Personality Assessment Inventory (PAI) Leslie C. Morey Texas A&M University, U.S.A. The Personality Assessment Inventory (PAI) is a self-report questionnaire designed to provide psychological assessment information pertaining to psychopathology, personality, and psycho-social environment. As an inventory, it has a broad coverage of several important psychological constructs that make it applicable to diverse types of settings. As a result, surveys of practicing clinicians indicate that it has become one of the most popular clinical measures in mental health, forensic/correctional, screening, and training contexts (e.g., Archer, Buffington-Vollum, Stredy, & Handel, 2006). Associated with the PAI are other measures that are related to, but not interchangeable with, the PAI. The Personality Assessment Inventory-Adolescent Version (PAI-A) provides a direct parallel to the PAI for use with individuals aged 12 to 18. The Personality Assessment Screener (PAS) is a very brief (22-item) screening tool that samples broadly from the problem domains assessed by the PAI, designed to provide a sensitive indicator for the likelihood of significant problems requiring additional specific assessment. Although these instruments can serve as important complements to the PAI, they are distinct measures and are not the focus of this discussion. This entry provides a brief overview of the theory and procedures employed in developing the PAI, as well as a discussion of PAI psychometric and validity data for various applications. More detailed coverage can be found in several primary sources listed in the “Further Reading” section found at the end of this entry. Description of the PAI The PAI is a self-report questionnaire that includes 344 questions; it can be administered by paper and pencil or by computer. Typically, completion time for the PAI is between 45 and 60 minutes, although this can vary depending upon the clinical presentation. The respondent provides answers to questions on a 4-point scale that ranges from “totally false, not at all true” to “very true.” The response selected corresponds to an item score (ranging from 0 to 3), and these scores are summed into total scores. These items are arranged into 22 full scales, and 10 of those full scales include subscales. Scale and subscale descriptions, as well as commonly used supplemental indices, are listed in Table 1. The PAI is designed for use with individuals aged 18 and older; as mentioned previously, there is an Adolescent Version that includes items and norms suited for use with those aged 12 to 18. The PAI is a self-report instrument, so it requires that the respondent be capable of understanding written test items and the use of the response scale. Analyses of the PAI items indicate that the statements are worded at a fourth-grade reading level, which is lower than comparable psychopathology inventories (Schinka & Borum, 1993). Furthermore, item differential functioning analyses were conducted to attempt to eliminate any bias related to gender, race, or age. With respect to cross-cultural application, the PAI has been translated into dozens of languages, although these translations vary considerably in the degree to which normative data are available when administered in these languages. Theoretical Basis and Test Development The development of the PAI was based on a construct validation framework, an approach The Encyclopedia of Clinical Psychology, First Edition. Edited by Robin L. Cautin and Scott O. Lilienfeld. © 2015 John Wiley & Sons, Inc. Published 2015 by John Wiley & Sons, Inc. DOI: 10.1002/9781118625392.wbecp284 Table 1 PAI Scales, Subscales, and Supplemental Indices. Scale Content measured Validity scales ICN INF NIM PIM Inconsistency Infrequency Negative Impression Management Positive Impression Management Inconsistent responding to similar items Idiosyncratic responses to items Negative response set due to cognitive distortions and/or feigning of psychopathology Positive response set due to lack of insight and/or intentional dissimulation Clinical scales SOM SOM-C SOM-S Somatic Complaints Conversion Somatization SOM-H ANX ANX-C ANX-A ANX-P ARD ARD-O Health Concerns Anxiety Cognitive Affective Physiological Anxiety-Related Disorders Obsessive-Compulsive ARD-P ARD-T DEP DEP-C DEP-A DEP-P MAN MAN-A MAN-G MAN-I PAR PAR-H PAR-P PAR-R SCZ SCZ-P SXZ-S SCZ-T Phobias Traumatic Stress Depression Cognitive Affective Physiological Mania Activity Level Grandiosity Irritability Paranoia Hypervigilance Persecution Resentment Schizophrenia Psychotic Experiences Social Detachment Thought Disorder BOR BOR-A BOR-I BOR-N BOR-S Borderline Features Affective Instability Identity Problems Negative Relationships Self-Harm Rare sensorimotor symptoms associated with conversion disorders Frequent, common physical symptoms or vague complaints of ill health or fatigue Preoccupation with physical functioning Ruminative worry, impaired concentration and attention Tension, difficulty relaxing, nervousness, and fatigue Physical signs