3. Test Structure
Assessment of normal personality
Referred to as “the sane man’s MMPI” (Thorndike, 1959)
Purpose of Test:
Predict one’s behaviors
Identify ways that person is described by others
Theory:
No theoretical basis, but has a model
172/434 questions from MMPI
True/false questions
Ex: I often lose my temper.
Immediate cross-cultural relevance
Functional validity
(Groth-Marnat, G., 2009)
4. Development
1957- CPI 480 (Harrison Gough)
18 Folk scales
1987- CPI 462
18 items omitted
2 Folk scales added
3 vector scales added- 23 scales total
1996- CPI 434
28 items omitted, retained same scales
2002- CPI 260
Special purpose scales
(Groth-Marnat, G., 2009)
13. Administration & Scoring
Administration
Originally designed for group
administration; however, it can
be administered individually
Length of time for
administration is 45-60
minutes
Level C Qualification to
Administer
Taken on a computer or with
pencil and paper
Scoring
Computer scoring programs
used for basic profile and
special scales
Raw scores transferred to
profile sheet and converted to
T-scores
-Standard Scores with a mean
of 50 and Standard Deviation
of 10 (Megargee, 1972)
16. Appropriate Use
Academic Counseling
Identifying Leaders
Predicting Success
“The test has generally proven to be a useful tool in the area of prediction and, as
a result, has been particularly helpful in counseling high school and college students as
well as in personnel selection” (Groth-Marnat, 2009, p. 341).
Career Counseling
Six special purpose scales
Clinics and Counseling Agencies
Evaluating Substance Abuse
Susceptibility to Physical Illness
Marital Discord
Juvenile Delinquency and Criminality
Social Immaturity
Cross Cultural and other Research
Consulting Psychologists Press, 1995
17. Inappropriate Use
For diagnostic purposes
To evaluate and predict a specific, internal, unidimensional
trait
To hypothesize construct-oriented life history indices
(Sarchione, et al.,1998)
To use with psychiatrically disturbed individuals
(Sarchione, et al., 1998)
18. Important to Know Prior to Use
Who you’re testing
Normal individuals ages 13 and older
Test requires a fifth-grade reading level
What you’re testing
Measure and evaluate interpersonal behavior and social
interaction
“The goal of the inventory is to give a true-to-life
description of the respondent, in clear, everyday
language, in formats that can help the client to
achieve a better understanding of self.” (Gough and
Bradley, 2005, p. 1).
20. Internal Validity
Extensive empirical evidence
Construct validity (Folk and Vector scales):
Moderate to strong correlations with other personality instruments (.4-.8)
Criterion validity:
California Q-sets (trained observers rated respondents on behavior characteristics):
.1 - .4 (low to moderate)
Adjective Check List (those who knew them rated them): .1-.4 (low to moderate)
Predictive validity
Most concerned with ability of scales to make accurate predictions
Less concerned with scales avoiding overlap or if scales are psychometrically valid
Not a measure of “traits” but the likelihood that someone will behave in a certain way
“Predictive power” consistent but weak (Gough & Bradley, 1996)
Certain subscales have better validity than others
Groth-Marnat, 2009; Gough & Bradley, 1996
21. Construct Validity
CPI and MCMI
High degree of overlap in scales (Holliman & Guthrie, 1989)
259 of the 360 possible MCMI-CPI scale combinations correlate
significantly at the p<.01 level (43% of CPI variance can be accounted for
by MCMI; 45% of MCMI variance can be accounted for by CPI)
Scales unique to each but measuring lots of similar personality
dimensions
CPI and NEO-PI
All of folk scales meaningfully related to one or more of five factors (McCrae,
Costa, & Piedmont, 1993)
Intra-class correlations: N = .57; E = .96; o = 59; A = .71; C = .88 (moderate
to good agreement) (McCrae, Costa, & Piedmont, 1993)
Four out of five factors correlated highly with CPI scales (Agreeableness
factor only minimally represented) (Groth-Marnat, 2009)
22. Construct Validity
CPI and MMPI
200 items overlap
Developed in same way
Empirical method of test construction
Internal consistency analyses
Crites, 1964
23. Internal Validity Subscales
Three scales, within Folk scales, that test for validity of test answers:
