003.8 6 Cylinder Gasoline 004.3 005.3 003.0 003.4 004.3 004.0 002.0 002.4 001.8 002.0 003.5 004.3 Multiple Full-Size Van Makes and Models CHEVROLET GMC JAGUAR TOYOTA GMC JAGUAR VOLKSWAGEN DODGE TOYOTA VOLKSWAGEN INFINITI CHEVROLET ASTRO 2WD CARGO K1500 YUKON 4WD S-TYPE TUNDRA 2WD SIERRA 1500 2WD S-TYPE 4.0 LITRE CABRIO STRATUS MATRIX JETTA QX4 4WD S10 PICKUP 4WD 126 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 31412 31988 30549 30814 31377 29909 30555 30731 Odds Ratio 18.598 18.598 18.607 18.609 18.612 18.635 18.637 18.673 Model Year 2003 2003 2002 2002 2003 2001 2002 2002 2047 18.690 2003 30419 30218 29592 33658 30018 29731 30574 30030 31624 18.706 18.712 18.713 18.717 18.727 18.746 18.793 18.810 18.812 2001 2001 2000 2006 2001 2000 2002 2001 2003 VOLKSWAGEN MERCEDES MITSUBISHI KIA FORD VOLKSWAGEN CHEVROLET FORD HYUNDAI GTI C240 MONTERO SPORT 2WD OPTIMA FOCUS 4DR SEDAN PASSAT IMPALA MUSTANG COUPE ACCENT 001.8 002.6 003.5 002.7 002.0 001.8 003.8 004.6 001.6 30820 18.852 2002 HONDA PASSPORT 2WD 003.2 31032 31088 32396 30396 31625 31169 29616 29381 31052 29985 32208 32271 29816 31115 18.856 18.868 18.874 18.896 18.925 18.953 18.998 19.027 19.041 19.078 19.096 19.100 19.105 19.107 2002 2002 2004 2001 2003 2002 2000 2000 2002 2001 2004 2004 2001 2002 OLDSMOBILE SATURN LAND ROVER TOYOTA HYUNDAI VOLKSWAGEN OLDSMOBILE FORD PONTIAC FORD DODGE FORD BMW SUZUKI ALERO SL DISCOVERY MR2 ELANTRA JETTA WAGON SILHOUETTE FWD WINDSTAR 3DR WAGON GRAND AM E250 ECONOLINE CARAVAN 2WD FREESTAR CARGO FWD 530I GRAND VITARA XL7 003.4 001.9 004.6 001.8 002.0 002.0 003.4 003.8 003.4 005.4 003.3 003.9 003.0 002.7 1993 19.113 2002 30409 30014 19.118 19.163 2001 2001 Make Model (Transmission Type) Engine Size DODGE TOYOTA CHEVROLET HONDA CHEVROLET CHRYSLER CHEVROLET FORD CARAVAN 2WD CELICA BLAZER 2WD CIVIC VENTURE FWD PT CRUISER (Auto) BLAZER 4WD FOCUS ZX3 003.8 001.8 004.3 002.0 003.4 002.4 004.3 002.0 Multiple Mid-Size Van Makes and Models Multiple Mid-Size Van Makes and Models TOYOTA FORD TUNDRA 2WD F150 SUPER CAB SHORT 6 Cylinder Gasoline 6 Cylinder Gasoline 004.7 004.6 127 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29194 29774 Odds Ratio 19.206 19.224 Model Year 2000 2001 50022 19.234 2003 31553 34736 29591 29465 31450 19.253 19.273 19.305 19.310 19.326 2003 2000 2000 2000 2003 1941 19.327 2001 34776 19.343 2003 1922 19.349 2001 30562 29747 30536 31077 31943 30217 31054 30418 29860 31551 33376 30423 31090 30918 31351 29974 30220 29123 30128 30587 30550 30593 34932 31092 19.369 19.379 19.384 19.396 19.409 19.435 19.437 19.443 19.462 19.493 19.642 19.683 19.698 19.716 19.745 19.777 19.788 19.846 19.886 19.981 20.079 20.109 20.118 20.119 2002 2000 2002 2002 2003 2001 2002 2001 2001 2003 2006 2001 2002 2002 2003 2001 2001 2000 2001 2002 2002 2002 2002 2002 50034 20.129 2006 Make Model (Transmission Type) Engine Size CHEVROLET AUDI G1500/2500 EXPRESS A4 QUATTRO 004.3 001.8 Multiple Heavy Duty Light Truck Makes and Models GMC CHEVROLET MITSUBISHI ISUZU DODGE SIERRA 1500 2WD C1500 SUBURBAN 2WD MONTERO SPORT 2WD RODEO 2WD STRATUS Multiple Ford Vans and Light Truck Models CHEVROLET S10 PICKUP 2WD Multiple Full-Size SUV Makes and Models CHEVROLET CAMARO VOLVO S40 CADILLAC DEVILLE SAAB 9-5 SUBARU IMPREZA AWD MAZDA TRIBUTE PONTIAC GRAND PRIX VOLKSWAGEN GOLF (Manual) CHEVROLET CAVALIER FORD WINDSTAR CARGO VAN CHEVROLET AVEO VOLKSWAGEN JETTA (Auto) SATURN VUE AWD MAZDA 626 CHEVROLET MONTE CARLO FORD CROWN VICTORIA MERCEDES C320 AUDI A6 QUATTRO ISUZU RODEO 4WD CHEVROLET MONTE CARLO CHEVROLET BLAZER 4WD CHEVROLET S10 PICKUP 2WD FFV CHRYSLER TOWN & COUNTRY 2WD SATURN VUE FWD Multiple Heavy Duty Light Truck Makes and Models 8 Cylinder Diesel 004.3 005.3 003.0 003.2 002.4 006.8 002.2 8 Cylinder Gasoline 005.7 002.0 004.6 002.3 002.0 003.0 003.8 002.0 002.2 003.8 001.6 002.8 003.0 002.0 003.4 004.6 003.2 002.7 003.2 003.4 004.3 002.2 003.8 002.2 8 Cylinder Diesel 128 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID Odds Ratio Model Year Make Model (Transmission Type) Engine Size 30992 20.145 2002 MITSUBISHI DIAMANTE SEDAN 003.5 32746 30182 20.246 20.273 2005 2001 CHEVROLET LINCOLN AVEO TOWN CAR 001.6 004.6 29735 20.286 2000 VOLKSWAGEN PASSAT 4MOTION 002.8 31889 31991 29184 20.289 20.329 20.330 2003 2003 2000 PONTIAC TOYOTA CHEVROLET GRAND AM HIGHLANDER 2WD C1500 TAHOE 2WD 003.4 002.4 004.8 34834 20.338 2000 CHEVROLET SILVERADO 3500 2WD 005.7 34746 29835 29836 30637 31565 30288 29179 34941 30417 34759 31529 20.355 20.406 20.406 20.458 20.458 20.513 20.572 20.606 20.630 20.651 20.659 2000 2001 2001 2002 2003 2001 2000 2002 2001 2002 2003 GMC BUICK BUICK DODGE GMC NISSAN CHEVROLET FORD VOLKSWAGEN CHEVROLET FORD C1500 YUKON XL 2WD LE SABRE PARK AVENUE CARAVAN 2WD ENVOY XL 2WD PATHFINDER 4WD SILVERADO 1500 2WD F250 SUPER DUTY GOLF (Auto) S10 PICKUP 2WD RANGER 2WD 1940 20.676 2001 31767 31056 29814 31306 30008 32760 30089 20.687 20.752 20.767 20.771 20.791 20.806 20.830 2003 2002 2001 2003 2001 2005 2001 1914 20.859 2001 29244 29628 29780 31936 29603 29802 30121 31291 20.867 20.896 20.904 20.937 20.973 20.974 21.000 21.027 2000 2000 2001 2003 2000 2001 2001 2003 005.3 003.8 003.8 003.8 004.2 003.5 004.3 005.4 002.0 002.2 002.3 8 Cylinder Gasoline 001.8 003.4 002.5 004.3 005.4 003.5 004.3 8 Cylinder