such as sweating, tremors, palpitations Intrusive thoughts, compulsive behaviors, perfectionism, affective constriction Common phobic fears Enduring effects of trauma exposure Worthlessness, hopelessness, difficulty concentrating Feelings of sadness, dysphoria Lowered drive, disruptions in sleep and eating patterns Disorganized over-involvement, accelerated behavior Inflated self-esteem, expansiveness Impatience, low frustration tolerance Tendency to closely monitor environment for threat Belief that others intentionally obstruct respondent Bitterness and cynicism, externalization of blame Unusual perceptions and ideas, magical thinking Social isolation, discomfort, and awkwardness Confusion, concentration difficulties, and disorganization Poor modulation of emotional responses Uncertainty about major life issues, lack of purpose History of intense, ambivalent relationships Impulsivity with disregard for negative consequences PERSONALITY ASSESSMENT INVENTORY (PAI) Table 1 3 (Continued.). ANT ANT-A ANT-E ANT-S Antisocial Features Antisocial Behaviors Egocentricity Stimulus-Seeking ALC DRG Alcohol Problems Drug Problems History of antisocial and illegal behavior Lack of empathy, exploitative approach to relationships Cravings for excitement, low boredom tolerance, recklessness Use of and problems with alcohol Use of and problems with drugs Treatment consideration scales AGG AGG-A AGG-V AGG-P SUI Aggression Aggressive Attitude Verbal Aggression Physical Aggression Suicidal Ideation STR NON Stress Nonsupport RXR Treatment Rejection Hostility, easily aroused anger Assertiveness, readiness to express anger to others Tendency and history for physical aggression Frequency and intensity of thoughts of self-harm or suicide Perception of an uncertain and unstable environment Perception that others are not available or willing to provide support Low motivation for treatment, little readiness to change Interpersonal scales DOM Dominance WRM Warmth Desire and tendency for control in relationships; low scores suggest meekness and submissiveness Interest and comfort with close relationships; low scores suggest hostility, anger, or mistrust Supplemental validity indices MAL RDF NDS DEF CDF ALCEst DRGEst Malingering Index Rogers Discriminant Function Negative Distortion Scale Defensiveness Index Cashel Discriminant Function ALC Estimated Score DRG Estimated Score Negative response set, malingering Malingering Negative response set, malingering Positive other or self- deception Intentional underreporting of specific problems ALC predicted by other elements of the profile DRG predicted by other elements of the profile Supplemental Predictive Indices TPI Treatment Process Index VPI Violence Potential Index SPI Suicide Potential Index Difficult treatment process, high probability of complicated treatment Profile features associated with increased likelihood of violence Profile features associated with increased likelihood of suicide that reflects a combination of a theoretical development and selection of items, as well as empirical assessment of their stability and correlates. Initial work on the PAI began in 1987 by the test author, Leslie C. Morey. As a first step, the theoretical and empirical literature for each of the constructs to be measured was closely examined to serve as a guide to the content of information to be sampled and to the subsequent assessment of content validity. 4 PERSONALITY ASSESSMENT INVENTORY (PAI) The development of the test then went through four iterations in a sequential construct validation strategy, beginning with data provided by clinical experts regarding the properties of proposed items, followed by three waves of data collection on preliminary versions of the test. The selection of items was based upon consideration of numerous item parameters; one unique aspect underlying PAI scale construction was the assumption that no single quantitative item parameter should be used as the sole criterion for item selection. This strategy was adopted because an overreliance on a single parameter typically leads to a scale with one desirable psychometric property and numerous undesirable ones. Both the conceptual nature and empirical adequacy of the items played an important role in their inclusion in the final version of the PAI. Two particular elements of construct validity were emphasized in PAI scale construction—content validity and discriminant validity. Content validity indicates the extent to which the measure in question provides a representative sampling of the relevant content of the construct of interest. In this regard, the construction of the PAI sought to develop scales that provided a balanced sampling of the most important elements of the constructs being measured, with respect to both the breadth (the diversity of elements subsumed within a construct) as well as depth (the full range of severity) of the construct. As one example, the PAI sought to insure breadth of content coverage through the use of subscales representing the major elements