CHAPTER 5 RESEARCH DESIGN LEARNING OBJECTIVES To understand what research design is, why it is significant, and the types To learn how exploratory research design may be used and the methods to conduct exploratory research To know the fundamental questions addressed by descriptive research and the two major types of descriptive research To know the different types and uses of panels in marketing research To explain what is meant by causal research, experiments and experimental design To know the different types of test marketing and how to select test market cities CHAPTER OUTLINE PROPER RESEARCH DESIGN IS ESSENTIAL TO VALID RESEARCH RECOMMENDATIONS The Significance of Research Design THREE TYPES OF RESEARCH DESIGNS Research Design: A Caution Exploratory Research Uses of Exploratory Research Gain Background Information Define Terms Clarify Problems and Hypotheses Establish Research Priorities Methods of Conducting Exploratory Research Secondary Data Analysis Experience Surveys Case Analysis Focus Groups Projective Techniques Descriptive Research Classification of Descriptive Research Studies Causal Research EXPERIMENTS 60 Chapter 5: Research Design Experimental Design After-Only Design One-Group, Before-After Design Before-After with Control Group How Valid Are Experiments? Types of Experiments TEST MARKETING Types of Test Markets Consumer Versus Industrial Test Markets "Lead Country" Test Markets Selecting Test Market Cities Pros and Cons of Test Marketing KEY TERMS Research design Exploratory research Secondary data analysis Experience surveys Case analysis Focus groups Projective techniques Descriptive research Cross-sectional studies Sample surveys Longitudinal studies Panels Continuous panels Discontinuous panels Omnibus panels Brand-switching studies Market tracking studies Causality Experiment Independent variables Dependent variables Extraneous variables Experimental design Pretest Posttest "True" experimental design After-only design Quasi-experimental designs One-group, before-after design Control group 61 Chapter 5: Research Design Experimental group Before-after with control group Internal Validity External validity Laboratory experiments Field experiments Test marketing Standard test market Controlled test markets Electronic test markets Simulated test markets Lead country test market Representativeness Degree of isolation Ability to control distribution and promotion ACTIVE LEARNING EXERCISES We just used the UWF/Listener Group® panel as an example of a discontinuous panel. Go to the website at www.uwf.edu/panel. This exercise requires students to visit an online panel web site and learn about how it operates. The question answers are: Read what the web site has to say about what a panel is… Panels Panels are simply groups of people who have agreed to answer survey questions periodically. Many of the top research firms, such as ACNielsen, use panels to collect research information. The UWF/Listener Group Panel This is a panel of male and female heads of households in the Pensacola MSA (Escambia and Santa Rosa counties). The panel is constructed in such a way as to be representative of all heads of households in the two county area. There are approximately 800 persons in the panel. …how often panel members may be contacted… How often are panel members surveyed? Panel members are contacted no more than once a month. …and how the panel protects the privacy of its panel members… 62 Chapter 5: Research Design How private is the information gathered by the panel? Although panel managers must have the panel member's telephone number (or email address) in order to contact the panel member, no other information is known about the identity of the panel member except for the panel member's first name. The first name is gathered simply to allow the survey to be directed to the correct member of the household. Any information reported by the panel member is never associated with the member's telephone number. Rather, data will be reported in the aggregate and/or reported by broad demographic categories such as by income group, zip code, or county of residence. … take a look at the kinds of information the panel has collected by looking at some of the survey results… Recent survey results are posted for inspection. For the UWF/Listener Group panel, how representative are the results and how can the panel members determine if the results are representative? The UWF/Listener Group Panel and Representativeness Panel is still a sample, a subset of a larger population, to which one wishes to make inferences. If data derived from the sample are to be inferred to the population, the question of representativeness arises. Is the UWF/Listener Group Panel representative? Heads of Households with Telephones First, the Panel was constructed so as to recruit heads of households having a telephone. Heads of household are often used in survey research because household heads are considered to be knowledgeable of a wide variety of decisions, have a broad base of experience, and so on, which makes them better qualified to answer a broad variety of survey questions. This is particularly important for panels whose members may be asked a variety of questions from one panel administration to another. It is important to note, therefore, that the UWF/Listener Group Panel does not represent the general population, but rather heads of households in the Pensacola having a telephone. (Approximately 98 percent of the households in the MSA have a telephone). How is representativeness achieved? The sample plan , not sample size, allows one to achieve sample representativeness. Essentially, sample plans describe how elements, or