Construct Validity of the E-LSRP in a Correctional Sample

Abstract

The Expanded–Levenson Self-Report Psychopathy Scale (E-LSRP) was developed by Christian and Sellbom to improve on the psychometric properties of scores on the Levenson Self-Report Psychopathy Scale. The current study investigated the construct validity of scores on the E-LSRP in 393 male inmates. Results provided support for the reliability and construct validity of E-LSRP scores. Specifically, confirmatory factor analysis results demonstrated support for a three-factor model. Additionally, correlation and multiple regression results provided evidence supporting the convergent and discriminant validity of E-LSRP scores against scores on measures assessing psychopathy-related personality traits (e.g., antagonism, disinhibition) and symptoms of internalizing disorders, respectively. Overall, these findings extend those of previous research by establishing that E-LSRP scores demonstrate validity in assessing psychopathy in correctional settings and thus, may be a useful tool for the assessment of psychopathy in these settings.

Keywords

psychopathy E-LSRP Levenson Self-Report Psychopathy Scale psychopathic antisocial personality

Psychopathy is characterized by a constellation of personality traits that broadly include deficits in interpersonal and emotional functioning as well as dysfunctional behavior (Hare & Neumann, 2008; Skeem et al., 2011). Additionally, psychopathy often has long-lasting expensive and damaging effects at the individual, societal, and occupational levels (Hare & Neumann, 2008; Patrick et al., 2009; Smith & Lilienfeld, 2013). Given the associations that have been demonstrated between psychopathic personality traits and tendencies toward criminality, interpersonal harm, and workplace deviance, it is evident that assessment instruments that can accurately and reliably assess psychopathy are important.

The Levenson Self-Report Psychopathy Scale (LSRP; Levenson et al., 1995) is a 26-item Likert-type self-report instrument that was originally developed to assess psychopathy in noninstitutional samples. It initially assessed a two-factor conceptualization of psychopathy, where factor one corresponds to Cleckley (1941/1988) and Karpman’s (1948) primary psychopathy and factor two corresponds to secondary psychopathy. Although previous research has provided good evidence in support of the convergent and discriminant validity of scores on the LSRP (Sellbom et al., 2018), researchers have questioned the reliability of the LSRP’s secondary subscale scores (Brinkley et al., 2008; Christian & Sellbom, 2016; Levenson et al., 1995; Lynam et al., 1999; Sellbom, 2011). Others have questioned whether the LSRP is best represented by a two- or a three-factor structure (Brinkley et al., 2008; Salekin et al., 2014; Sellbom, 2011; Shou et al., 2017; Somma et al., 2014). Specifically, several studies have demonstrated superior fit for a three-factor 19-item LSRP structure aligning more closely with Cooke and Michie’s (2001) interpersonal, affective, and behavioral model of psychopathy, with factors now representing egocentricity, callousness, and antisociality (Brinkley et al., 2008; Salekin et al., 2014; Sellbom, 2011; Shou et al., 2017; Somma et al., 2014). However, despite the superior model fit of this three-factor structure, subsequent work investigating the psychometric properties of scores on the 19-item LSRP continued to call into question the reliability and construct validity of this measure. This is because reliability estimates for the Callous and Antisocial subscale scores were low (often around .60 in university and correctional samples; Sellbom et al., 2018), Callous subscale scores have demonstrated atheoretical relations with scores on external criterion measures in some studies (Few et al., 2013; Salekin et al., 2014), and scores on the Antisocial subscale were oversaturated with elements of distress and negative emotionality in other studies (Sellbom, 2011).

Due to these persisting psychometric limitations of scores on the 19-item LSRP, Christian and Sellbom (2016) developed the Expanded–Levenson Self-Report Psychopathy Scale (E-LSRP) in an attempt to improve on both the internal reliability estimates and construct validity of scores on this instrument. Employing a two-study design, they first utilized a university sample to expand on the LSRP’s item content. A second sample recruited from the community was used to assess the psychometric properties of the E-LSRP. Overall, Christian and Sellbom’s (2016) results indicated that they were successful in improving on both the internal reliability estimates and the construct validity of scores on the E-LSRP. Specifically, they found increased internal reliability estimates for scores on the E-LSRP when compared with scores on the 19-item LSRP. Regarding construct validity, they found that after allowing for item residuals of certain items to correlate a three-factor structure representing Egocentricity, Callous, and Antisocial factors was an optimal fit for their data and that this factor solution demonstrated superior indices of model fit when compared with the 19-item LSRP. Additionally, they found that scores on the E-LSRP subscales demonstrated expected relations with theoretically relevant external criterion measures. For example, Egocentricity scores were positively correlated with scores on a measure of narcissism, Callous scores were negatively correlated with scores on a measure of empathy, and Antisocial scores were positively correlated with scores on measures of disinhibition and negative emotionality—without demonstrating evidence of an oversaturation of negative emotionality. Thus, Christian and Sellbom’s (2016) findings provided strong evidence to suggest that the E-LSRP had improved psychometric properties when compared with the original measure.