of the measured constructs to safeguard against extremely homogeneous item content coverage that would ultimately lower the content validity of the full measure. The second area of emphasis in PAI scale construction involved discriminant validity, or the extent to which a measure is a relatively “pure” indicator of a particular construct that is not contaminated by undue associations with other constructs. Although discriminant validity has been long recognized as an important facet of construct validity, traditionally it has not played a major role in the construction of psychological tests, and it continues to represents one of the largest challenges in the assessment of psychological constructs. There are a variety of threats to validity where discriminability plays a vital role. One such area involves test bias. A test that is intended to measure a psychological construct should not be inadvertently measuring a demographic variable, such as gender, age, or sex. However, such bias is just one form of potential problems with discriminant validity. It is particularly common in the field of clinical assessment to find that a measure that supposedly measures one construct (such as anxiety or schizophrenia) is in fact highly related to many constructs. It is this tendency that makes many instruments quite difficult to interpret; for example, it is difficult to evaluate an elevated score on a scale measuring “schizophrenia” if it is known that the scale also appears to be measuring alienation, indecisiveness, family problems, and depression. To assure discriminant validity, PAI items were selected that had maximal associations with indicators of the pertinent construct and minimal associations with other constructs—whether they be demographic features, response style, or substantive measures of other forms of psychopathology. Normative Data The PAI scale and subscale raw scores are transformed to T-scores (mean of 50, standard deviation of 10) in order to provide interpretation relative to a standardization sample of 1,000 community-dwelling adults. This sample was carefully selected to match 1995 U.S. census characteristics on the basis of gender, race and age; the educational level of the standardization sample (mean of 13.3 years) was representative of a community group with the required fourth-grade reading level. For each scale and subscale, the T-scores were linearly transformed from the means and standard deviations derived from the census-matched standardization sample. Unlike several similar instruments, the PAI does not calculate PERSONALITY ASSESSMENT INVENTORY (PAI) T-scores differently for men and women; instead, combined norms are used for both genders. The PAI item selection parameters included several procedures intended to eliminate items that might be biased due to demographic features, and items that displayed any signs of being interpreted differently as a function of these features were eliminated in the course of selecting final items for the test. As it turns out, with relatively few exceptions, differences as a function of demography were negligible in the community sample. The most noteworthy effects involve the tendency for younger adults to score higher on the BOR and ANT scales, and the tendency for men to score higher on ANT and ALC relative to women. T scores are derived from a community sample, so they provide a useful means for determining if certain problems are clinically significant, because relatively few normal adults will obtain markedly elevated scores—roughly 84% of community adults obtain a score below 60T on a given scale, while roughly 96% obtain scores below 70T (Morey, 2007). A score of 70T or above, reflecting two standard deviations above the community mean, typically indicates problem severity in a range that merits clinical attention. However, other comparisons are often of equal importance in clinical decision making—for example, it is often informative to understand how the respondent’s scores compare with those obtained in clinical settings. As one example, nearly all patients report some depression at their initial evaluation; the question confronting the clinician considering a diagnosis of major depressive disorder is one of relative severity of symptomatology. Knowing that an individual’s score on the PAI Depression scale is elevated in comparison to the community standardization sample is of value, but a comparison of this elevation relative to a clinical sample may be more critical in distinguishing major depressive disorder from some other condition. To facilitate these comparisons, the PAI profile form also indicates the T scores that correspond to marked elevations when referenced against a representative clinical sample. The 5 “skyline” shown on the profile form indicates the score for each scale that represents a raw score that is 2 standard deviations above the mean for a clinical sample of 1,246 patients selected from a wide variety of professional settings. Scores approaching or exceeding this