sample units, are drawn from the population. For example, we could “draw” elements by standing on a street corner and asking passersby if they would participate in the panel. This is a type of sample plan (often referred to as the “man on the street” interview). However, this type of plan, part of a number of plans known as “non-probability” sample plans, does NOT guarantee representativeness. Rather, “probability” sample plans guarantee representativeness. A probability sampling plan is one in which every member of the population (defined as heads of households with telephones) has a chance of 63 Chapter 5: Research Design being drawn into the sample. The UWF/Listener Group Panel was recruited using a probability sample plan known in the marketing research industry as a “RDD Sample.” In this sample plan, computer programs generate lists of randomly selected telephone numbers. The numbers are then screened to eliminate not working (or unassigned) blocks of numbers. By randomly generating numbers, we are assured that everyone in the defined population has a chance of being included in the sample. The RDD sample for this project was supplied by Survey Sampling, Inc.http://www.surveysampling.com/ssi_home.html, a major sampling firm in the U.S. So, in summary, the UWF/Listener Group Panel is a representative sample of heads of households with telephones in the Pensacola MSA. (There is more on this topic on the website.) TEACHING SUGGESTIONS 1. (The following will require giving an assignment to students and having them bring to class their findings.) Although students have not been fully exposed to exploratory research methods, they can perform three of them to some extent. Divide the class up into three groups and assign each group one of the following types of exploratory research: (1) secondary data analysis, (2) experience surveys, and (3) case analysis. Here are some possible “assignment” situations. It is necessary to keep the situations local. The university is experiencing a decline (increase) in enrollment and wonders why. (Chose “decline” or “increase” to fit the situation of your institution. Alternatively, consider using a particular major or a college as it suits your situation and students’ abilities.) The university bookstore is considering going to an “all rent” policy for textbooks. (Some universities have this policy in place. Note: the experience survey will pertain to students’ and instructors’ experiences with buying and selling back textbooks.) The director of your “student life” office is concerned about the high use and possible abuse of credit cards by college students. Under consideration is a selfhelp program for students to learn about credit card abuse and money management. Have the groups present their findings and use class discussion to generate hypotheses. 2. Focus groups are covered in detail in chapter 8, and projective techniques are described in greater detail in that chapter as well. Resist the temptation to tell all about these methods to students at this time unless you have something else to present for your qualitative research techniques coverage later in the course. 64 Chapter 5: Research Design 3. Illustrate the result of a cross-sectional descriptive study by presenting findings from such a study to the class. If you have used team projects in the past, you have many from which to pick. Alternatively, you may use a consulting report, or even the descriptive findings from an academic study may suffice. Use the tables and figures to illustrate the descriptive nature of the findings. For instance, show how the demographic profile describes the sample. If there is product use information, use it to show how it describes the types of product usage. 4. Some students may not readily understand the value of a longitudinal panel. An analogy is that a cross-sectional survey is like a snapshot; whereas a longitudinal panel is like a video. 5. A tangential discussion can be generated by asking students the implications of attrition with an omnibus panel versus a longitudinal panel. In the former case, the lost panel member can be replaced with someone whose profile is identical or very similar. That is, if a female elderly sole survivor living on the West Coast drops out, you would replace her with an elderly sole survivor who lives on the West Coast. However, with a drop out in a longitudinal panel, the string of observations over time is broken, so attrition is a much more serious problem. 6. You can have fun with causality because humans have the inborn tendency to make causal attributions without using good experimental design. Here are some examples. It has been documented that human births increase as stork populations grow, so storks bring babies. (Actually, rain causes crops to grow, so agrarian people expand their families, while water affords more protection for stork nests so more storks are sighted.) Lizards sometimes jump out of fires, so medieval people believed that lizards were created by fire. (Actually, lizards jump out of logs when lit on fire, because they hide in logs and the fire forces them to leave their hiding places.) The sun revolves around the earth because one can see it rising and setting. It was believed for centuries that the sun revolved around the earth because of the fallacious assumption that the earth stood still. (In truth, the sun’s gravity causes the earth to revolve around it while the earth spins on its axis.) 7. The after-only experimental design can be set up as a convenient straw man to demonstrate the effects of not controlling other independent and extraneous variables. Ask students to identify the independent variable, dependent variable, and causal statement in one or more of the following examples. A dry cleaning company institutes a “frequent cleaner” plan where $1 is taken off the price of a dry cleaning order for every 10 times a customer uses the dry cleaning company. Sales at the end of three months are $10,000. 