Examination of differential validity of MMPI-2-RF scores by gender and ethnicity in predicting future suicidal and violent behaviors in a forensic sample.

Abstract

Given the diversity of individuals who undergo psychological assessment, examining whether cultural bias exists in psychological assessment instruments (i.e., differential validity) is crucial. This issue occurs when a measure systematically over-or underpredicts a criterion across demographic groups or is associated with the criterion unequally across the groups. We tested the differential validity of a widely used psychological test, the Minnesota Multiphasic Personality Inventory (MMPI)-2-Restructured Form (MMPI-2-RF), as a function of gender (male, female) and ethnicity (Caucasian, African American, and Hispanic/Latino American) in large samples of forensic psychiatric inpatients. Regression models were estimated in a multigroup framework. The analyses yielded negligible to small statistical evidence of differential validity in MMPI-2-RF scores predicting the number of future suicidal behaviors and violent behaviors in the samples. This evidence supports use of the MMPI-2-RF as a generally unbiased instrument for predicting key criteria across genders and ethnicities in a forensic psychiatric population. Public Significance StatementThis study supports use of the MMPI-2-RF in forensic settings as a generally unbiased instrument for predicting suicidal and violent behaviors across genders (male, female) and ethnicities (Caucasian, African American, and Hispanic/Latino American).

MMPI-3 scale predictors of reactions to ostracism in a sample of racially diverse college students.

“…Another alternative is that MMPI-3 scales may not measure the same constructs across people who differ on racial group membership. Although research on prior MMPI iterations suggests minimal evidence for race-based test bias (e.g., Whitman et al, 2019), measurement invariance has rarely been considered for past MMPI instruments (see Wang et al, 2021, for a notable exception), and such data are not yet available for the MMPI-3. Given emerging evidence that other measures of pathological personality traits are not equivalent across White and Black Americans (e.g., Bagby et al, 2021), tests of measurement invariance across demographic and other groups should be the focus of future assessment research (see Han et al, 2019, for a methodological review).…”

“…As the MMPI-3 is newly published, a limited literature exists regarding its performance and associations in racially diverse populations beyond those reported in the test manuals. Research on the previous version of the instrument, the MMPI-2-Restructured Form (MMPI-2-RF; Ben-Porath & Tellegen, 2011), focused on differential predictive validity and found minimal evidence of test bias in specific samples (e.g., bariatric surgery candidates, Marek et al, 2015; forensic psychiatric inpatients, Whitman et al, 2019), and research on the MMPI-2 and prior test iterations likewise found minimal evidence for test bias across racialized groups (e.g., Arbisi et al, 2002; Castro et al, 2008; McNulty et al, 1997). Scholars have noted the extreme paucity of psychological science research with representation of POC populations (Buchanan et al, 2021; Roberts et al, 2020), and calls to consider racialized context extend to the domain of psychological assessment (e.g., Byrd et al, 2021).…”

Suicidal ideation across race in a justice‐involved sample: An item response theory approach

“…Although prior research has found group‐level differences in suicidal ideation across racial categories (Davidson & Wingate, 2011), this study provides the first evidence that item properties function differently across race in a justice‐involved sample. In research contexts, prior literature has largely focused on the ability of total and/or factor scores to predict outcomes of interest (e.g., suicide attempt) as evidence of the psychometric soundness of measures (Whitman et al, 2019). However, the lack of investigation of item properties (and the overemphasis on test properties) represents a disconnect between research settings and clinical realities, where clinicians often use item responses to inform follow‐up assessment and, potentially, intervention.…”

“…In addition to race‐related differences in undergraduate/community samples, recent work in an offender sample found the Minnesota Multiphasic Personality Inventory (MMPI)‐2‐Restructured Form (MMPI‐2‐RF; Ben‐Porath & Tellegen, 2008/2011) suicidal/death ideation scale demonstrated relatively similar predictive validity across ethnicity, although there was evidence for statistically significant, but small, differences across African American and Caucasian respondents (Whitman et al, 2019). However, this work does not provide information about whether test items demonstrate similar psychometric properties across race, which is essential because item‐level data are often informative for clinical decision‐making.…”

An uncomfortable tension: Reconciling the principles of forensic psychology and cultural competency.

“…Many of these instruments were initially normed on predominantly White male samples, and research conducted on older versions of some assessments has found statistical differences in scores between the norming sample and women, ethnically diverse populations, and individuals with lower levels of acculturation (Perlin & McClain, 2009; Tsai & Pike, 2000; Weiss & Rosenfeld, 2012). More recent research, however, suggests that the differences in scores are generally not substantial enough to be the sole cause for clinical scale elevations and thus may be appropriate for use with some diverse populations (Benuto et al, 2020; Whitman et al, 2018).…”