skyline represent a marked elevation relative to those of patients in clinical settings. There are assessment contexts where other forms of normative comparisons are of interest. For example, the publisher of the PAI offers specialty interpretive reports that provide transformations with respect to norms from applicants for public safety positions and from inmates being assessed in correctional settings. Published normative information is also available for diverse populations, including chronic pain patients (Karlin et al., 2005), patients presenting for bariatric surgery (Corsica et al., 2010) and deployed military troops (Morey et al., 2011). Such information serves as a valuable supplement to interpretation; for example, when the PAI is used in a vocational screening context, some elevation of indicators of defensive responding is typical, which should serve as a consideration in the determination of profile validity. Although it is appropriate to make normative comparisons to a sample most representative of the individual being assessed, for most clinical and research applications the use of T scores derived from the full community normative data is strongly recommended because of its representativeness and large sample size. Reliability The reliability of the PAI scales and subscales has been examined in terms of internal consistency, test–retest reliability, and configural stability across several different studies, many of which are summarized in the test manual (Morey, 2007). Internal consistency alphas for the full scales are generally found to be in the .80s in a variety of different settings, and test–retest reliability in nonclinical samples yields similar estimates. Such reliability values provide an estimate of the standard error of 6 PERSONALITY ASSESSMENT INVENTORY (PAI) measurement of roughly three to four T-score points, with 95% confidence intervals of +/−six to eight T-score points (Morey, 2007). Validity The initial studies of PAI validity presented in the original manual (Morey, 1991) described the convergence of the PAI scales with a number of the best available corresponding clinical indicators, administered concurrently to various samples. Since that time, numerous studies have investigated the validity of the PAI in a variety of populations and for numerous purposes. A comprehensive presentation of available validity evidence for the various scales is beyond the scope of this entry; the second edition of the PAI manual (Morey, 2007) summarizes hundreds of such studies. Some of the more noteworthy findings from such studies are presented in the following paragraphs, divided into the four broad classes of PAI scales: Validity, Clinical, Treatment Consideration, and Interpersonal. Validity Scales The assessment of profile validity is an important component of any method intended for an evaluative context. This issue is compounded in that instruments such as the PAI are frequently used in settings where distorted or dishonest responding might be common, such as pre-employment screens, forensic or psycholegal settings, or disability determinations. The PAI validity scales were developed to provide an assessment of the potential influence of certain response tendencies on PAI test performance, including both random and systematic influences upon test responding. In addition, several supplemental validity indicators have been developed since the test was first introduced in 1991. The extent to which such scales and indicators are useful for clarifying interpretation of psychological tests has been a point of some controversy (McGrath, Mitchell, Kim, & Hough, 2010; Morey, 2012) but there is substantial evidence supporting utility of the various PAI validity markers. Random response influences on the PAI can be identified using the ICN and INF scales. Research demonstrating the utility of these scales to detect random responding has generally compared profiles derived from random response simulations with those obtained from bona fide respondents. Such studies demonstrate a clear separation of scores of actual respondents from the random simulations (Morey, 1991; Clark, Gironda, & Young, 2003), such that nearly all random profiles are identified as scores on either ICN or INF that exceed the clinical skyline. However, Clark et al. (2003) noted that these scales were not particularly sensitive to “back random responding,” where a respondent discontinues attending to item content midway through the examination. To assist in the detection of these partially-random response sets; Morey and Hopwood (2004) developed an indicator involving partial to full-scaled score discrepancies on two PAI scales (ALC and SUI), with satisfactory positive and negative predictive power across different levels and base rates of back random responding. Systematic profile distortion involves response patterns that affect the test results in predictable ways, and can occur in positive or negative directions, and even in both directions in a given profile (with different