65 Chapter 5: Research Design A movie theater adds more parking spaces in its lot, and 3,000 tickets are sold in a month. Goodyear adds a line of ultra-high performance radials and sells 200,000 of them in six months. As an after-only design, the causal statement is that “X caused O.” That is, the frequent cleaner plan caused $10,000 in sales; the parking spaces caused 3,000 tickets to be sold, and that addition of the radials caused Goodyear to sell 200,000 of them. However, students will generate a great number of other possible reasons why the sales were as described. Other independent variables may be at work, such as an especially popular movie being shown during the month. Extraneous variables are unaccounted for as well as competitors’ actions or changes in customers. 8. The descriptions of internal and external validity in the chapter do not utilize terms used in more comprehensive treatments. For internal validity, the terms are history, maturation, instrumentation, pretest effect, equivalence, and mortality, while for external validity, they are sample representativeness, artificiality, and generalizability. Some instructors may want to introduce these terms to students when reviewing the two types of validity. 9. The two experimental designs without control groups can be used to illustrate the role of a control group. The after-only and one-group, before-after designs suffer from internal validity problems, namely extraneous effects and changes in subjects in both cases, and measure error and subjects guessing for the before-after design. The way to overcome these problems is to let them happen to both the experimental and the control group. That is, you must have a control group to “capture” the internal validity problem, and then subtract it out. The description of the before-after with control group and the causal statement formula where E = (O2 - O1) - (O4 - O3) shows how the control groups O3 and O4 do the subtracting out. 10. There are experimental designs that are more complex and which are not covered in the chapter. Some instructors may want to expand students’ knowledge by bringing these into the classroom. 11. Students can relate to a taste test as an example of a laboratory experiment, and they can be questioned on why, for example, when Diet Pepsi does a blind taste test against Diet Coke, findings show a slight preference for Diet Pepsi. However, Diet Coke sales are more than those for Diet Pepsi. The artificiality of the blind taste test can be exploited. There are extraneous and independent variables affecting brand loyalty, plus who does a blind taste test of two colas to decide which one to buy? 12. Consider doing an abbreviated taste test in class. You can select some students to be the administrators, and two or three to be the taste testers. One approach is to tell the administrators to read up on taste tests and to design and implement one themselves. 66 Chapter 5: Research Design Another is to bring the materials to class and select the administrators and tasters from class, but you will need to orient the administrators on how to conduct the test. Here are the instructions for three colas with six subjects. It is best to use Dixie cupsized cups as subjects must take several drinks. It is fun to use national brands like Coke or Pepsi along with private brands like Sam’s Choice or Dr. Thunder. Instructions for Conducting an In-Class Taste Test 1. Pour the colas into identical large containers suitable for storage and decanting. Assign the three colas letters (A, B, C). 2. Select subjects at random. 3. Assign the specific order of the taste tests to each subject: ABC, ACB, BCA, BAC, CAB, CBA 4. Subjects drink bottled water. 5. Pour the first letter cola for each subject; each tastes the cola and rates it. 6. Subjects drink bottled water. 7. Pour the second letter cola for each subject; each tastes the cola and rates it. 8. Subjects drink bottled water. 9. Pour the third letter cola for each subject; each tastes and rates the cola. 10. Subjects identify 1st choice with respect to best taste rating. For class enjoyment: Let the subjects guess what cola brands they have tasted, then divulge the brands to them and the class. 13. There are four types of test markets described: standard, controlled, electronic, and simulated. Ask students who uses each type and why. One factor is the distribution system of the company wishing a test market; another is the type of product being tested; a third factor is the type of information desired. Standard test markets fit companies with fully developed distribution systems that are marketing convenience goods such as grocery products, personal care items, or over-the-counter drugs. This is a “do-it-yourself” approach with the information need being primarily to track the test product’s sales. Controlled test markets fit companies without fully developed distribution systems. The test market company establishes the retail structure during the test. Convenience goods marketers are a good fit. There is more competitive sales 67 Chapter 5: Research Design information available because all brands are tracked, so the information needs are more specific to what brand is gaining or losing during the test. Electronic test markets are limited to those companies testing products that are sold in the stores where the electronic tracking