psychometric properties of a psychological assessment of your choice.
Personality Assessment Inventory–Adolescent
Morey, Leslie C.
Psychological Assessment Resources, Inc., 16204 N. Florida Avenue, Lutz, FL, 33549-8119, firstname.lastname@example.org, www.parinc.com
R. A. Spies, J. F. Carlson, & K. F. Geisinger (Eds.), The eighteenth mental measurements yearbook. 2010.
Kade, H. Dennis; Sandoval, Jonathan
An ‘objective test of personality designed to provide information on critical client variables in professional settings.’
53: Somatic Complaints (Conversion, Somatization, Health Concerns, Total), Anxiety (Cognitive, Affective, Physiological, Total), Anxiety-Related Disorders (Obsessive-Compulsive, Phobias, Traumatic Stress, Total), Depression (Cognitive, Affective, Physiological, Total), Mania (Activity Level, Grandiosity, Irritability, Total), Paranoia (Hypervigilance, Persecution, Resentment, Total), Schizophrenia (Psychotic Experiences, Social Detachment, Thought Disorder, Total), Borderline Features (Affective Instability, Identity Problems, Negative Relationships, Self-Harm, Total), Antisocial Features (Antisocial behaviors, Egocentricity, Stimulus-Seeking, Total), Alcohol Problems, Drug Problems, Aggression (Aggressive Attitude, Verbal Aggression, Physical Aggression, Total), Suicidal Ideation, Stress, Nonsupport, Treatment Rejection, Dominance, Warmth, Inconsistency, Infrequency, Negative Impression, Positive Impression.
2021: $520 per introductory kit including professional manual (2007, 190 pages), 2 hardcover reusable item booklets, 2 administration folios, 25 hand-scored answer sheets, 25 profile forms-adolescent, and 25 critical items forms-adolescent; $123 per professional manual (print or digital); $59 per hardcover reusable item booklet; $70 per 10 softcover reusable item booklets; $93 per 25 hand-scored answer sheets; $65 per 25 profile forms-adolescent; $56 per 25 critical items forms-adolescent.
Designed to complement its parent instrument, the Personality Assessment Inventory (18:93)
For reviews by H. Dennis Kade and Jonathan Sandoval, see 18:92.
Spanish forms available.
Published Test Description:
Personality Assessment Inventory-Adolescent. Purpose: An “objective test of personality designed to provide information on critical client variables in professional settings.” Population: Ages 12-18. Publication Dates: 1990-2007. Acronym: PAI-A. Scores, 53: Somatic Complaints (Conversion, Somatization, Health Concerns, Total), Anxiety (Cognitive, Affective, Physiological, Total), Anxiety-Related Disorders (Obsessive-Compulsive, Phobias, Traumatic Stress, Total), Depression (Cognitive, Affective, Physiological, Total), Mania (Activity Level, Grandiosity, Irritability, Total), Paranoia (Hypervigilance, Persecution, Resentment, Total), Schizophrenia (Psychotic Experiences, Social Detachment, Thought Disorder, Total), Borderline Features (Affective Instability, Identity Problems, Negative Relationships, Self-Harm, Total), Antisocial Features (Antisocial Behaviors, Egocentricity, Stimulus-Seeking, Total), Alcohol Problems, Drug Problems, Aggression (Aggressive Attitude, Verbal Aggression, Physical Aggression, Total), Suicidal Ideation, Stress, Nonsupport, Treatment Rejection, Dominance, Warmth, Inconsistency, Infrequency, Negative Impression, Positive Impression. Administration: Group. Price Data, 2008: $295 per complete kit including professional manual (2007, 190 pages), 2 reusable item booklets, 2 administration folios, 25 hand-scored answer sheets, 25 profile forms-adolescent, 25 critical items forms-adolescent, and 1 professional report service answer sheet in a soft-sided attaché case; $32 per reusable item booklet; $34 per 10 soft cover item booklets; $48 per 25 hand-scored answer sheets; $30 per 25 profile forms-adolescent; $35 per 25 critical items forms-adolescent; $70 per professional manual. Time: 45(55) minutes. Comments: Designed to complement its parent instrument, the Personality Assessment Inventory. Authors: Leslie C. Morey. Publisher: Psychological Assessment Resources, Inc.