scales influenced by different factors). Such distortion can also be either effortful/intentional or noneffortful. Effortful distortion occurs when respondents intentionally present themselves in a manner that is at odds with their experience or historical fact (i.e., they might malinger, or fake good). In noneffortful distortion, respondents may present themselves in a manner consistent with their subjective experience but which an experienced clinician might see as an exaggeration (if overly negative) or a lack of insight (if overly positive). Such distortion can be related to specific forms of psychopathology. For example, cognitive symptoms of depression often involve a magnification of personal difficulties, while narcissistic or manic disorders can manifest limited capacity for a critical appraisal of one’s PERSONALITY ASSESSMENT INVENTORY (PAI) real abilities and prospects. The PAI indices of systematic profile distortion were developed to assess self- and other-deception, in both the positive and negative directions. With respect to detecting negative distortion, the NIM scale and supplemental indicators including MAL, RDF, and NDS have all received research support. The most common study methodology has involved simulations where nonclinical respondents attempt to feign the responses of patients with various diagnoses; however, other approaches have included convergence studies using other measures of feigning or negative distortion, as well as against criteria such as staff determinations of patient malingering. A meta-analysis of negative distortion detection research (Hawes & Boccaccini, 2009) concluded that all reviewed negative distortion indicators on the PAI demonstrated large average effect sizes across several studies. The PIM scale and supplemental indicators such as DEF and CDF that were developed to assess positive dissimulation directly have also been examined in several empirical studies. Many of these studies have involved individuals instructed to respond in socially desirable ways under various scenarios (e.g., pre-employment screens, parole determinations, or custody evaluations). However, other studies have used criteria such as the detection of denied substance misuse that is later confirmed by drug screens, or the role of such indicators in moderating the prediction of job performance among individuals hired to be police officers (Lowmaster & Morey, 2012). Although multiple studies have supported the utility of these positive distortion indicators, the effects tend to be smaller than those for indicators of negative distortion (Morey, 2007). Clinical Scales The clinical scales of the PAI were assembled to provide information about clinical constructs relevant in a variety of different contexts. Hundreds of different studies have been conducted exploring the convergent 7 and discriminant validity of these clinical scales, and many of these studies conducted up to 2007 were summarized in the updated test manual (Morey, 2007). A few interesting findings merit mention as illustrations of the diversity of applications for which the PAI has been used. For example, the SOM scale, particularly SOM-C, has proven useful in cases where functional vs. organic origins of physical complaints are an issue, such as efforts to distinguish between epileptic and nonepileptic seizures (e.g., Locke et al., 2011). The ARD scale has subscales that measure diverse anxiety-related conditions. Elevations on the ARD-T subscale are associated with a variety of traumatic stress reactions, including combat-related PTSD as well as the aftereffects of childhood abuse. The DEP and ANX scales, while related to other commonly used measures of negative affect, have served as helpful measures of treatment response, being sensitive to changes associated with treatments for common mental disorders as well as for conditions as varied as right temporal lobectomy, trichotillomania, and irritable bowel syndrome. Elevations on the SCZ scale are associated with thought disorder, involving disruptions in both the process and the content of thinking. In comparison, the PAR scale is more directly focused upon atypical thought content, such as prominent suspiciousness or persecutory delusions. The BOR scale represents a broad marker of personality immaturity related to borderline personality functioning, and elevations are associated with marked impairment in functioning and outcome, and with greater chronicity of problems. Despite being associated with more chronic problems, the BOR scale has been found to be sensitive to changes associated with treatment, such as indicating response to dialectical behavior therapy (Harley, Blais, Baity, & Jacobo, 2007). The ANT scale, a measure of antisociality and psychopathy, has received considerable study in forensic and correctional settings, where it has been found to predict factors such as disciplinary infractions during incarceration and recidivism among inmates after release. 