Well-being (Wb): faking bad (at or below 30)
Good impression (Gi): faking good (at or above 70) or faking bad
(at or below 30)
Communality (Cm): standard approach (at or above 50) or invalid
results (at or below 30)
Groth-Marnat, 2009
24. External Validity
Old Normative data:
Large sample size: 3,000 males and 3,000 females
High-school (50%) and undergraduate (16.7%) students strongly represented
Negative:
Not random or representative
Information lacking regarding ethnicities, geographic locations, and socioeconomic background
Certain groups underrepresented (adults working in professional occupations)
New Normative data:
New norms in manual for 52 samples of males and 42 samples of females
1000 men and women who are more representative of population using it (Van Hutton, 1990)
Much research has been done to show that CPI can be used with diverse populations
Result:
Need to also compare normed scores with raw scores of similar population groups, such as:
CPI manual has a lot of reference tables for this purpose
Research of CPI with diverse population groups
Conclusion:
Mixed data on its external validity
Gough, & Bradley, 1996
25. Reliability
Test-Retest Reliabilities:
Individual scales: range from .51 (Flexibility) to .84
(Femininity/Masculinity)
Overall median reliability: .68 (CPI 434) and .66 (CPI 260)
Internal Consistency:
Considerable variability among subscales but adequate
Individual scales: .43 (Masculinity/Femininity) - .85 (Well Being)
Lots of Variance = bad (speculation on reasons)
Three Vector scales: .77 - .88
Cronbach’s alphas for scales: .62 - .84
Correlations between CPI 434 and CPI 260: .81 to .97 = High
Thus most of validity numbers apply to both
Conclusion: Decent reliability but lots of variability between subscales
Groth-Marnat, 2009; Gough & Bradley, 1996
26. Reliability of Specific Subscales
Result: Due to variation among subscales in reliability and, evaluate CPI on
specific dimensions of interest
Examples:
CPI-So subscale: good concurrent reliability and acceptable internal
consistency in alcoholic patients (Kadden, Litt, Donovan, & Cooney, 1996)
Significantly predict treatment and outcomes among alcoholic patients (Kadden, Cooney,
Getter, & Litt, 1989)
CPI: predictive of criminal behaviors
Study by Gough & Bradley (1992): mean differences found on 25 subscales for men and
26 subscales for women, out of 27 scales (CPI-So subscale: best differentiator with point-
biserial correlations of .54 for men and .58 for women)
CPI-So subscale: Hundreds of studies show that it predicts antisocial and prosocial
behavior (Collins & Bagozzi, 1999)
Meta-analysis by Collins & Griffin (1998): p = .61 (criminal behavior); p = .35 (antisocial but
not illegal behavior)
27. Reliability
Factor Analysis (aka “Cluster Analysis”):
Establishes reliability (compares) whole test with other personality tests
Establishes reliability of subscales (new and old)
Factor structure also within each subscale
Factor Analysis Factor Structure (4-5 factors) –[Slide 6]
Exception = male and female populations (different factor structures)
Main Factor Structure similar to the core five factors of personality
(Neuroticism, Extraversion, Openness to Experience, Agreeableness,
and Conscientiousness)
Measuring core aspects of personality
Agreeableness not as well represented
Gough & Bradley, 1996; Van Hutton, 1990; Groth-Marnat, 2009
28. Reliability
Purpose of Test:
Predict one’s behaviors
Identify ways that person is described by others
Factor Analysis inconsistent with test’s purpose/goals but:
Criticism that subscales weren’t based on it
Suggestion that if built upon certain factors, would have less variance
Groth-Marnat, 2009; Gough & Bradley, 1996
30. Cultural Applications
• Developed using “Folk concepts”
• Translated into more than 40 languages
• Appropriate for normal persons, so addresses issues that interest diverse
groups
• A choice for cross-cultural personality study because its scales were
designed to represent “dispositions having universal status” (Gough, 1965,
p.379)
31. Cultural Applications
• External validity has been tested across cultures:
• Often focusing on an individual scale of the CPI
(example: Socialization and Femininity/ Masculinity)