Gasoline 002.5 003.8 002.7 002.2 003.0 002.5 003.5 004.6 Multiple Full-Size Van Makes and Models MERCEDES PONTIAC BMW CHEVROLET FORD CHEVROLET GMC C230 KOMPRESSOR MONTANA FWD 525I BLAZER 2WD F150 REG CAB SHORT COLORADO 2WD SAFARI 2WD PASSENGER Multiple Light Truck Makes and Models CHRYSLER PONTIAC AUDI SATURN NISSAN BMW INFINITI CADILLAC SEBRING CONVERTIBLE FIREBIRD/TRANS AM A6 QUATTRO VUE FWD (Manual) MAXIMA 325CI QX4 2WD DEVILLE 129 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30589 30825 29804 29306 30386 Odds Ratio 21.028 21.053 21.059 21.110 21.149 Model Year 2002 2002 2001 2000 2001 Make Model (Transmission Type) Engine Size CHEVROLET HYUNDAI BMW FORD TOYOTA PRIZM ACCENT 325I CONTOUR CELICA 1977 21.171 2002 FORD Light Trucks and SUVs 32292 30256 30870 30296 32220 31935 30427 21.210 21.254 21.317 21.365 21.370 21.475 21.480 2004 2001 2002 2001 2004 2003 2001 1895 21.487 2001 31927 30117 29994 31356 32388 34766 30475 21.495 21.522 21.566 21.587 21.635 21.664 21.678 2003 2001 2001 2003 2004 2002 2002 GMC CANYON 2WD MERCURY VILLAGER JEEP GRAND CHEROKEE 2WD OLDSMOBILE ALERO DODGE INTREPID SATURN VUE FWD (Auto) VOLKSWAGEN NEW BEETLE Multiple Full-Size Passenger Car Makes and Models SATURN L300 HYUNDAI XG 300 FORD EXPEDITION CHEVROLET S10 PICKUP 2WD KIA RIO VOLKSWAGEN JETTA AUDI A6 QUATTRO 1913 21.678 2001 Multiple Full-Size Light Truck Makes and Models 30622 21.736 2002 CHRYSLER 001.8 001.5 002.5 002.5 001.8 8 Cylinder Gasoline 003.5 003.3 004.7 002.4 002.7 002.2 001.8 8 Cylinder Gasoline 003.0 003.0 005.4 004.3 001.6 001.8 003.0 8 Cylinder Gasoline 002.7 2049 21.738 2003 FORD 31875 29910 29844 31958 30388 30315 29524 30573 34753 29803 29733 21.767 21.807 21.839 21.854 21.861 21.891 21.909 21.910 21.916 21.947 21.954 2003 2001 2001 2003 2001 2001 2000 2002 2001 2001 2000 OLDSMOBILE CHRYSLER CHEVROLET SUZUKI TOYOTA PONTIAC MAZDA CHEVROLET MITSUBISHI BMW VOLKSWAGEN SEBRING SEDAN Multiple Vans and Light Trucks ALERO PT CRUISER (Manual) ASTRO 2WD PASSENGER AERIO COROLLA GRAND AM 626 IMPALA ECLIPSE 325CI CONVERTIBLE PASSAT (Auto) 006.8 003.4 002.4 004.3 002.0 001.8 002.4 002.0 003.4 003.0 002.5 002.8 130 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30909 30084 34908 Odds Ratio 21.957 21.967 21.969 Model Year 2002 2001 2001 Make Model (Transmission Type) Engine Size LINCOLN GMC JEEP LS K1500 YUKON 4WD CHEROKEE 2WD 003.0 005.3 004.0 29951 30004 30627 34889 30149 29990 21.972 22.011 22.014 22.044 22.070 22.095 2001 2001 2002 2001 2001 2001 DODGE FORD CHRYSLER FORD JEEP FORD RAM 1500 4WD F150 REG CAB LONG VOYAGER F250 SUPER DUTY GRAND CHEROKEE 2WD ESCAPE 005.9 005.4 003.3 006.8 004.0 003.0 1892 22.100 2001 34865 34760 30471 29573 31051 34894 33604 30869 22.143 22.164 22.195 22.207 22.246 22.261 22.310 22.314 2001 2002 2002 2000 2002 2001 2006 2002 1984 22.323 2002 30290 51859 30430 31837 35087 33444 30390 22.329 22.336 22.337 22.350 22.366 22.373 22.393 2001 2002 2001 2003 2004 2006 2001 NISSAN CHEVROLET VOLKSWAGEN MITSUBISHI FORD CHRYSLER TOYOTA QUEST SILVERADO 2500 DIESEL PASSAT LANCER E350 PACIFICA 2WD HIGHLANDER 2WD 003.0 005.2 003.0 0025 002.2 006.8 002.7 004.0 6 Cylinder Gasoline 003.3 006.6 001.8 002.0 005.4 003.5 002.4 31337 22.431 2003 CHEVROLET IMPALA 003.8 29546 29634 30254 30304 31106 22.522 22.522 22.534 22.538 22.546 2000 2000 2001 2001 2002 MAZDA PONTIAC MERCURY PLYMOUTH SUZUKI PROTEGE MONTANA FWD SABLE NEON AERIO 001.8 003.4 003.0 002.0 002.0 31397 22.586 2003 CHRYSLER TOWN & COUNTRY 003.8 29810 31219 29932 22.697 22.706 22.725 2001 2003 2001 BMW AUDI DODGE 330CI A4 QUATTRO DAKOTA 2WD 003.0 003.0 004.7 Multiple Compact Passenger Car Makes and Models BMW DODGE AUDI MERCURY PONTIAC FORD HYUNDAI JEEP 330CI CONVERTIBLE RAM VAN 1500 2WD A4 QUATTRO COUGAR GRAND AM EXCURSION TIBURON GRAND CHEROKEE 2WD Multiple Mini-Van Makes and Models 4 Cylinder Gasoline 131 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29657 29966 34888 30039 29952 30180 30259 29931 52255 29299 30894 30429 29235 30542 29940 32258 32010 31430 30250 29545 30422 30151 Odds Ratio 22.737 22.814 22.817 22.825 22.825 22.873 22.881 22.976 22.982 23.003 23.042 23.077 23.135 23.184 23.203 23.213 23.252 23.263 23.321 23.336 23.358 23.385 Model Year 2000 2001 2001 2001 2001 2001 2001 2001 2003 2000 2002 2001 2000 2002 2001 2004 2003 2003 2001 2000 2001 2001 1894 23.387 2001 31352 23.395 2003 2073 23.453 2004 29854 32041 30729 30181 30832 33015 32818 29189 29992 30655 29857 35080 23.507 23.529 23.550 23.557 23.579 23.581 23.585 23.607 23.646 23.647 23.651 23.659 2001 2003 2002 2001 2002 2005 2005 2000 2001 2002 2001 2004 Make Model (Transmission Type) Engine Size SATURN DODGE FORD FORD DODGE LINCOLN MITSUBISHI DODGE VOLKSWAGEN DODGE LAND ROVER VOLKSWAGEN CHEVROLET CADILLAC DODGE FORD TOYOTA DODGE MERCURY MAZDA VOLKSWAGEN JEEP SC STRATUS F250 SUPER DUTY RANGER REG CAB SHORT RAM 2500 (CA Emission) LS ECLIPSE DAKOTA 2WD JETTA TDI STRATUS RANGE ROVER NEW BEETLE (Manual) VENTURE FWD SEVILLE DURANGO 4WD F150 2WD TUNDRA 2WD NEON COUGAR PROTEGE JETTA GRAND CHEROKEE 4WD 001.9 003.0 005.4 002.5 005.9 003.9 002.4 003.9 001.9 002.5 004.6 002.0 003.4 004.6 005.9 004.2 003.4 002.0 002.5 001.6 002.0 004.0 Multiple Mid-Size Passenger Car Makes and Models CHEVROLET MONTE CARLO General Motors Light Trucks