8 PERSONALITY ASSESSMENT INVENTORY (PAI) Among the PAI clinical scales are two indicators of substance use and misuse, ALC and DRG. The items on these scales inquire about behaviors and consequences related to alcohol and drug use, abuse, and dependence. These scales change over time for individuals in substance abuse treatment, suggesting that the scale (unlike formal diagnoses) is sensitive to the biological and psychosocial stabilization that occurs in successfully treatment patients. Because the items for ALC and DRG inquire directly about substance use, it should be recognized that these scales are susceptible to denial or underreporting. Various empirical strategies (including the ALC Est and DRG Est scores) have been developed to assess the likelihood that a profile under-represents the extent of alcohol or drug problems. Treatment Consideration Scales The treatment consideration scales of the PAI were assembled to provide indicators of potential complications in treatment that are not necessarily apparent from diagnostic constructs. For example, judgments about potential aggression or suicide have direct implications for decisions around psychiatric hospitalization, but psychiatric diagnoses associated with these constructs (such as Antisocial Personality or Major Depression) are neither sufficiently sensitive nor specific for such important decisions. Thus, scales such as AGG and SUI were constructed to provide a direct assessment of these constructs, while supplemental indicators such as VPI and SPI provide an overview of factors that might exacerbate risk for dangerousness to self or other. For example, high scores on the AGG scale are associated with greater disciplinary adjustment and aggressive infractions in correctional settings, as well as being associated with recidivism post-release. Predictions of aggressiveness can also be supplemented by the VPI, which is comprised of markers of heightened violence risk and has also been found to be predictive of aggressive behavior. With respect to the assessment of suicide risk, SUI scale elevations indicate pervasive suicidal thoughts and ruminations; SUI scores as well as the supplemental SPI indicator have been found to predict subsequent suicidal gestures. The SUI scale also has been found to be sensitive to the effects of treatments targeting self-damaging behaviors (Harley et al., 2007). The STR and NON scales provide indicators of the respondent’s perception of their environment with respect to the extent of environmental stressors and the availability of social supports to buffer this stress. These scales have been found to be related to the degree to which respondents feel connected to their community and to participation in support groups, with low scores on both NON and STR reflecting the combination with the most favorable prognosis, followed by low NON but elevated STR (typically indicative of an adjustment reaction). Finally, the RXR scale is designed to measure lack of treatment motivation and lack of readiness to change. This scale has been found to be related to treatment noncompliance and premature treatment termination. In studies of treatment dropout, RXR typically interacts with measures of problem severity to predict successful treatment completion. In this regard, the TPI supplemental index can serve to alert the clinician to factors that will predict a difficult treatment, with elevated scores also potentially predictive of risk for treatment attrition. Interpersonal Scales The interpersonal scales of the PAI were designed to provide an assessment of the interpersonal style of subjects along two dimensions: (a) a warm and affiliative vs. a cold and rejecting axis, and (b) an axis characterized by a dominating, controlling vs. a meekly submissive style. These axes correspond to the two main vectors of the “interpersonal circumplex” that has provided a theoretical foundation for a number of different assessments and treatments. A number of studies have demonstrated that the combination of high DOM and low WRM scores, reflecting “hostile control,” is particularly problematic. For example, this pattern has been associated with noncompliance, PERSONALITY ASSESSMENT INVENTORY (PAI) aggressive acts, and overall poor treatment progress in offender populations. Limitations As a self-report questionnaire, it should be recognized that some individuals with limited reading comprehension, intellectual or cognitive disability, or clouding of consciousness may not be capable of completing the PAI; when the test is used in such populations, the validity scales of the instrument should be consulted to assist in this determination. Other particular assessment contexts may also have unique influences upon the response characteristics of the PAI. For example, application of the test in pre-employment hiring determinations, or in correctional contexts, can have systematic effects upon PAI performance, and researchers have collected special norms in these settings to assist in understanding these influences. As a clinical instrument, the PAI is designed to