system is in place (grocery and drug stores primarily). Because electronic test market customers are in a database, the information would profile the buyers versus nonbuyers and provide market segmentation implications. The simulated test market is suited to a new product where the innovation is tried and the repurchase rate estimated by mathematical model. The distribution system may be nonexistent, or the cost of other marketing variables needed in the test (e.g., training salespeople to demonstrate it, or a large-scale educational advertising campaign) may be prohibitive. There is an implicit evaluative process and adoption decision on the part of the consumer, and its outcome is the primary information need. 14. Principles of marketing textbooks sometimes list 10 to 20+ U.S. test market cities. One exercise is to take these listings over the past 10 years and see which cities have been added or deleted from the lists. If one of these is known to students due to its proximity to the university, students may be able to comment on why it was dropped or added. 15. The representativeness issue is paramount in selecting a test market city. Although the cities cited in some textbooks (Tulsa, Oklahoma; Charleston, West Virginia; Lexington-Fayette, Kentucky; Wichita, Kansas; Rockford, Illinois; Omaha, Nebraska; and Midland, Texas) are well known for being similar, in many respects, to the “average U.S. city,” they do not fit the target market definitions of all marketers. Ask students to indicate products, services, or brands whose target markets are not “average Americans” and who would be ill-advised to use these cities in their market tests. Here are some examples of product markets that are not comprised of average Americans: In-line skates Fly fishing equipment Jazz music CDs White water rafting adventures Your students can probably generate a much better list of nonqualifiers. ANSWERS TO END-OF-CHAPTER QUESTIONS 1. How would you match research designs with various research objectives? Review question. To answer this question, students can use Table 5.1. 68 Chapter 5: Research Design This is done in Table 5.1, repeated below. Research Objective To gain background information, to define terms, to clarify problems and hypotheses, to establish research priorities Appropriate Design Exploratory To describe and measure marketing phenomena at a point in time Descriptive To determine causality, to make “if-then” statements Causal 2. Give some examples illustrating the uses of exploratory research. Review question. Students must demonstrate an understanding of exploratory research. Exploratory research is unstructured, informal research that is undertaken to gain background information about the general nature of the research problem. It is usually conducted when the researcher does not know much about the problem and needs additional information or desires new or more recent information. The purposes of exploratory research are to (1) gain background information, (2) define terms, (3) clarify problems and hypotheses, and (4) establish research priorities. Any number of examples exists. The types described in the chapter are: (1) secondary data analysis, (2) experience surveys, (3) case analysis, (4) focus groups, and (5) projective techniques. For class discussion, ask students what types of exploratory research they would undertake if they were managers of a food service center on campus, and the school newspaper just published a blistering editorial on the poor quality of food and service in their unit. Some exploratory research methods include: Research method Example Secondary data analysis stories.” Reading about college food service “success Experience surveys Discussing the matter with student leaders. Asking the newspaper editor for more information. Asking cashiers what comments students have made to them. 69 Chapter 5: Research Design Case analysis in the food Observing what students do and do not eat service center. Focus group Inviting students to participate in a open meeting on how the food service can do better. Projective techniques Giving students a sheet of paper and asking them to complete the statement, “This food service center would be a better place if...” 3. What type of research design answers the questions of who, what, where, when, and how? Review question. To answer this question, students will need to realize that these questions are basic elements of descriptive research This is part of the definition of descriptive research. Who are our customers? What brands do they buy and in what quantities? Where do they buy? When do they buy? How do they buy? 4. What are the differences between longitudinal studies and cross-sectional studies? Review question. Students must distinguish between these two types of studies. By definition, cross-sectional studies measure a population at only one point in time, while longitudinal studies repeatedly measure the same population over a period of time. One might make the analogy that a cross-sectional study is a snapshot and a longitudinal study is a video. 5. In what situation would a continuous panel be more suitable than a discontinuous panel? Under what situation would a discontinuous panel be more suitable than a continuous panel? Review question. This question tests students’ comprehension of these two panel types. Continuous panels ask panel members the same questions on each panel measurement. Discontinuous panels vary questions from one panel measurement to the next. Unlike the continuous panel, discontinuous panels may be used for a variety of purposes because the information collected by an discontinuous panel varies from one panel measurement to the next. Continuous panels are noted in the chapter as 70 Chapter 5: Research Design suited to brand switching and market tracking studies, while discontinuous panels are of the “snapshot” variety. That is, comparisons are made between groups (say MasterCard owners versus Discover Card owners) existing within the entire discontinuous panel. 