Translate Full Text:
Personality Assessment Inventory–Adolescent
Review of the Personality Assessment Inventory-Adolescent by H. DENNIS KADE, Director of Developmental & Behavioral Health, Norfolk Department Of Public Health and Tidewater Child Development Clinic and Adjunct Assistant Professor of Psychology, Old Dominion University, Norfolk, VA:
DESCRIPTION. The Personality Assessment Inventory-Adolescent (PAI-A) is a self-report personality questionnaire for clinical assessment at ages 12 to 18 years. It was developed as an extension of the adult Personality Assessment Inventory (PAI) for adolescents. The manual states that the PAI-A is not appropriate for comprehensive assessment of normal personality domains. Analysis of the test instructions and items shows reading comprehension at a 4.0 grade level is required. The Flesch Reading Ease score is 80.9, suggesting a fifth grade level. This is within the easy range, perhaps harder to read than a comic but easier than a teen magazine.
The test manual indicates there are 264 items that take adolescents 30-45 minutes to complete and examiners 10-15 minutes to score, yielding 22 nonoverlapping scales. Each item is answered on a 4-point scale ranging from false/not at all true, to slightly true, mainly true, or very true. The written instructions for the adolescent are succinct, but contain no example item. The question booklet and answer sheet items are arranged in columns. Breaks in one do not correlate with breaks in the other, but the answer sheet layout facilitates scoring without keys. Alternating colors are used to help the adolescent visually separate the rows on the answer sheet. A critical item sheet facilitates inspection of 17 answers with space for notes from subsequent queries. With no age or gender differences, only one profile sheet is needed to convert raw scale and subscale scores to T-scores. Reference to normative tables in the manual is unnecessary. The examiner’s manual is well organized with a detailed table of contents that facilitates its use as a reference.
Hand score and optical scan score answer sheets are available; the latter are also used for the publisher’s mail-in scoring service. Computer-based administration, scoring, and interpretive report software is available. Though its purchase price is more than the PAI-A test kit itself, additional per-use fees are charged only for on-screen administrations. Hand scoring is quite straightforward and a scoring example is included in the manual. Guidelines for interpretation included in the manual extend beyond single scales at different elevations to suggest profiles based on cluster analysis. Unfortunately, no example clinical cases are supplied that might illustrate integration of the interpretive guidelines. The manual states that PAI-A interpretation requires training in psychometric assessment and psychopathology.
The four validity scales included in the PAI-A support decisions about the validity of an individual’s protocol. There are 10 pairs of items in the Inconsistency scale; completely random responding produced an average T-score of 82 on this scale and scores above 77T render the protocol invalid. The 7 items of the Infrequency scale were rarely endorsed in the normal and clinical samples; random responding produced an average T-score of 82, and scores above 78T indicate invalid protocols. The 8 items of the Negative Impression scale were infrequently endorsed in the clinical and particularly in the normal samples; random responding produces an average T-score of 77, and a sample instructed to malinger averaged 108T. Scores above 84T indicate invalid protocols. The 8 items of the Positive Impression scale were endorsed most frequently by a sample instructed to respond with positive impression enhancement (average score of 68T), least often by clinical cases, with the normal sample falling in between. Scores above 71T indicate invalid protocols. The manual states that protocols with more than 13 unanswered items are problematic and scales or subscales with more than 19% unanswered should not be interpreted.
DEVELOPMENT. Both the structure and most of the items from the PAI were retained in the PAI-A. Thus, constructs relate to the diagnostic categories of mental disorder and the experience of symptoms and other factors relevant to treatment (e.g., Aggression, Suicidal Ideation, Stress, Nonsupport, Treatment Rejection) that occur across the life span. A few items had to be reworded to make them more age-appropriate (e.g., changing “work” to “school”), but new items were not introduced because the PAI-A was not intended to assess psychopathology unique to adolescents. Pilot testing used 275 clinical cases who were 65% male, 54% from forensic (juvenile detention) settings, 26.7% from inpatient mental health, and 22.8% from outpatient mental health. Internal consistency and mean interitem correlations within scales were used to eliminate 80 items. One item was omitted from scoring on the Infrequency scale because adolescents endorsed it too often. The PAI-A item pairs used to score the Inconsistency scale were selected empirically, rather than using the PAI items.
TECHNICAL. The test manual provides information on the PAI-A’s empirical structure such as scale intercorrelations, exploratory factor analyses, confirmatory factor analyses of subscales, and cluster analyses. Normative data for the U.S. Census-matched standardization sample and (for comparison purposes) the clinical sample are included in appendices, as well as data on item means and standard deviations in both samples, scale and subscale correlations with the validation measures in the community sample, and frequency distributions of clinical scale codetypes across nine diagnoses plus the entire clinical sample.