provide information relevant to clinical diagnosis, treatment planning, and screening for psychopathology. It should be recognized that there are a host of clinical conditions, such as eating disorders, neurodevelopmental disorders, or sexual dysfunction, that are not assessed by the test, although in such cases test results might provide useful supplemental information for purposes of differential diagnosis or treatment planning. Furthermore, the inventory is not designed to provide a comprehensive assessment of the domains of normal personality. In clinical settings, diagnostic and screening decisions should not be based exclusively on the results of the PAI. Important decisions necessarily require multiple sources of information, which may include but are not limited to historical data, current mental status, and the results of other assessment instruments. Integration of information from such diverse sources requires the expertise of a qualified professional with training in both psychometric assessment and descriptive psychopathology. In particular, although 9 computer-based interpretive reports are available for the PAI, such reports should never serve as the sole basis of any professional decisions and any such information should always be considered to be one of many sources of hypotheses for professionals making decisions regarding diagnoses, treatment plans, or screening. SEE ALSO: Antisocial Personality Disorder/ Psychopathy; Anxiety Disorders; Circumplex Models; Construct Validity; Forensic Testing; Morey, Leslie (b. 1956); Self-Report Questionnaires References Archer, R. P., Buffington-Vollum, J. K., Stredy, R. V., & Handel, R. W. (2006). A survey of psychological test use patterns among forensic psychologists. Journal of Personality Assessment, 87, 84–94. Clark, M. E., Gironda, R. J., & Young, R. W. (2003). Detection of back random responding: Effectiveness of MMPI-2 and Personality Assessment Inventory validity indices. Psychological Assessment, 15, 223–234. Corsica, J. A., Azarbad, L., McGill, K., Wool, L., & Hood, M. (2010). The Personality Assessment Inventory: Clinical utility, psychometric properties, and normative data for bariatric surgery candidates. Obesity Surgery, 20, 722–731. Harley, R., Blais, M. A., Baity, M. R, & Jacobo, M. C. (2007). Use of DBT skills training for borderline personality disorder in a naturalistic setting. Psychotherapy Research, 17, 362–370. Hawes, S. W., and Boccaccini, M. T. (2009). Detection of overreporting of psychopathology on the Personality Assessment Inventory: A meta-analytic review. Psychological Assessment, 21, 112–124. Karlin, B. E., Creech, S. K., Grimes, J. S., Clark, T. S., Meagher, M. W., & Morey, L. C. (2005). The use of the Personality Assessment Inventory with individuals with chronic pain. Journal of Clinical Psychology, 61, 1571–1585. Locke, D. E. C., Kirlin, K. A., Wershba, R., Osborne, D., Drazkowski, J. F., Sirven, J. I., & Noe, K. H. (2011). Randomized comparison of the Personality Assessment Inventory and the Minnesota Multiphasic Personality Inventory-2 in the epilepsy monitoring unit. Epilepsy and Behavior, 21, 397–401. 10 PERSONALITY ASSESSMENT INVENTORY (PAI) Lowmaster, S. E., & Morey, L. C. (2012). Predicting law enforcement officer job performance with the Personality Assessment Inventory. Journal of Personality Assessment, 94, 254–261. McGrath, R. E., Mitchell, M., Kim, B. H., & Hough, L. (2010). Evidence for response bias as a source of error variance in applied assessment. Psychological Bulletin, 136, 450–470. Morey, L. C. (1991). The Personality Assessment Inventory professional manual. Odessa, FL: Psychological Assessment Resources, Inc. Morey, L. C. (2007). The Personality Assessment Inventory professional manual (2nd ed.). Odessa, FL: Psychological Assessment Resources, Inc. Morey, L. C. (2012). Detection of response bias in applied assessment. Psychological Injury and Law, 5, 153–161. Morey, L. C., & Hopwood, C. J. (2004). Efficiency of a strategy for detecting back random responding on the Personality Assessment Inventory. Psychological Assessment, 16, 197–200. Morey, L. C., Lowmaster, S. E., Coldren, R. L., Kelly, M. P., Parish, R. V., & Russell, M. L. (2011). Personality Assessment Inventory profiles of deployed combat troops: An empirical investigation of normative performance. Psychological Assessment, 23, 456–462. Schinka, J. A., & Borum, R. (1993). Readability of adult psychopathology inventories. Psychological Assessment, 5, 384–386. Further Reading Blais, M. A, Baity, M. R., & Hopwood, C. J. (2010). Clinical applications of the Personality Assessment Inventory. New York, NY: Routledge. Morey, L. C. (1996). An interpretive guide to the Personality Assessment Inventory. Odessa, FL: Psychological Assessment Resources, Inc. Morey, L. C. (2003). Essentials of PAI assessment. New York, NY: John Wiley & Sons. Morey, L. C., & Hopwood, C. J. (2007). Casebook for the Personality Assessment Inventory: A structural summary approach. Odessa, FL: Psychological Assessment Resources.