6. Explain why studies of “if-then” variety are considered to be causal studies. Review question. Students must relate the concept of manipulating an independent variable. Causality may be thought of as understanding a phenomenon in terms of conditional statements of the form “If X, then Y.” These “if-then” statements, then, become ways of manipulating independent variables of interest. 7. What is the objective of good experimental design? Explain why certain designs are called “quasi- experimental” design. Review question. This question tests students’ understanding of experimental design. An experimental design is a procedure for devising an experimental setting such that a change in a dependent variable may be attributed solely to the change in an independent variable. Experimental designs are procedures that allow experimenters to control for the effects on a dependent variable by an extraneous variable. In this way, the experimenter is assured that any change in the dependent variable was due only to the change in the independent variable. A “true” experimental design is one that truly isolates the effects of the independent variable on the dependent variable while controlling for effects of any extraneous variables. Designs that do not properly control for the effects of extraneous variables on the dependent variable are known as quasi-experimental designs. In the quasiexperimental design diagrams there is no measure of E, the “experimental effect,” on the dependent variable due solely to the independent variable. This is true in all quasi-experimental designs. 8. Explain the two types of validity in experimentation and also explain why different types of experiments are better suited for addressing one type of validity versus another. Review question. Students are required to discuss internal and external validity. An experiment is valid if (a) the observed change in the dependent variable is, in fact, due to the independent variable, and (b) if the results of the experiment apply to the “real world,” outside the experimental setting. Two forms of validity are used to assess the validity of an experiment: internal and external. Internal validity is concerned with the extent to which the change in the dependent variable was actually due to the independent variable. External validity refers to the extent that the 71 Chapter 5: Research Design relationship observed between the independent and dependent variable during the experiment is generalizable to the “real world.” Also, refer to Table 5.5. We distinguished between laboratory and field experiments. Laboratory experiments are those in which the independent variable is manipulated and measures of the dependent variable are taken in a contrived, artificial setting for the purpose of controlling the many possible extraneous variables that may affect the dependent variable. So lab experiments are better suited for addressing internal validity. Field experiments are those in which the independent variables are manipulated and the measurements of the dependent variable are made on test units in their natural setting. Field experiments are better suited for addressing external validity. 9. Distinguish among the various types of test marketing. Review question. In order to answer this question, students must review the various types of test markets described in the chapter. Test marketing is the phrase commonly used to indicate an experiment, study, or test that is conducted in a field setting. The chapter classifies test markets into four types: standard, controlled, electronic, and simulated. The standard test market is one in which the firm tests the product and/or marketing mix variables through the company’s normal distribution channels. Controlled test markets are conducted by outside research firms who guarantee distribution of the product through pre-specified types and numbers of distributors. Electronic test markets are those in which a panel of consumers has agreed to carry an identification card that they present when buying goods and services. Simulated test markets (STMs) are those in which some limited amount of data on consumer response to a new product are fed into a model containing certain assumptions regarding planned marketing programs, which generates likely product sales volume. There are also consumer versus industrial test markets. In consumer tests, multiple versions of a more-or-less finished product are tested by consumers. In industrial test markets, the key technology is presented to selected industrial users who offer feedback on desired features and product performance levels. 10. Think of a past job that you have held. List three areas in which you, or some other person in the organization, could have benefited by having information generated by research. What would be the most appropriate research design for each of the three areas of research you have listed? Application question. This exercise asks students to relate how information could help an organization with which they are familiar. The purpose of this question is to have students relate research design to something with which they are familiar. Students’ work experiences will vary greatly and may prove interesting as a basis for class discussion. 