To date, one additional study has further aimed to examine the construct validity of E-LSRP scores. Maheux-Caron et al. (2018) examined a French-adapted version of the E-LSRP in a sample of community participants in Canada. Their results showed improved internal consistency reliability estimates for the E-LSRP over the original LSRP. Furthermore, their results generally supported the construct validity of the three-factor structure of the E-LSRP. Furthermore, although Maheux-Caron et al. (2018) have presented the only work thus far that directly aimed to examine the construct validity of scores on the E-LSRP, it is also important to note that findings from Sellbom et al.’s (2019) study on the Comprehensive Assessment of Psychopathic Personality–Self-Report (CAPP-SR) have provided further supporting evidence for the improved reliability and construct validity of scores on the E-LSRP in university samples. More specifically, scores on E-LSRP subscales demonstrated good internal consistency reliability estimates and expected associations with scores on theoretically relevant CAPP-SR symptom scales. Most recently, Lee and Sellbom (2021) reported similar findings in a large university sample drawn from the same population, with good evidence for convergent and discriminant validity of E-LSRP scores against the Elemental Psychopathy Assessment–Short Form (Lynam et al., 2011)—the topic of their research project.

The findings reported in Christian and Sellbom (2016), Maheux-Caron et al. (2018), Sellbom et al. (2019), and Lee and Sellbom (2021) have provided promising evidence for the construct validity of the E-LSRP scores as a measure of psychopathy, as well as an improvement over the original LSRP. However, these findings have been limited to university and community samples. Given that individuals with high levels of psychopathic personality traits comprise a large portion of incarcerated samples (e.g., Kiehl & Hoffman, 2011), replicating these findings in a correctional sample against additional criterion measures would allow for further elaboration on the nomological network associated with E-LSRP scale scores.

Current Study

The current investigation therefore aimed to examine the reliability of E-LSRP scores, their internal structure, and their convergent and discriminant validity in a prison inmate sample. Based on the research described above, several hypotheses were generated to investigate the construct validity of scores on the E-SLRP. Specifically, we hypothesized that internal reliability estimates in this sample would demonstrate a similar pattern of results to that presented in previous research (Christian & Sellbom, 2016; Maheux-Caron et al., 2018; Sellbom, Cooke, & Shou, 2019). We hypothesized that a three-factor structure would be supported, as it has in previous research (Christian & Sellbom, 2016; Maheux-Caron et al., 2018). We also hypothesized that scores on the Egocentric, Callous, and Antisocial subscales would demonstrate a conceptually expected pattern of associations with scores on theoretically relevant external criterion measures based on the previous findings of research studies discussed earlier. More specifically, we hypothesized that Egocentricity scores would demonstrate significant and meaningful relations with meanness from the triarchic psychopathy model and grandiose narcissism. We expected that Callous would correlate most strongly with meanness from the triarchic psychopathy model, but no other criterion measures. We hypothesized that Antisocial would be meaningfully associated with disinhibition from the triarchic psychopathy model, criminal versatility, hypersensitive narcissism, anger, and general dysphoria. The latter hypothesis was included as the Antisocial scale continues to some degree to be associated with general demoralization and unhappiness (e.g., Christian & Sellbom, 2016; Maheux-Caron et al., 2018). Lilienfeld (1994) also postulated based on a review of the literature that psychopathy would be associated with general dysphoria rather than specific emotions of anxiety or fear. Last, in terms of discriminant validity, we hypothesized that none of the E-LSRP scores would be meaningfully correlated with boldness from the triarchic psychopathy model (Christian & Sellbom, 2016), as none of the LRSP versions have been meaningfully associated with such psychopathy features in the past (e.g., Sellbom et al., 2018). We also hypothesized that E-LSRP scores would demonstrate smaller associations with scores on scales assessing common symptoms of internalizing disorders (with the exception of dysphoria and anger-related problems as just mentioned), as part of the expansion effort was to reduce saturation with negative affectivity (Christian & Sellbom, 2016).

Method

Participants and Procedure

The current study’s sample consisted of 533 adult male incarcerated offenders from a midwestern intake correctional facility.¹ To account for missing data, cases where individuals did not answer 10% or more of a measure’s item content were excluded from analyses. Additionally, while content scales from the Minnesota Multiphasic Personality Inventory–2–Restructured Form (MMPI-2-RF; Ben-Porath & Tellegen, 2008) were not utilized for our investigation into the E-LSRP’s construct validity, we did use its validity scales to identify and exclude MMPI-2-RF profiles that were invalid (as outlined by Ben-Porath & Tellegen, 2008)² as this would have implications for all self-report inventories administered. These exclusions resulted in a final sample of 393 participants. All of these remaining 393 participants completed the E-LSRP and responded to at least 90% of its items. However, given the realities of data collected from a naturalistic setting, the current sample’s n varied throughout the data analysis process, as some participants failed to complete all measures in the study or left more than 10% of an instrument’s items unanswered. Thus, n’s deviating from the total study valid cases sample of 393 are reported where necessary (see Table 1). Within our final sample, the majority of participants identified their race/ethnicity as White (55%), although there was a significant portion of Black participants (30%). However, 2% of participants identified their race/ethnicity as Hispanic, 6% identified with a Mixed Race, and 7% identified with an “Other” race. The mean age of participants was 33.41 years (SD =10.24, Range = 18-98). Last, the proportional breakdown for participant’s level of education attained in the current sample was as follows: 37.4% equivalent to a high school diploma or GED, 29.1% some college, 22.1% no high school diploma or equivalent, 7.10% unknown, and 4.30% bachelor’s degree or higher level education. This study was approved by both university and the prison institutional review boards.