CHEVROLET VOLVO FORD LINCOLN HYUNDAI KIA CHRYSLER CHEVROLET FORD DODGE CHEVROLET FORD SILVERADO 1500 2WD S40 FOCUS WAGON NAVIGATOR SONATA RIO SEBRING CAMARO ESCORT 4DR NEON C1500 TAHOE 2WD F350 SUPER DUTY 6 Cylinder Gasoline 003.8 003.5 005.3 001.9 002.0 005.4 002.7 001.6 002.7 003.8 002.0 002.0 005.3 005.4 132 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 31600 29450 30721 31053 31937 Odds Ratio 23.664 23.805 23.805 23.819 23.832 Model Year 2003 2000 2002 2002 2003 Make Model (Transmission Type) Engine Size GMC HYUNDAI FORD PONTIAC SATURN 004.3 002.5 004.2 003.1 003.0 30715 23.846 2002 FORD 30431 30298 30063 29843 29607 29839 30712 29165 29658 29250 23.851 23.877 23.934 23.940 23.965 23.980 24.061 24.086 24.094 24.108 2001 2001 2001 2001 2000 2001 2002 2000 2000 2000 VOLKSWAGEN OLDSMOBILE GMC CHEVROLET NISSAN CADILLAC FORD CADILLAC SATURN DAEWOO SONOMA 2WD SONATA F150 SUPER CAB (Auto) GRAND PRIX VUE FWD F150 REG CAB SHORT (Auto) PASSAT ALERO C1500 YUKON 2WD ASTRO 2WD CARGO SENTRA DEVILLE F150 REG CAB (Manual) CATERA SL LEGANZA 30428 24.157 2001 VOLKSWAGEN NEW BEETLE (Auto) 002.0 30642 29935 29842 30049 31428 33202 32340 30158 30889 29834 30709 29503 34761 32222 30119 30281 30762 30671 29859 30893 24.197 24.221 24.228 24.229 24.264 24.359 24.430 24.475 24.507 24.508 24.542 24.549 24.642 24.654 24.710 24.718 24.783 24.801 24.842 24.895 2002 2001 2001 2001 2003 2005 2004 2001 2002 2001 2002 2000 2002 2004 2001 2001 2002 2002 2001 2002 DODGE DODGE CADILLAC FORD DODGE SUZUKI HYUNDAI KIA KIA BUICK FORD LAND ROVER DODGE DODGE INFINITI NISSAN FORD DODGE CHEVROLET LAND ROVER DAKOTA 2WD DAKOTA 4WD SEVILLE TAURUS INTREPID FORENZA TIBURON RIO SPORTAGE 4WD CENTURY F150 REG CAB LONG RANGE ROVER RAM VAN 2500 2WD NEON I30 FRONTIER 2WD (Manual) WINDSTAR 4DR STRATUS 4-DR CAMARO FREELANDER 003.9 004.7 004.6 003.0 002.7 002.0 002.7 001.5 002.0 003.1 004.2 004.6 005.2 002.0 003.0 002.4 003.8 002.7 005.7 002.5 004.2 002.8 003.4 005.3 004.3 001.8 004.6 004.2 003.0 001.9 002.2 133 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30379 30264 30472 30148 30561 32232 31501 30653 30271 Odds Ratio 24.913 24.917 24.941 24.948 25.047 25.056 25.060 25.126 25.137 Model Year 2001 2001 2002 2001 2002 2004 2003 2002 2001 Make Model (Transmission Type) Engine Size TOYOTA MITSUBISHI AUDI JEEP CHEVROLET DODGE FORD DODGE MITSUBISHI CAMRY GALANT A6 CHEROKEE 4WD CAMARO STRATUS F150 REG CAB LONG INTREPID MONTERO SPORT 2WD 002.2 003.0 003.0 004.0 003.8 002.4 004.2 002.7 003.0 30287 31416 32194 29882 31411 30024 30412 29946 25.188 25.194 25.198 25.228 25.233 25.292 25.332 25.396 2001 2003 2004 2001 2003 2001 2001 2001 NISSAN DODGE CHRYSLER CHEVROLET DODGE FORD VOLKSWAGEN DODGE PATHFINDER 2WD DAKOTA 2WD SEBRING K1500 TAHOE 4WD CARAVAN 2WD MUSTANG CONVERTIBLE CABRIO NEON 003.5 003.9 002.4 005.3 003.3 003.8 002.0 002.0 30266 25.400 2001 MITSUBISHI MIRAGE 001.8 29499 30610 31049 32923 30345 30626 25.423 25.455 25.616 25.664 25.668 25.798 2000 2002 2002 2005 2001 2002 LAND ROVER CHEVROLET PONTIAC GMC SATURN CHRYSLER DISCOVERY SER II VENTURE FWD FIREBIRD CANYON 2WD SC VOYAGER 004.0 003.4 003.8 003.5 001.9 002.4 29878 25.824 2001 CHEVROLET SILVERADO 1500 4WD 005.3 30113 31336 29243 30092 31690 30059 29939 29252 25.866 25.898 25.908 25.958 25.972 25.999 26.003 26.040 2001 2003 2000 2001 2003 2001 2001 2000 HYUNDAI CHEVROLET CHRYSLER GMC JEEP GMC DODGE DAEWOO SONATA IMPALA SEBRING SONOMA 2WD GRAND CHEROKEE 4WD SIERRA 1500 2WD DURANGO 4WD NUBIRA 2038 26.056 2003 30159 31401 30446 26.072 26.095 26.140 2001 2003 2001 002.4 003.4 002.5 004.3 004.0 004.8 004.7 002.0 6 Cylinder Gasoline 001.8 003.3 001.9 Multiple Mini-Van Makes and Models KIA CHRYSLER VOLVO SEPHIA/SPECTRA VOYAGER/TOWN&CTRY V40 134 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 29965 29948 30080 51985 29993 29904 Odds Ratio 26.270 26.302 26.312 26.361 26.396 26.427 Model Year 2001 2001 2001 2002 2001 2001 Make Model (Transmission Type) Engine Size DODGE DODGE GMC GMC FORD CHRYSLER 002.4 005.2 005.3 006.6 004.6 003.5 30722 26.433 2002 FORD 30060 29837 29848 31493 30763 30421 26.528 26.565 26.653 26.658 26.688 26.735 2001 2001 2001 2003 2002 2001 GMC BUICK CHEVROLET FORD FORD VOLKSWAGEN STRATUS RAM 1500 2WD SIERRA 1500 4WD SIERRA 2500 DIESEL EXPEDITION 300M F150 SUPER CAB SHORT (Manual) SIERRA 1500 2WD REGAL BLAZER 4WD F150 2WD WINDSTAR CARGO VAN JETTA 1910 26.758 2001 30213 31713 30635 30104 30668 30408 32568 34748 30346 26.952 26.991 27.003 27.017 27.018 27.021 27.025 27.063 27.096 2001 2003 2002 2001 2002 2001 2004 2001 2001 MAZDA LAND ROVER DODGE HONDA DODGE TOYOTA SUZUKI AUDI SATURN PROTEGE/PROTEGE MPS FREELANDER CARAVAN 2WD PASSPORT 2WD STRATUS TUNDRA 2WD FORENZA A4 SL 001.6 002.5 002.4 003.2 002.4 003.4 002.0 001.8 001.9 29890 27.114 2001 CHEVROLET S10 PICKUP 2WD 004.3 32817 31702 29991 30440 29870 27.128 27.183 27.208 27.243 27.283 2005 2003 2001 2001 2001 CHRYSLER KIA FORD VOLVO CHEVROLET SEBRING RIO ESCORT S40 IMPALA 002.4 001.6 002.0 001.9 003.4 30280 27.330 2001 NISSAN FRONTIER 2WD (Auto) 002.4 29888 30887 30272 29968 29912 31452 27.451 27.495 27.592 27.608 27.612 27.612 2001 2002 2001 2001 2001 2003 CHEVROLET KIA MITSUBISHI DODGE CHRYSLER DODGE PRIZM SPECTRA MONTERO SPORT 2WD STRATUS 4-DR SEBRING COUPE STRATUS 4-DR 001.8 001.8 003.5 002.7 