T-scores are based on a community sample of 707 students in junior and senior high school and college who were between the ages of 12 and 18 years from 21 states stratified by race/ethnicity and gender within age bands to match the 2003 U.S. Census. Analysis showed only two scales had more than 5% of their variance accounted for by ethnicity, gender, or age: higher scores for males on Antisocial Features and higher scores for non-Caucasians on Paranoia. These differences are about 5 T-Score points.
Data were also gathered on 1,160 representative clinical cases from 78 sites: outpatient mental health (49.7%), juvenile/correctional (33.3%), inpatient mental health (12.7%) and other settings. The clinical sample data were used to generate a clinical “skyline” on the profile form that exceeds 98% or is 2 standard deviations above the mean. Evidence is provided that the most common diagnoses in treatment are represented, but there are fewer than 5% with an anxiety disorder.
Standard error of measurement values for scales in the community sample average 4.68 T-score units (range of 3.3 to 5.9) and subscales average 4.94 (range of 3.42 to 6.42). Coefficient alpha internal consistency values are reported in community and clinical samples for scales and subscales. Internal consistency for scales (except Inconsistency and Infrequency) averaged .79 for the community standardization sample and .80 for the clinical sample. Except for Dominance in the clinical sample, all scale values meet or exceed the .70 traditionally considered acceptable for research use, but only Aggression in the community sample and Suicidal Ideation in the clinical sample reach the .90 standard for making decisions about individuals. Only 13 of the 31 subscales meet or exceed .70 in the community sample and 18 in the clinical sample; none reach the .90 standard. These results are not surprising given the brevity of the scales and subscales. The test author is to be commended for addressing the stability of profiles, though the only datum reported was a median inverse correlation of .84 for the relative position of scale scores within the profiles of test-retest cases. The latter 100 participants were reassessed to yield an average test-retest stability coefficient of .78 for the scales (except Inconsistency and Infrequency) after a mean of 18 days (range from 9 to 35 days).
Six validity studies with clinical and community samples ranging from 77 to 1,160 are reported that provide a wealth of support for criterion-related validity. Only clinical diagnosis and symptom checklist were used in the largest sample, but other samples used multiple measures of personality and psychopathology: Minnesota Multiphasic Personality Inventory-Adolescent (MMPI-A), Adolescent Psychopathology Scales (APS), Personality Inventory for Youth (PIY), NEO Five-Factor Inventory (NEO-FFI), Symptom Assessment-45 (SA-45), College Adjustment Scales (CAS), Clinical Assessment of Depression (CAD), Adolescent Anger Rating Scale (AARS), Beck Depression Inventory (BDI), and State-Trait Anxiety Inventory (STAI). Exploratory factor analysis using a principal components approach with orthogonal rotation produced four factors accounting for 41%, 10%, 9%, and 6% of the variance with the first interpreted as representing psychological distress.
COMMENTARY. Like other tests linked to diagnostic nosology, the adequacy of the theoretical model behind the PAI-A is limited by the adequacy of the diagnostic model itself. The test’s focus on core characteristics of diagnoses without directly attempting to replicate all diagnostic criteria may prove advantageous if more peripheral criteria change over time or are unimportant to decisions made by the user. As the manual points out, research continues to evolve on some diagnoses such as bipolar mood disorder in adolescents, so that there is no current gold standard by which to judge the PAI-A for some disorders. It would be useful if more information was presented showing the PAI-A’s power to discriminate constructs and diagnoses, such as depression/mood disorders and anxiety/anxiety disorders. In the clinical sample, less than 4% had diagnoses of posttraumatic stress disorder or mania/hypomania, and less than 2% had borderline personality disorder (the only Axis II diagnosis tabulated). Current clinical practice with adolescents often requires a differential between these and other diagnoses, particularly when serving a population with a history of early maltreatment and trauma. Future editions of the manual would also benefit from data on the scales’ sensitivity to treatment effects, the relationship of PAI-A scores to diagnoses made by research criteria, gender/racial/ethnic/cultural differences by diagnosis such as males versus females with Attention Deficit Hyperactivity Disorder, and on profile stability in adolescents 18 years of age and older who complete the PAI after the PAI-A.