72 Chapter 5: Research Design 11. Design an experiment. Select an independent variable and a dependent variable. What are some possible extraneous variables which may cause problems? Explain how you would control for the effects these variables may have on your dependent variable. Is your experiment a valid experiment? Application question. Students must apply experimental design concepts. The purpose is to have students apply an experimental design and thus pick out independent, dependent, and extraneous variables as well as assess the internal and external validity of the experiment. Class discussion can be generated by having a student “present” his or her design to the class. 12. The Maximum Company has invented an extra-strength instant coffee brand to be called “Max-Caff,” and positioned to be stronger tasting than any competing brands. Design a taste-test experiment that compares Max-Caff to the two leading instant coffee brands to determine which brand consumers consider to taste the strongest. Identify and diagram your experiment. Indicate how the experiment is to be conducted, and assess the internal and external validity of your experiment. Application question. A taste test is an experiment, so students must demonstrate comprehension of experimental design concepts. There is no need to perform a pretest as there is no baseline to taste tests. Max-Caff is to be compared to the leading brands, A and B (Maxwell House and Folger’s, for instance). With a taste test, the design can be within subjects or across subjects. For the within design, the same subject tastes all three coffees and rates the taste of each. Typically, a neutral stimulus such as a soda cracker or a drink of water is administered between the randomly ordered stimuli. The panel design would look like the following: X1 O1 N X2 O2 N X3 O3 Where the Xis are the randomly ordered stimuli, the Ois are the measurements, and the Ns are the neutralizers. With an across subjects design, three groups are used, each with its own stimulus. Subjects are assigned randomly to each group, and the after-only design diagram looks like this: X1 X2 X3 O1 O2 O3 No control group is used as it does not make sense to have subjects rate the strength of a nonstimulus. 73 Chapter 5: Research Design Internal validity is addressed in the following ways. Extraneous factors – eliminated if the panel design is random and if the three groups take place in the same circumstances and time frame Changes in subjects – obviated by the neutral stimulus in the panel and the short time frame in the three-group design Measure error – use the same measure Subjects guessing – could happen but will be random and cancel out Equivalent groups – use random selection or matching criteria Drop-out – should be monitored External validity is an issue as the taste test is conducted in an artificial setting (perhaps at a shopping mall or some other facility), and the subjects are told that they are to rate the taste which means that they will be especially vigilant to the taste. Also, the long-term taste aspects of the coffees are not tested, so if Max-Caff “stays with” the taster, this factor is not captured by the design. Of course, the sample should be representative of coffee drinkers. 13. Coca-Cola markets PowerAde as a sports drink that competes with Gatorade. Competition for sports drinks is fierce where they are sold in the coolers of convenience stores. Coca-Cola is thinking about using a special holder that fits in a standard cooler but moves PowerAde to eye level and makes it more conspicuous than Gatorade. Design an experiment that determines whether or not the special holder increases the sales of PowerAde in convenience stores. Identify and diagram your experiment. Indicate how the experiment is to be conducted, and assess the internal and external validity of your experiment Application question. This question describes an experimental design that students must recognize and analyze. There is one treatment (the cooler holder for PowerAde) and one control situation (the current stocking of PowerAde without the cooler holder). A before-after with control group or an after-only with control group design would be satisfactory. The after-only with control group design looks like the following: X1 O1 O2 The groups are randomly assigned convenience stores such as Circle Ks or 711s, and the Os are store sales, and the effect of the PowerAde cooler holder is 74 Chapter 5: Research Design O1:O2, and if the cooler holder average PowerAde sales are greater than the control group average PowerAde sales, the experiment has shown that the cooler holder has caused the greater sales Internal validity is addressed in the following ways: Extraneous factors – eliminated groups take place in the same circumstances and time frame Changes in subjects – if groups are matched or random, changes will be the same for all Measure error – use the same measure – sales of PowerAde Subjects guessing – no pretest, so no problem Equivalent groups – use random selection or matching criteria Drop-out – probably no problem, but should be monitored External validity is taken care of if the groups are representative of all convenience stores. The test is realistic and generalizable. 