003.0 002.4 Multiple Light Truck Makes and Models 004.2 005.3 003.8 004.3 004.2 003.8 001.8 4 Cylinder Gasoline 135 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID Odds Ratio Model Year 1939 27.616 2001 30636 29853 31000 34931 31711 31890 30263 29812 30052 30051 27.652 27.755 27.762 27.836 27.850 27.879 27.883 28.060 28.109 28.132 2002 2001 2002 2002 2003 2003 2001 2001 2001 2001 DODGE CHEVROLET MITSUBISHI CHRYSLER LAND ROVER PONTIAC MITSUBISHI BMW FORD FORD CARAVAN 2WD SILVERADO 1500 2WD LANCER TOWN & COUNTRY 2WD DISCOVERY GRAND PRIX GALANT 330I WINDSTAR 4DR WAGON WINDSTAR 3DR WAGON 8 Cylinder Gasoline 003.3 004.8 002.0 003.3 004.6 003.1 002.4 003.0 003.8 003.8 34878 28.216 2001 CHRYSLER TOWN & COUNTRY 2WD 003.8 30764 30695 30127 32855 30939 29884 31511 29632 30028 29937 30260 29926 31031 29847 30022 28.220 28.274 28.310 28.506 28.554 28.670 28.674 28.711 28.735 28.804 28.832 28.847 28.909 28.950 29.059 2002 2002 2001 2005 2002 2001 2003 2000 2001 2001 2001 2001 2002 2001 2001 FORD FORD ISUZU DODGE MAZDA CHEVROLET FORD PONTIAC FORD DODGE MITSUBISHI DODGE OLDSMOBILE CHEVROLET FORD WINDSTAR WAGON ESCORT RODEO 2WD STRATUS MPV MALIBU F150 SUPER CAB SHORT GRAND PRIX MUSTANG COUPE (Auto) DURANGO 2WD ECLIPSE CARAVAN 2WD ALERO BLAZER 2WD MUSTANG 003.8 002.0 003.2 002.4 003.0 003.1 004.2 003.1 003.8 004.7 003.0 002.4 002.2 004.3 003.8 29915 29.074 2001 CHRYSLER SEBRING CONVERTIBLE 002.7 30274 30053 30161 31688 30382 30892 30178 29949 31706 29.233 29.428 29.431 29.554 29.624 29.696 29.716 29.790 29.811 2001 2001 2001 2003 2001 2002 2001 2001 2003 MITSUBISHI FORD KIA JEEP TOYOTA LAND ROVER LINCOLN DODGE KIA MONTERO SPORT 4WD WINDSTAR VAN SPORTAGE 4WD GRAND CHEROKEE 2WD CAMRY SOLARA DISCOVERY SER II LS RAM 1500 2WD SPECTRA 003.5 003.8 002.0 004.0 002.2 004.0 003.0 005.9 001.8 Make Model (Transmission Type) Multiple Full-Size Van Makes and Models Engine Size 136 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30888 34749 29928 29947 30291 30265 29543 29871 31468 31552 30317 29249 29883 30038 Odds Ratio 29.893 29.908 29.962 30.007 30.031 30.152 30.218 30.480 30.840 30.893 31.841 32.081 32.346 32.567 Model Year 2002 2001 2001 2001 2001 2001 2000 2001 2003 2003 2001 2000 2001 2001 1930 32.584 2001 30401 30114 29852 30284 32.743 32.784 32.831 32.870 2001 2001 2001 2001 TOYOTA HYUNDAI CHEVROLET NISSAN 29891 32.894 2001 29886 32571 29776 30311 30184 33.123 33.582 33.595 33.600 33.648 30268 30012 29887 31762 30160 30005 30884 30684 30156 29927 29903 30258 Make Model (Transmission Type) Engine Size KIA CHEVROLET DODGE DODGE NISSAN MITSUBISHI MAZDA CHEVROLET FORD FORD PONTIAC DAEWOO CHEVROLET FORD SPORTAGE 2WD S10 PICKUP 2WD CARAVAN 2WD RAM 1500 2WD SENTRA MIRAGE MPV IMPALA E150 ECONOLINE WINDSTAR WAGON GRAND AM LANOS LUMINA RANGER REG CAB SHORT SIENNA SONATA SILVERADO 1500 2WD MAXIMA 002.0 002.2 003.8 003.9 001.8 001.5 002.5 003.8 004.2 003.8 003.4 001.6 003.1 002.3 6 Cylinder Gasoline 003.0 002.5 004.3 003.0 CHEVROLET S10 PICKUP 2WD FFV 002.2 2001 2004 2001 2001 2001 CHEVROLET SUZUKI AUDI PONTIAC MAZDA MONTE CARLO VERONA A6 AZTEK FWD 626 003.4 002.5 002.8 003.4 002.0 33.916 2001 MITSUBISHI MONTERO 003.5 33.972 34.010 34.117 34.122 34.368 34.464 34.520 34.606 34.666 35.058 35.183 2001 2001 2003 2001 2001 2002 2002 2001 2001 2001 2001 FORD CHEVROLET MAZDA KIA FORD KIA FORD KIA DODGE CHEVROLET MITSUBISHI F150 SUPER CAB (Auto) MONTE CARLO MPV SPORTAGE 2WD F150 REG CAB SHORT OPTIMA E150 ECONOLINE OPTIMA CARAVAN 2WD VENTURE FWD DIAMANTE SEDAN 004.2 003.8 003.0 002.0 004.2 002.7 004.2 002.4 003.3 003.4 003.5 Multiple Mini-Van Makes and Models 137 Appendix C: VLT ID Numbers Most Likely to Fail the OBD II Functional Test VLT ID 30302 29922 30111 31633 52135 34877 30002 29858 29914 Odds Ratio 35.378 35.610 35.804 35.863 36.005 36.346 36.417 36.480 36.501 Model Year 2001 2001 2001 2003 2003 2001 2001 2001 2001 Make Model (Transmission Type) Engine Size OLDSMOBILE DAEWOO HYUNDAI HYUNDAI FORD CHRYSLER FORD CHEVROLET CHRYSLER 003.5 002.2 002.4 002.7 006.0 003.3 004.2 003.8 002.7 30013 36.526 2001 FORD 30938 30029 29918 29941 30319 30072 29921 30318 30163 30312 29979 30299 30320 30211 36.608 36.613 36.666 37.184 37.807 37.998 38.451 38.909 39.043 39.576 40.481 40.813 41.487 43.542 2002 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 MAZDA FORD CHRYSLER DODGE PONTIAC GMC DAEWOO PONTIAC LAND ROVER PONTIAC FORD OLDSMOBILE PONTIAC MAZDA INTRIGUE LEGANZA SANTA FE TIBURON EXCURSION DIESEL TOWN & COUNTRY 2WD F150 REG CAB LONG CAMARO SEBRING SEDAN F150 SUPER CAB SHORT (Manual) MILLENIA MUSTANG COUPE (Manual) VOYAGER INTREPID GRAND PRIX JIMMY 2WD LANOS GRAND PRIX DISCOVERY SER II BONNEVILLE E150 ECONOLINE AURORA MONTANA FWD MPV 004.2 002.5 003.8 003.3 002.7 003.8 004.3 001.6 003.1 004.0 003.8 004.2 003.5 003.4 002.5 138 Appendix D: VLT ID Numbers Demonstrating Zero Failures of the Visual Inspection VLT ID 2073 2236 2245 Model-Year 2004 2007 2007 Make Model General Motors Light Trucks Multiple Mid-Size Light Truck Makes and Models Multiple Mid-Size SUV Makes and Models Engine Size 003.5 6 Cylinder Gasoline 6 Cylinder Gasoline 2264 2007 Multiple Full-Size Van Makes and Models 8 Cylinder Gasoline 2269 2008 SMART 2274 2281 2318 2325 2335 2354 2372 2398 2426 2433 2450 2487 2489 2513 2516 2524 2008 