SUMMARY. The PAI-A achieves its stated purpose of providing a parallel form that extends the PAI to adolescents. More information on discriminative validity and on the ability to detect more subtle denial and malingering is needed. The PAI-A provides an attractive option to the growing number of other self-report measures available for this age group in clinical decision-making (many summarized in Sattler & Hoge, 2006; extensive reviews of several are in Reynolds & Kamphaus, 2004). Self-report measures that parallel parent and teacher ratings typically lack validity scales. The Millon Adolescent Personality Inventory has significant shortcomings (Sattler & Hoge, 2006), and the other Millon scales for this age group have only clinical norms. The PAI-A has the advantage of ease and economy of scoring over the MMPI-A, another downward extension of an adult test of psychopathology. Also, the MMPI-A takes 60-90 minutes to complete and a seventh grade reading level is recommended, but many items are even more difficult (Sattler & Hoge, 2006). The PIY might be considered as an alternative to the PAI-A when empirical scales rather than diagnosis-based scales are acceptable (see review by Sattler & Hoge, 2006). If not, then the APS is a good alternative, but clinical experience has shown the APS has difficulty detecting cases of sophisticated denial. When there is an anticipated need to reevaluate an adolescent over the age of 18, then a clear advantage to the PAI-A is allowing future testing to be based on a very similar measure, the PAI.
Sattler, J. M., & Hoge, R. D. (2006). Assessment of children: Behavioral, social and clinical applications (5th ed.) San Diego: Jerome M. Sattler.
Reynolds, C. R., & Kamphaus, R. W. (Eds.). (2003). Handbook of psychological and educational assessment of children: Personality, behavior and context (2nd ed.). New York: Guilford Press.
Review of the Personality Assessment Inventory-Adolescent by JONATHAN SANDOVAL, Professor of Education, University of the Pacific, Stockton, CA:
DESCRIPTION. The Personality Assessment Inventory-Adolescent (PAI-A) is an extension to 12- through 18-year-olds of the Personality Assessment Inventory (Morey, 1991). Both measures may be used with 18-year-olds. The self-report personality measure consists of 264 items that generate 4 Validity scales, 11 Clinical scales, 5 Treatment Consideration scales, and 2 Interpersonal scales. Nine of the 11 Clinical scales and the Aggression Treatment Consideration scales have from three to four subscales based on 5 to 6 items each. Other scales typically are derived from about 8 items. In total there are 22 scales and 31 subscales expressed as linear T scores. On the PAI-A, a respondent indicates if a statement is False, Not At All True; Slightly True; Mainly True; or Very True about himself or herself. The intent is to measure constructs relevant to the presence of mental disorder. It is not intended to be a general measure of personality in normal individuals. The materials include a professional manual, an item booklet, hand-scored and optical scan answer sheets, a Critical Items form, and a Profile form. Computer administration and scoring options are available from the test publisher. The Critical Items form consists of 17 items from the PAI-A suggesting behavior or psychopathology demanding immediate attention, such as Suicidal Ideation. The form permits a follow-up interview to explore the critical response. The profile form permits the examination of patterns of scores across the 22 areas, particularly 10 clusters of modal scores identified by statistical analysis that are associated with particular groups such as substance abusers.
The test author states that the interpretation of profiles and test score patterns must be done by qualified and experienced clinical professionals. The examiner’s manual is straightforward and provides information about administration considerations, administration procedures, scoring procedures and options, and test feedback to test-takers. Included is a chapter on interpretation that discusses the rationale and construct underlying each scale and subscale and possible meanings of high scores. The PAI-A is not intended to be used mechanically to provide a definitive diagnosis but is rather to provide information relevant to clinical diagnosis, treatment planning, and screening for psychopathology. The information from the PAI-A should supplement information from multiple sources such as case histories, clinical interviews, and other mental tests.
DEVELOPMENT. The test author designed the PAI-A to retain the structure and most of the item content of the adult inventory. A few items were reworded to fit adolescent experience. Items were carefully selected based on 15 criteria involving expert opinion, discriminant validity, reading level (fourth grade), positive and negative response sets, and item statistics such as differential item functioning. A total of 344 items were pilot tested with 275 adolescents with mental health diagnoses. Some 80 items with different statistical characteristics for adolescents and adults or other problems were eliminated from the standardization version.
TECHNICAL. Norms for the scores were based on 707 adolescents obtained from sites in 21 states in the U.S. The norm sample closely matched the U.S. Census data from 2003 with regard to gender, age, race/ethnicity, but not geographical region, although all regions were represented as well as urban and rural areas. Participants were recruited in a variety of ways including through schools in targeted areas. The sample was not random, but purposive, and data were collected on 1,032 adolescents to create the demographic match. Norms are not broken down by age, ethnicity, or gender, as observed group differences were typically within the range of measurement error and account for less than 5% of the variance in scores.