• Socialization was researched in 10 different countries and with every country having
supportive results
• Over 17 different countries examined sex differences (Femininity/Masculinity) and in
every country the prediction of respondent gender was significantly supported
• Minimum degree of Cultural Bias
32. Ethnicity
• European Americans
• African Americans
• Native Americans
• Research conducted by Davis, Hoffman, & Nelson, (1990) examined the difference of
CPI results between Native Americans and Whites of similar age, education, and
socioeconomic status
• Men: less conventional and less sensitive to violations of norms when compared with European
American men
• Women: more passive, less verbally controlling, more likely to be comfortable in the
background, and likely to solicit input and support in decision- making when compared with
European American women
• CPI responses need to be compared to cultural norms and considerations of ethnic
background taken into account
Davis, Hoffman, & Nelson, 1990
33. Gender
• Men and women score differently on the CPI
• CPI tests for Femininity and Masculinity common traits that apply to a vast amount
of cultures of men and women
• Gender was found to be significantly different across cultures but not within
cultures
34. Around the World
• Factor structure of CPI tested cross-culturally in different areas, other than the
United States
• Research in a wide variety of countries supports CPI’s validity, even in
countries culturally quite different from the United States
• CPI able to make accurate predictions cross-culturally, such as:
• predictions of academic achievement in Greece
• Detect “faking good” among Norwegians
• Distinguishing from delinquents from non-delinquents in Sweden
• Japan
35. Cultural Limitations
• Additional research needs to be conducted on the
relationship between CPI scores and race, socioeconomics
status, and other demographic variables
• Future research need to be conducted on the ability of the
CPI to predict behaviors in a specific cultural group context
• CPI responses need to be compared to cultural norms, and
considerations of ethnic background taken into account
36. Criticisms
Initial lack of appropriate, representative norming
samples
Mainly representative of Caucasian, college students
Now better norming samples
High level of variance among subscales
Certain scales more valid and reliable than others
Reliability and validity could be better
Not developed based on factor structure
May have helped high levels of variance
Developing factor structure later not consistent with
test’s original goals (Gough & Bradley, 1996)
37. Criticisms
Item overlap among subscales
Lack of theoretical guidelines
Lack of justification of criteria used to develop folk scales
(Gough & Bradley, 1996)
Not easily available
High cost
Manual, Item Booklet, Interpretation Guide and a
Packet of Answer Sheets = $462 (Consulting
Psychological Press, 1995)
Level C Qualification required (doctoral degree)
38. Strengths
Comprehensive coverage of personality traits
26 scales! (not including special purpose scales)
Empirically supported over time (lots of research!)
Strong predictive and construct validity (MCMI, NEO, MMPI)
Item overlap (Gough & Bradley, 1996)
Easy scoring (computer)
Easy to understand
5th grade reading level and True/False questions
Adaptable
Functional validity cross-culturally and among various
subscales (especially Socialization)
Two different test formats (long or short)
Group or individual administration
39. References
Collins, J., & Bagozzi, R. (1999). Testing the equivalence of the socialization factor structure for criminals and noncriminals.
Journal Of Personality Assessment, 72(1), 68-73.
Collins, J., & Griffin, R. (1998). The psychology of underlying counterproductive job performance. In R. W. Griffin, A.
O’Leary-Kelly, & J. M. Collins (Eds.), Dysfunctional work behavior in organizations: Monographs in
organizational behavior and industrial relations (Vol. 23, part B). Stanford, CT: JAI.
Consulting Psychologists Press (1995). CPI 434: Narrative Report. CPP Inc. Retrieved from:
https://www.cpp.com/Pdfs/smp210128.pdf
Consulting Psychologists Press (2002). Technical Brief for the CPI 260® Instrument. CPP Inc.