2008 2008 2009 2009 2009 2009 2010 2010 2011 2011 2012 2012 2012 2012 2012 Multiple Luxury Car Makes and Models Multiple Mid-Size Passenger Car Makes and Models Multiple Full-Size Van Makes and Models Multiple Mid-Size Passenger Car Makes and Models Multiple Mid-Size SUV Makes and Models Multiple Full-Size SUV Makes and Models Multiple Light Truck Makes and Models Multiple Compact Light Truck Makes and Models Multiple Full-Size Van Makes and Models Multiple Mid-Size Passenger Car Makes and Models Multiple Compact Light Truck Makes and Models Multiple Mid-Size Passenger Car Makes and Models Multiple Full-Size Passenger Car Makes and Models Multiple Mid-Size SUV Makes and Models Multiple Full-Size SUV Makes and Models Multiple Mini-Van Makes and Models 10 Cylinder Gas 6 Cylinder Gasoline 8 Cylinder Gasoline 5 Cylinder Gasoline 6 Cylinder Gasoline 8 Cylinder Gasoline 8 Cylinder Gasoline 6 Cylinder Gasoline 8 Cylinder Gasoline 5 Cylinder Gasoline 4 Cylinder Gasoline 5 Cylinder Gasoline 8 Cylinder Gasoline 4 Cylinder Gasoline 8 Cylinder Gasoline 6 Cylinder Gasoline 2534 2012 Multiple Full-Size Van Makes and Models 8 Cylinder Gasoline 29568 29575 29639 29641 29776 29817 30283 30452 30537 30794 30889 30903 30974 31048 31067 31485 2000 2000 2000 2000 2001 2001 2001 2002 2002 2002 2002 2002 2002 2002 2002 2003 MERCEDES MERCURY PORSCHE PORSCHE AUDI BMW NISSAN ACURA CADILLAC GMC KIA LEXUS MERCEDES PONTIAC PORSCHE FORD SMART CAR SL500 MOUNTAINEER BOXSTER BOXSTER S A6 540I FRONTIER 4WD 3.2CL ELDORADO SIERRA 1500 DENALI AWD SPORTAGE 4WD LX 470 SL500 BONNEVILLE BOXSTER EXPLORER 2WD 001.0 005.0 004.0 002.7 003.2 002.8 004.4 003.3 003.2 004.6 006.0 002.0 004.7 005.0 003.8 002.7 004.6 139 Appendix D: VLT ID Numbers Demonstrating Zero Failures of the Visual Inspection VLT ID 31565 31593 Model-Year 2003 2003 Make GMC GMC Model ENVOY XL 2WD K1500 YUKON DENALI XL Engine Size 004.2 006.0 31596 2003 GMC SAFARI 2WD PASSENGER 004.3 31640 31722 31798 31800 32066 32073 32094 32109 32244 32350 32366 32403 32406 32413 32456 32458 2003 2003 2003 2003 2004 2004 2004 2004 2004 2004 2004 2004 2004 2004 2004 2004 INFINITI LEXUS MERCEDES MERCEDES AUDI AUDI BMW BUICK FORD INFINITI JAGUAR LEXUS LEXUS LINCOLN MERCEDES MERCEDES I35 GX 470 ML320 ML500 A4 A6 QUATTRO 5-SERIES PARK AVENUE E250 QX56 2WD XJ8 GS 430 LX 470 LS ML500 S500 003.5 004.7 003.2 005.0 003.0 002.7 003.0 003.8 004.6 005.6 004.2 004.3 004.7 003.9 005.0 005.0 32486 2004 MITSUBISHI MONTERO SPORT 003.5 32499 32532 32542 32575 32582 32624 32640 32705 32707 32717 32739 32740 32752 32812 32835 32836 32884 32895 2004 2004 2004 2004 2004 2004 2004 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 NISSAN PORSCHE SATURN TOYOTA TOYOTA VOLKSWAGEN VOLVO BMW BMW BUICK CADILLAC CADILLAC CHEVROLET CHRYSLER DODGE DODGE FORD FORD PATHFINDER 4WD BOXSTER L300 4RUNNER 2WD CAMRY SOLARA PASSAT S40 M3 X3 LACROSSE STS STS C1500 SUBURBAN CROSSFIRE DURANGO 2WD DURANGO 2WD EXPLORER SPORT TRAC FIVE HUNDRED 003.5 003.2 002.2 004.7 002.4 002.8 002.5 003.2 003.0 003.8 003.6 004.6 005.3 003.2 004.7 005.7 004.0 003.0 140 Appendix D: VLT ID Numbers Demonstrating Zero Failures of the Visual Inspection VLT ID 32976 33003 33012 33026 33039 33067 33079 33080 33084 33089 33095 33133 33158 33166 33170 33188 33209 33224 33232 33233 33266 33321 33324 33339 33344 33360 33367 33369 Model-Year 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2006 2006 2006 2006 2006 2006 2006 Make INFINITI JEEP KIA LAND ROVER LEXUS MAZDA MERCEDES MERCEDES MERCEDES MERCEDES MERCEDES NISSAN PONTIAC PORSCHE SAAB SUBARU TOYOTA TOYOTA TOYOTA TOYOTA VOLVO BMW BMW BMW BUICK CADILLAC CADILLAC CADILLAC Model FX35 RWD GRAND CHEROKEE 4WD AMANTI LR3 SC 430 TRIBUTE CLK320 CLK500 E500 ML350 SL500 MURANO AWD VIBE CAYENNE 9-3 FORESTER AWD 4RUNNER 2WD HIGHLANDER 2WD RAV4 2WD RAV4 4WD S40 3-SERIES 5-SERIES Z4 COUPE LUCERNE DTS STS STS Engine Size 003.5 003.7 003.5 004.4 004.3 003.0 003.2 005.0 005.0 003.7 005.0 003.5 001.8 004.5 002.0 002.5 004.0 002.4 002.4 002.4 002.4 002.5 003.0 003.0 004.6 004.6 003.6 004.6 33375 2006 CHEVROLET AVALANCHE 1500 4WD 005.3 33455 33469 33494 33508 2006 2006 2006 2006 CHRYSLER DODGE FORD FORD TOWN & COUNTRY 2WD DURANGO 2WD E150 EXPLORER 4WD 003.8 005.7 005.4 004.0 33529 2006 FORD FREESTAR WAGON FWD 004.2 33541 33601 33605 33608 2006 2006 2006 2006 FORD HYUNDAI HYUNDAI HYUNDAI RANGER 4WD SONATA TUCSON TUCSON 4WD 004.0 002.4 002.0 002.7 141 Appendix D: VLT ID Numbers Demonstrating Zero Failures of the Visual Inspection VLT ID 33609 33610 33624 33657 33665 33670 33680 33691 33692 33696 33723 33724 33737 33743 33749 33752 33763 33768 33772 33777 33788 33796 33837 33839 33840 33873 33876 Model-Year 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 Make INFINITI INFINITI JAGUAR KIA KIA LAND ROVER LAND ROVER LEXUS LEXUS LEXUS MAZDA MAZDA MERCEDES MERCEDES MERCEDES MERCEDES MERCEDES MERCURY MERCURY MERCURY MITSUBISHI MITSUBISHI PORSCHE PORSCHE PORSCHE SUBARU SUZUKI Model FX35 FX35 AWD S-TYPE OPTIMA SPORTAGE LR3 SPORT LS 430 LX 470 SC 430 TRIBUTE TRIBUTE CLK500 E500 ML500 S350 SLK350 MARINER MILAN MOUNTAINEER 2WD ENDEAVOR OUTLANDER BOXSTER CAYENNE CAYENNE OUTBACK AERIO Engine Size 003.5 003.5 003.0 002.4 002.7 004.0 004.2 004.3 004.7 004.3 002.3 003.0 005.0 005.0 005.0 003.7 003.5 003.0 