Normative data of a different sort were obtained from a clinical sample of 1,160 adolescents identified by clinicians working in 78 institutions such as inpatient or outpatient mental health settings and juvenile correctional settings as having a mental illness or emotional disturbance. The clinical sample was 58.4 % male and 72.3% Caucasian, 19.8% African American, 4.5% Hispanic, and 4.4% other with a pattern of diagnoses comparable to those found in another large-scale study, with the exception that depression was somewhat overrepresented and anxiety disorder underrepresented, compared to the other study.
Evidence of internal consistency and stability of the PAI-A indicates the test has reasonable reliability. Coefficient alpha estimates for the normative sample scores on the 20 substantive scales range from .70 for the Positive Impression Scale to .90 for Aggression with a mean of .79. A similar pattern of alpha values was found for the clinical sample, with remarkable consistency across ages, genders, and ethnicities. The values for the subscales, based on fewer items, was lower, ranging from .47 for Anxiety-Related Disorders-Phobias to .85 Anxiety-Related Disorders-Traumatic Stress, with a mean of .69 for the normative sample. Again, the same pattern and levels of alpha was true of the clinical sample. Test-retest stability over a 9- to 35-day interval based on 100 adolescents in the normative sample for the substantive scales ranged from .65 for Positive Impression to .89 for Somatic Complaints with a mean of .78. Coefficients for the subscales ranged from .59 for Mania-Irritability to .88 for Aggression-Physical Aggression with a mean of .76. The manual also reports stability of PAI-A profiles over time.
The test author provides validity information based on a number of studies correlating the PAI-A scores with clinician ratings and with related scores on other commonly used diagnostic measures such as the Minnesota Multiphasic Personality Inventory-Adolescent and the Beck Depression Inventory. Also examined were differences in performance between the normative and clinical samples and the level of performance of the clinical sample on particular scales. The validity scales of Inconsistency, Infrequency, Negative Impression, and Positive Impression were also studied by computer modeling and experimentation with groups of adolescents instructed to “fake bad” or “fake good.” In general, correlations with scales from other measures were in the expected moderate range of .40 to .70. The highest correlations were found for measures of Anxiety and Depression. More modest correlations were found for the Suicidal Ideation, Stress, Treatment Rejection, and Dominance Scales of the PAI-A and parallel measures. In general, patterns of scores of groups from the clinical sample were consistent with their identified disorder. An impressive amount of validity data from studies by the test author and others is provided in the manual and indicates the measures are as valid as other comparable tests. Studies of the internal structure of the PAI-A include an examination of scale intercorrelations, exploratory and confirmatory factor analysis, and cluster analysis. The pattern of intercorrelations is similar for the standardization sample and the clinical sample and related scales have moderate to high correlations within expected domains such as Anxiety, Antisocial Behavior, and Psychosis. A factor analysis of the scales in both samples yielded four orthogonal factors related to internalization, externalization, substance abuse, and social alienation. These factors are consistent with the findings on other measures. The confirmatory factor analysis validated the subscale structures of the PAI-A scales indicating excellent goodness of fit between the subscales and their assigned scales. The manual discusses the method of cluster analysis used to identify 10 profiles on the PAI-A, and to the extent that these profiles make sense, provides some support for the validity of the test.
COMMENTARY. Given the vague description of how the normative and clinical samples were obtained, it is difficult to know what biases may have crept into the norms. This is not a critical issue, but should be taken into account in interpreting the test. In any case, because of the developmental variability of adolescents, care should be taken to consider the measurement error around an obtained score. Otherwise, the test was carefully developed using modern psychometric techniques. The issue of test-fairness for minority groups was considered in test development, but more validity work in this area would be welcome. Clearly the test shows promise in clinical applications and may be considered a viable option to other traditional self-report measures for adolescents. The fourth grade reading level and somewhat fewer self-report items are an advantage in soliciting cooperation.
SUMMARY. The PAI-A is a well-designed downward extension of the PAI. This self-report inventory, with good measures for detecting uninterpretable results, should aid in clinical diagnosis when used in conjunction with information from other sources. The norms, reliability, and validity information justify the careful and judicious use of the test by qualified clinicians.
Morey, L. C. (1991). Personality Assessment Inventory professional manual. Odessa, FL: Psychological Assessment Resources.