Consulting Psychologists Press (2003). CPI 260® Client Feedback Report. CPP Inc. Retrieved from:
https://www.cpp.com/Pdfs/smp219250.pdf
Crites, J. (1964). Test reviews: The California Psychological Inventory: I. As a measure of the normal personality. Journal Of
Counseling Psychology, 11(2), 197-202.
Gough, H., & Bradley, P. (1992). Delinquent and criminal behavior as assessed by the revised California Psychological
Inventory. Journal Of Clinical Psychology, 48(3), 298-308.
Gough, H., & Bradley, P. (1996). CPI manual (3rd ed.). Mountain View, CA: CPP, Inc.
Gough, H. & Bradley, P. (2005). CPI 260TM Manual. Mountain View, CA: CPP, Inc.
Groth-Marnat, G. (2009). Handbook of Psychological Assessment. John Wiley & Sons.
Holliman, N., & Guthrie, P.(1989). A comparison of the Millon Clinical Multiaxial Inventory and the California
Psychological Inventory in assessment of a nonclinical population. Journal Of Clinical Psychology, 45(3),
373-382.
40. References
Kadden, R., Cooney, N., Getter, H., & Litt, M. (1989). Matching alcoholics to coping skills or interactional
therapies: Posttreatment results. Journal of Consulting and Clinical Psychology, 57, 698-704.
Kadden, R., Litt, M., Donovan, D., & Cooney, N. (1996). Psychometric properties of the California
Psychological Inventory Socialization scale in treatment-seeking alcoholics. Psychology of
Addictive Behaviors, 10, 131-146.
Lanning, K., & Gough, H. (1991). Shared variance in the California Psychological Inventory and the
California Q-Set. Journal Of Personality And Social Psychology, 60(4), 596-606.
McCrae, R., Costa, P., & Piedmont, R.(1993). Folk concepts, natural language, and psychological
constructs: The California Psychological Inventory and the five-factor model. Journal Of
Personality, 61(1), 1-26.
Megargee, E. (1972). The California Psychological Inventory Handbook. San Francisco: Jossey-Bass, Inc.
Publishers.
Sarchione, C., Cuttler, M., Muchinsky, P., & Nelson-Gray, R. (1998). Prediction of Dysfunctional Job
Behaviors Among Law Enforcement Officers. Journal of Applied Psychology, 83(6), 904-912.
Van Hutton, V. (1990). Test review: The California Psychological Inventory. Journal Of Counseling &
Development, 69(1), 75-77.
Editor's Notes
Describing vs. predicting
Not created to describe personality.
Functional validity
Readily understood by a wide range of people and has a high degree of power in predicting behavior.
Folk scales are in terms that most people can understand (going back to the functional validity)
Empirical criterion keying: criterion groups formed- how they respond to a set of question is then applied.
Empirical relationship is more important than the “truth” of the content.
Ex: “I have never done anything hazardous just for the thrill of it.”
Does not matter whether or not the person ever did do anything hazardous for the thrill of it; it helps differentiate responsible from irresponsible.
Rational Approach: Questions that form a conceptual point of view or seemed to assess characteristics that the scale was trying to measure.
Interpersonal Aspects
Self- confidence, poise, ascendancy, and social effectiveness
Internal Values and Normative Expectations
Maturity, personal values, self-control, and sense of responsibility
Achievement Needs and Cognitive Tendencies
Motivation, persistence, and organization
Stylistic Preferences
Insightfulness, adaptability, and sensitivity
Client Feedback Report
Page 5 of CPI Form 434 Narrative Report
Abbreviations on scale match abbreviations on table of descriptions on handout.
Graph gives a sample of what one’s score for Folk Scales would look like.
The higher the score the more like the trait one is.
First two vectors used for placement
Participating/ Private (v.1)- horizontal line
Approving/ Questioning (v. 2)- vertical line
Third vector provides meaning to placement
Fulfillment (v.3)