003.0 004.0 003.8 002.4 002.7 003.2 004.5 003.0 002.3 33878 2006 SUZUKI GRAND VITARA 002.7 33904 33905 33955 33956 33965 34005 34025 34027 34044 34047 2006 2006 2007 2007 2007 2007 2007 2007 2007 2007 TOYOTA TOYOTA ACURA ACURA AUDI BMW CADILLAC CADILLAC CHEVROLET CHEVROLET RAV4 2WD RAV4 4WD MDX RDX A4 X3 DTS ESCALADE AWD SILVERADO 1500 2WD C1500 SUBURBAN 003.5 002.4 003.7 002.3 002.0 003.0 004.6 006.2 004.3 005.3 142 Appendix D: VLT ID Numbers Demonstrating Zero Failures of the Visual Inspection VLT ID 34053 34070 Model-Year 2007 2007 Make CHEVROLET CHEVROLET Model COLORADO 2WD IMPALA Engine Size 003.7 003.5 34081 2007 CHEVROLET K1500 SUBURBAN 005.3 34099 34132 34154 34168 34173 34174 34183 34188 34199 34211 34264 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 CHEVROLET DODGE DODGE FORD FORD FORD FORD FORD FORD FORD HONDA UPLANDER CARAVAN 2WD NITRO EDGE EXPEDITION 2WD EXPLORER 2WD F150 2WD FIVE HUNDRED FUSION TAURUS ELEMENT 003.9 003.3 003.7 003.5 005.4 004.0 004.2 003.0 002.3 003.0 002.4 34351 2007 LAND ROVER RANGE ROVER 004.4 34359 34374 34375 34399 34403 34413 34414 34420 34452 34456 34457 34461 34462 34467 34469 34470 34509 34531 34547 34549 34560 34571 34573 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 LEXUS LINCOLN LINCOLN MERCEDES MERCEDES MERCEDES MERCEDES MERCEDES MITSUBISHI NISSAN NISSAN NISSAN NISSAN NISSAN NISSAN NISSAN SATURN SUBARU TOYOTA TOYOTA TOYOTA TOYOTA TOYOTA GS 350 NAVIGATOR TOWN CAR C230 CLK350 GL450 ML350 S550 OUTLANDER ALTIMA ARMADA MAXIMA MURANO TITAN 2WD VERSA XTERRA 2WD AURA OUTBACK 4RUNNER 2WD 4RUNNER 4WD FJ CRUISER 4WD RAV4 2WD RAV4 4WD 003.5 005.4 004.6 002.5 003.5 004.6 003.5 005.5 003.0 003.5 005.6 003.5 003.5 005.6 001.8 004.0 003.5 002.5 004.0 004.0 004.0 003.5 003.5 143 Appendix D: VLT ID Numbers Demonstrating Zero Failures of the Visual Inspection VLT ID 34574 34592 34817 34820 34825 35068 35168 35181 35229 35243 35257 35276 35277 50039 50041 50042 50045 52961 53016 53038 53069 Model-Year 2007 2007 2006 2007 2007 2004 2005 2006 2007 2007 2007 2007 2007 2009 2010 2010 2012 2008 2009 2010 2005 Make Model TOYOTA SEQUOIA 2WD VOLKSWAGEN PASSAT CHEVROLET EXPRESS 2500 BMW 525I CHEVROLET EXPRESS 2500 BMW 325I SPORT WAGON MINI MINI COOPER BMW 330CI CONVERTIBLE BMW 328I COUPE CHEVROLET SILVERADO 2500 2WD FORD E350 MINI MINI COOPER MINI MINI COOPER S Multiple Mid-Size Passenger Car Makes and Models Multiple Mid-Size Passenger Car Makes and Models Multiple Light Truck Makes and Models Multiple Mid-Size Passenger Car Makes and Models DODGE SPRINTER 2500 DIESEL DODGE RAM 2500 DIESEL FORD F350 DIESEL FORD F350 SRW SUPER DUTY Engine Size 004.7 002.0 006.0 002.5 004.8 002.5 001.6 003.0 003.0 006.0 005.4 001.6 001.6 Diesel Engine Diesel Engine Diesel Engine Diesel Engine 003.0 006.7 006.4 006.0 144 Appendix E: VLT ID Numbers Most Likely to Fail the Visual Inspection VLT ID Odds Ratio Model Year 50022 1.605 2003 31853 52293 29289 29349 29729 29102 29217 1.627 1.647 1.647 1.652 1.659 1.660 1.708 2003 2004 2000 2000 2000 2000 2000 NISSAN DODGE DODGE FORD VOLKSWAGEN ACURA CHEVROLET 350Z RAM 2500 DIESEL NEON FOCUS ZX3 3DR NEW BEETLE (Auto) INTEGRA MALIBU Engine Size 6 and 8 Cylinder Diesel 003.5 005.9 002.0 002.0 002.0 001.8 003.1 29346 1.722 2000 FORD FOCUS 4-DR SEDAN 002.0 29631 1.728 2000 PONTIAC GRAND AM 003.4 30642 1.742 2002 DODGE DAKOTA 2WD 003.9 34749 30109 30587 33281 29580 31095 29731 30617 30730 1.742 1.754 1.764 1.771 1.771 1.772 1.774 1.776 1.776 2001 2001 2002 2005 2000 2002 2000 2002 2002 CHEVROLET HYUNDAI CHEVROLET VOLVO MERCURY SUBARU VOLKSWAGEN CHRYSLER FORD S10 PICKUP 2WD ACCENT MONTE CARLO XC90 SABLE IMPREZA AWD PASSAT PT CRUISER (Auto) FOCUS ZX3 002.2 001.6 003.4 002.5 003.0 002.0 001.8 002.4 002.0 29730 1.784 2000 VOLKSWAGEN NEW BEETLE (Manual) 002.0 29277 31428 32505 30654 30107 1.789 1.790 1.794 1.802 1.813 2000 2003 2004 2002 2001 DODGE DODGE NISSAN DODGE HONDA DAKOTA 4WD INTREPID SENTRA INTREPID S2000 004.7 002.7 002.5 003.5 002.0 29200 1.815 2000 CHEVROLET IMPALA 003.4 29764 30106 52265 30422 1.815 1.821 1.821 1.825 2001 2001 2004 2001 ACURA HONDA CHEVROLET VOLKSWAGEN INTEGRA PRELUDE SILVERADO 2500 DIESEL JETTA 001.8 002.2 006.6 002.0 30735 1.826 2002 FORD MUSTANG CONVERTIBLE (Auto) 004.6 29990 52341 29299 1.827 1.849 1.854 2001 2004 2000 FORD FORD DODGE ESCAPE EXCURSION DIESEL STRATUS 003.0 006.0 002.5 Make Model (Transmission Type) Multiple Domestic Light Truck Makes and Models 145 Appendix E: VLT ID Numbers Most Likely to Fail the Visual Inspection VLT ID 30421 32521 Odds Ratio 1.855 1.857 Model Year 2001 2004 Make Model (Transmission Type) VOLKSWAGEN PONTIAC JETTA GRAND AM Engine Size 001.8 003.4 32228 1.858 2004 DODGE RAM 1500 4WD 004.7 29724 52171 31336 29322 31522 52493 29323 1.865 1.868 1.868 1.878 1.879 1.880 1.887 2000 2003 2003 2000 2003 2005 2000 VOLKSWAGEN FORD CHEVROLET FORD FORD DODGE FORD JETTA F250 DIESEL IMPALA ESCORT MUSTANG RAM 2500 DIESEL ESCORT 4DR 001.8 006.0 003.4 002.0 004.6 005.9 002.0 31173 1.891 2002 VOLKSWAGEN NEW BEETLE (Manual) 002.0 29728 33453 29727 30469 30268 29629 29493 1.896 1.896 1.899 1.900 1.919 1.925 1.932 2000 2006 2000 2002 2001 2000 2000 VOLKSWAGEN CHRYSLER VOLKSWAGEN AUDI MITSUBISHI PONTIAC JEEP NEW BEETLE SRT-8 NEW BEETLE A4 QUATTRO (Auto) MONTERO FIREBIRD/TRANS AM WRANGLER 