14. SplitScreen is a marketing research company that tests television advertisements. SplitScreen has an agreement with a cable television company in a medium-sized city in Iowa. The cable company can send up to four different television ads simultaneously to different households. SplitScreen also has agreements with the three largest grocery store chains, which will provide scanner data to SplitScreen. About 25 percent of the residents have SplitScreen scan cards that are scanned when items are bought at the grocery store and which allows SplitScreen to identify who bought what grocery products. For allowing SplitScreen access to their television hook-ups and their grocery purchases information, residents receive bonus points that can be used to buy products in a special points catalog. Identify and diagram the true experimental designs possible using the SplitScreen system. Assess the internal and external validity of SplitScreen’s system. Application question. This question introduces students to scanner data experimental design considerations and possibilities. Virtually any true experimental design is possible with the Split Screen system, but the most powerful ones would be before-after with control group or after-only design. The system allows for up to four groups – either four experimental groups or three experimental groups and one control group. By combining or not using groups, fewer groups can be used for a less complex design. 75 Chapter 5: Research Design All internal validity problems are under control assuming that groups are equivalent, drop-outs are replaced, commercials are seen at the same time, stores remain constant, etc. The external validity may concern some students who will point out the artificiality of the scanner cards, bonus points, and the use of a medium-sized city in Iowa. If Split Screen recruits households and does not use them right away and thus lets them settle back into their normal shopping patterns after what might be a period of unusual shopping because they feel “watched,” the artificiality will be minimized. The medium-sized city in Iowa is consistent with good market test selection criteria. CASE SOLUTIONS Case 5.1 Quality Research Associates Case Objective This case presents students with six different situations where they must determine the appropriate research design. Answers to Case Questions 1. What research design do you think Sam Fulkerson should select for each of his clients? 2. For each research design you specify in question 1, describe the reason(s) you selected it. The answers are provided in the table below. Client Hamptons Bank Research Design Descriptive research Wesleyan College Exploratory research M&M Mars candy bars M&M Mars cereal Exploratory research McBride’s Markets Causal research Phillips, Yarbrough and Rogers Descriptive research Causal research 76 Reason(s) Need to measure the bank’s image with a cross-sectional survey Early stage in this decision; need to clarify opinions (probably with focus groups) Will require an extensive literature search with secondary data Need to conduct a taste test type study but out of the laboratory A “what-if” situation. The “if” is variations of ad copy and layout, and the “what” is some measure of attention The “more evidence” required by retail chain buyers would be satisfied with a large survey. A pretest of the two Chapter 5: Research Design promotional campaign themes could be embedded in this survey. CASE 5.2 The Hobbit’s Choice: Research Design Case Objective This case requires students to identify appropriate research designs for the Hobbit’s Choice Restaurant marketing research project. Answers to Case Questions 1. What type of research design did Cory Rogers use to be prepared for the second meeting with his client, Jeff Dean? What are some methods that he likely used? Given that he was armed with financial and performance information about restaurants, Cory most certainly used exploratory research in the form of secondary data. He probably used an Internet search engine or, if available to him, he used a library-type data base system of some sort. The case notes that he has ordered a marketing research report on upscale restaurants, and this is secondary data. 2. Consider the second paragraph above which refers to the critical issues in the case. Would exploratory research be the most appropriate research design to choose, given the research objectives that would be generated to provide the information necessary to resolve these issues? Explain why or why not. The critical issues are: “a. determining demand, b. determining the best choice of restaurant design and operating characteristics, c. determining where to locate the restaurant, and d. determining how to promote The Hobbit’s Choice Restaurant.” These issues are all in the realm of primary demand: he needs specific information about each one. Students might mistakenly recommend focus groups; however, although some useful information would be gained, the generalizability of this information would be very low. 3. If you elected not to use exploratory research design in question 2, which of the two remaining research designs would be more appropriate - descriptive or causal? Select one and provide the rational for you choice. Descriptive research is definitely called for as we need answers to who, what, when, how, and so on. So, this requires a survey. Jeff needs information on how to design the restaurant, where to locate it, how to promote it, and so forth. 77 Chapter 5: Research Design 4. How does selection of the research design you chose in question 3 aid Cory Rogers in planning the research project for The Hobbit’s Choice Restaurant? The choice of a research design such as descriptive research has a number of implicit decisions. For example, the objectives must be set, questionnaire designed, data collection method decided, sample plan identified, sample size determined, and so on. So, when the basic research design is decided, the researcher knows what decisions must be made (and in what order if he/she refers to the 11-step process for marketing research described in this textbook.) 78