4WD 001.8 006.1 001.8 001.8 003.5 005.7 002.5 29613 1.954 2000 OLDSMOBILE ALERO 003.4 31434 30217 30019 34766 30994 31889 30418 29632 29584 52197 29915 29736 29223 31205 30250 29922 29870 29269 1.958 1.959 1.969 1.973 1.977 1.988 1.996 1.997 2.001 2.005 2.010 2.014 2.021 2.022 2.027 2.029 2.035 2.035 2003 2001 2001 2002 2002 2003 2001 2000 2000 2003 2001 2000 2000 2003 2001 2001 2001 2000 DODGE MAZDA FORD VOLKSWAGEN MITSUBISHI PONTIAC VOLKSWAGEN PONTIAC MITSUBISHI FORD CHRYSLER VOLKSWAGEN CHEVROLET ACURA MERCURY DAEWOO CHEVROLET DODGE RAM 1500 2WD TRIBUTE FOCUS WAGON JETTA ECLIPSE GRAND AM GOLF (Manual) GRAND PRIX ECLIPSE F350 DIESEL SEBRING CONVERTIBLE PASSAT WAGON S10 PICKUP 2WD FFV RSX COUGAR LEGANZA IMPALA CARAVAN 2WD 003.7 003.0 002.0 001.8 003.0 003.4 002.0 003.1 003.0 006.0 002.7 001.8 002.2 002.0 002.5 002.2 003.4 003.3 146 Appendix E: VLT ID Numbers Most Likely to Fail the Visual Inspection VLT ID 30428 Odds Ratio 2.036 Model Year 2001 Make Model (Transmission Type) VOLKSWAGEN NEW BEETLE (Auto) Engine Size 002.0 29909 2.038 2001 CHRYSLER PT CRUISER 002.4 33812 30753 29249 30618 29886 34839 32103 29357 29442 29719 30650 2.044 2.049 2.055 2.060 2.064 2.066 2.066 2.068 2.079 2.080 2.084 2006 2002 2000 2002 2001 2000 2004 2000 2000 2000 2002 NISSAN FORD DAEWOO CHRYSLER CHEVROLET CHRYSLER BMW FORD HONDA VOLKSWAGEN DODGE SENTRA RANGER SUPER CAB 2DR SH LANOS PT CRUISER (Manual) MONTE CARLO VOYAGER X5 MUSTANG COUPE (Auto) PRELUDE GOLF DURANGO 2WD 002.5 003.0 001.6 002.4 003.4 003.3 004.4 004.6 002.2 002.0 005.9 51985 2.085 2002 GMC SIERRA 2500 DIESEL 006.6 30659 29245 30417 29117 29624 29294 34739 30814 30825 33462 31610 32246 30030 32559 33866 29379 30260 29444 33249 29118 30148 52059 29725 2.105 2.124 2.126 2.132 2.137 2.141 2.162 2.177 2.186 2.211 2.236 2.262 2.263 2.263 2.267 2.284 2.284 2.289 2.309 2.313 2.335 2.341 2.353 2002 2000 2001 2000 2000 2000 2000 2002 2002 2006 2003 2004 2001 2004 2006 2000 2001 2000 2005 2000 2001 2003 2000 DODGE CHRYSLER VOLKSWAGEN AUDI PLYMOUTH DODGE CHEVROLET HONDA HYUNDAI DODGE HONDA FORD FORD SUBARU SUBARU FORD MITSUBISHI HYUNDAI VOLKSWAGEN AUDI JEEP CHEVROLET VOLKSWAGEN RAM 1500 4WD TOWN & COUNTRY 2WD GOLF (Auto) A4 QUATTRO VOYAGER 2WD RAM 1500 4WD S10 PICKUP 2WD CIVIC ACCENT CHARGER CIVIC ESCAPE MUSTANG COUPE IMPREZA WRX IMPREZA TAURUS WAGON ECLIPSE ACCENT (Auto) JETTA A4 QUATTRO CHEROKEE 4WD SILVERADO 2500 DIESEL JETTA 004.7 003.3 002.0 001.8 003.3 005.9 002.2 002.0 001.5 006.1 002.0 003.0 004.6 002.0 002.5 003.0 003.0 001.5 001.8 001.8 004.0 006.6 002.0 147 Appendix E: VLT ID Numbers Most Likely to Fail the Visual Inspection VLT ID 30653 52063 29591 29966 32021 30731 29910 34764 51859 29219 30419 30668 32436 29921 31027 30729 29586 30108 31695 29445 30739 29593 52087 29358 31452 Odds Ratio 2.362 2.367 2.370 2.385 2.406 2.433 2.459 2.466 2.475 2.489 2.494 2.494 2.500 2.515 2.535 2.545 2.566 2.584 2.590 2.639 2.662 2.697 2.700 2.707 2.712 Model Year 2002 2003 2000 2001 2003 2002 2001 2002 2002 2000 2001 2002 2004 2001 2002 2002 2000 2001 2003 2000 2002 2000 2003 2000 2003 Make Model (Transmission Type) DODGE CHEVROLET MITSUBISHI DODGE VOLKSWAGEN FORD CHRYSLER FORD CHEVROLET CHEVROLET VOLKSWAGEN DODGE MAZDA DAEWOO NISSAN FORD MITSUBISHI HYUNDAI JEEP HYUNDAI FORD MITSUBISHI DODGE FORD DODGE 30736 2.714 2002 FORD 29286 33794 29965 33193 52209 29941 29968 31943 31528 29714 29298 2.734 2.746 2.768 2.783 2.831 2.867 2.908 2.943 2.950 2.958 3.044 2000 2006 2001 2005 2003 2001 2001 2003 2003 2000 2000 DODGE MITSUBISHI DODGE SUBARU GMC DODGE DODGE SUBARU FORD VOLKSWAGEN DODGE INTREPID SILVERADO 3500 DIESEL MONTERO SPORT 2WD STRATUS JETTA FOCUS ZX3 PT CRUISER (Manual) RANGER SUPER CAB 4DR SILVERADO 2500 DIESEL METRO GTI STRATUS TRIBUTE LANOS SENTRA (Manual) FOCUS WAGON GALANT ACCENT LIBERTY 4WD ACCENT (Manual) MUSTANG COUPE MONTERO SPORT 4WD RAM 2500 DIESEL MUSTANG COUPE (Manual) STRATUS 4-DR MUSTANG CONVERTIBLE (Manual) INTREPID LANCER EVOLUTION STRATUS IMPREZA STI SIERRA 2500 DIESEL INTREPID STRATUS 4-DR IMPREZA AWD MUSTANG COUPE CABRIO STRATUS Engine Size 002.7 006.6 003.0 003.0 001.8 002.0 002.4 003.0 006.6 001.3 001.8 002.4 003.0 001.6 002.5 002.0 003.0 001.5 003.7 001.5 004.6 003.0 005.9 004.7 002.4 004.6 002.7 002.0 002.4 002.5 006.6 002.7 002.7 002.0 004.6 002.0 002.4 148 Appendix E: VLT ID Numbers Most Likely to Fail the Visual Inspection VLT ID 29885 31163 32614 52095 29239 32019 30412 31158 Odds Ratio 3.082 3.100 3.116 3.175 3.394 3.439 3.706 4.331 Model Year 2001 2002 2004 2003 2000 2003 2001 2002 Make Model (Transmission Type) CHEVROLET VOLKSWAGEN VOLKSWAGEN DODGE CHRYSLER VOLKSWAGEN VOLKSWAGEN VOLKSWAGEN METRO GTI GTI RAM 3500 DIESEL CONCORDE GTI CABRIO CABRIO Engine Size 001.3 001.8 001.8 005.9 002.7 001.8 002.0 002.0 149 REFERENCES Acock, A. C. (2010). A gentle introduction to Stata. College Station, TX: Stata Press Air Resources Board. (2009). On-board diagnostic II (OBD II) systems – fact sheet / FAQs. California Environmental Protection Agency, Air Resources Board. Retrieved from http://www.arb.ca.gov/msprog/obdprog/obdfaq.htm Austin, T. C., McClement, D., & Roeschen, J. D. (2009). Evaluation of the California smog check program using random roadside data. Sacramento, CA: Sierra Research, Inc. Barone, S. (2006). Statistics-driven development of OBD systems: an overview. Quality and Reliability Engineering International 22(2006), 615-628. Bin, O. (2003). A logit analysis of vehicle emissions using inspection and maintenance testing data. Transportation Research Part D: Transport and Environment, 8(3), 215-227. Retrieved from: http://www.sciencedirect.com/science/article/pii/S136192090300004X Bureau of Automotive Repair. (2009). Smog check inspection procedures manual. Retrieved from http://www.bar.ca.gov/80_BARResources/05_Legislative/Regulatory Actions/Smog%20Check%20Manual%20w-diesel%208-09%20V3.pdf Bureau of Automotive Repair. (2010a). Directed vehicles. Retrieved from http://www.autorepair.ca.gov/80_BARResources/02_SmogCheck/TestOnly_Directed_ Vehicles.html Bureau of Automotive Repair. (2012a). 2012 Smog check performance report. Retrieved from http://www.autorepair.ca.gov/80_BARResources/ftp/pdfdocs/2012%20Smog%20Chec k%20Performance%20Report.pdf Bureau of Automotive Repair. (2012b). Executive summary report. Smog check statewide. Year: 2011. Retrieved from http://www.bar.ca.gov/80_BARResources/02_ SmogCheck/Engineering/ExecutiveSummaryReports/ArchivesContents.asp Bureau of Automotive Repair. (2012c). STAR program. Retrieved from http://www.bar.ca.gov /80_BARResources/03_Standards&Training/Star/StarProgram.html Bureau of Automotive Repair. (2012d). New equipment and inspection procedures workshops presentation. http://www.smogcheck.ca.gov/80_BARResources/05_ Legislative/RegulatoryActions/NewEquipandInspProceduresWorkshops.html Bureau of Automotive Repair. (2013). Executive summary report. Smog check statewide. Year: 2012. Retrieved from http://www.bar.ca.gov/80_BARResources/02_ SmogCheck/Engineering/ExecutiveSummaryReports/ArchivesContents.asp 150 REFERENCES California Inspection and Maintenance Review Committee. (2000). Smog check II evaluation. Retrieved from http://www.feat.biochem.du.edu/assets/reports/CA%20IMRC%20 Report/IMRC20000619Part1.PDF California State Assembly. (1994). Bill analysis for Assembly Bill 2018: Proposed Conference Report No. 1. Retrieved from: http://www.leginfo.ca.gov/pub/93-94/bill/asm/ab_20012050/ab_2018_cfa_940316_174230_asm_floor Cascio, F., Console, L., Guagliumi, M., Osella, M., Panati, A., Sottano, S. & Dupre, D. T. (1999). Generating on-board diagnostics of dynamic automotive systems based on qualitative models. AI Communications 12 (1999) 33-43 Choo, S., Shafizadeh, K., & Niemeier, D. (2007). The development of a prescreening model to identify failed and gross polluting vehicles. Transportation Research Part D: Transport and Environment. 12(3), 208-218. Retrieved from: http://www.sciencedirect.com/science/article/pii/S1361920907000132 Eisinger, D. S. (2010). Smog check: science, federalism, and the politics of clean air. New York, NY: Taylor and Francis Group Environmental Protection Agency. (2013). Status of state and local OBD I/M programs. United States Environmental Protection Agency. Retrieved from http://www.epa.gov/obd/status.htm Lyons, A., & McCarthy, M. (2009) Transitioning away from smog check tailpipe emission testing in California for OBD II equipped vehicles. Sacramento, CA: California Air Resources Board, Mobile Source Control Division Moghadam, A. K. & Livernois, J. (2010). The abatement cost function for a representative vehicle inspection and maintenance program. Transportation Research Part D: Transport and Environment. 15(5), 285-297. Retrieved from: http://www.sciencedirect.com/science/article/pii/S1361920910000180 Singer, B. C. & Wenzel, T. P. (2003) Estimated emission reductions from California's enhanced smog check program. Environmental Science & Technology. 37(11), 2588-2595 Sosnowski, D. & Gardetto, E. (2001). Performing onboard diagnostic system checks as part of a vehicle inspection and maintenance program. United States Environmental Protection Agency, Office of Transportation and Air Quality, Transportation and Regional Programs Division. Publication number EPA420-R-01-015, June 2001 Singleton, JR., R. A. & Straits, B. C. (2010). Approaches to social research. New York, NY: Oxford University Press Studenmund, A. H. (2011). Using econometrics: A practical guide. Boston, MA: AddisonWestley 151 REFERENCES Supnithadnaporn, A., Noonan, D. S., Samoylov, A., & Rodgers, M. O. (2011). Estimated validity and reliability of on-board diagnostics for older vehicles: comparison with remote sensing observations. Journal of the Air & Waste Management Association, 61(10), 996-1004. United States Environmental Protection Agency. (1992). 1992 Enhanced I/M support document. Retrieved from: http://www.epa.gov/otaq/epg/regs.htm Washburn, S., Seet, J., & Mannering, F. (2001). Statistical modeling of vehicle emissions from inspection/maintenance testing data: an exploratory analysis. Transportation Research Part D: Transport and Environment. 6(1), 21 – 36. Retrieved from: http://www.sciencedirect.com/science/article/pii/S1361920900000110 Wenzel, T. P. & Ross, M. (2003) Comment on “Statistical modeling of vehicle emissions from inspection/maintenance testing data: an explanatory analysis”. Transportation Research Part D: Transport and Environment. 8(1), 69-71. Retrieved from: http://www.sciencedirect.com/science/article/pii/S1361920902000093