Table 1.

Descriptive Statistics and Internal Consistency Reliability for All Study Scale Scores.

Measure	Range		M	SD	n	ω
Measure	Minimum	Maximum	M	SD	n	ω
E-LSRP Total	1.08	3.40	2.26	0.44	393	.88
E-LSRP Egocentric	1.00	3.91	2.09	0.63	393	.83
E-LSRP Callous	1.00	3.67	2.04	0.50	393	.75
E-LRSP Antisocial	1.00	3.77	2.35	0.54	393	.80
TriPM Total	83.00	205.00	140.89	18.87	368	.85
TriPM Boldness	29.00	75.00	53.65	7.88	368	.75
TriPM Meanness	19.00	65.00	36.32	9.44	368	.86
TriPM Disinhibition	20.00	76.00	50.92	11.11	368	.84
ABQ	16.00	46.00	33.30	7.23	391	.88
NGS	1.00	5.00	2.77	0.81	343	.97
HSNS	1.00	4.80	2.69	0.64	343	.72
IDAS-II Dysphoria	10.00	42.00	21.55	6.93	380	.86
IDAS-II Well-Being	8.00	40.00	23.60	6.72	380	.84
IDAS-II Insomnia	6.00	30.00	15.00	5.86	380	.87
IDAS-II Lassitude	5.00	29.00	11.47	4.27	380	.80
IDAS-II Panic	7.00	32.00	11.54	4.16	380	.83
IDAS-II Ill Temper	4.00	25.00	9.75	4.45	380	.74
IDAS-II Suicidality	6.00	30.00	8.18	3.66	380	.87
IDAS-II Traumatic Intrusions	3.00	20.00	8.13	3.89	380	.86
IDAS-II Traumatic Avoidance	2.00	20.00	10.59	4.16	380	.85
IDAS-II Appetite Loss	2.00	14.00	5.53	2.17	380	.59
IDAS-II Appetite Gain	3.00	15.00	6.74	2.50	380	.64
IDAS-II Ordering	2.00	44.00	11.72	4.86	380	.73
IDAS-II Mania	3.00	25.00	10.64	4.46	380	.86
IDAS-II Checking	3.00	15.00	6.83	3.02	380	.84
IDAS-II Euphoria	2.00	22.00	10.69	4.26	380	.80

Note. n = sample size; ω = McDonald’s omega internal reliability estimate; E-LSRP = Expanded–Levenson Self-Report Psychopathy Scale; TriPM = Triarchic Psychopathy Measure; ABQ = Antisocial Behavior Questionnaire; NGS = Narcissistic Grandiosity Scale; HSNS = Hypersensitive Narcissism Scale; IDAS-II = Inventory of Depression and Anxiety Symptoms–Second Edition.

Measures

The following are descriptions of measures used in the current study. Descriptive statistics and McDonald’s omega internal consistency reliability estimates for all scale scores in this study are provided in Table 1. A matrix displaying correlations between all variables included in this study is available in the online supplemental materials (Table S1).

Expanded–Levenson Self-Report Psychopathy Questionnaire (Christian & Sellbom, 2016; Levenson et al., 1995)

The E-LSRP is a 36-item Likert-type self-report measure assessing an individual’s level of psychopathy. The E-LSRP includes a Total score and three subscales assessing interpersonal (Egocentric), affective (Callous), and behavioral (Antisocial) components of psychopathy. Item responses were averaged for each scale. Evidence supporting the reliability and construct validity of the E-LSRP’s scores was described in the introduction.

Triarchic Psychopathy Measure (TriPM; Patrick, 2010)

The TriPM is a 58-item Likert-type self-report inventory assessing an individual’s levels of psychopathic personality traits according to the Triarchic Model of Psychopathy (Patrick et al., 2009). The TriPM consists of three scales assessing individuals’ levels of meanness, boldness, and disinhibition, as well as a total score. Item responses were summed for each scale. Previous research has provided support for the construct validity of scores on the TriPM by demonstrating associations between TriPM scores and scores on other measures of psychopathic personality traits in a variety of samples (see Sellbom et al., 2018, for a review).

Antisocial Behavior Questionnaire (ABQ; Sellbom & Verona, 2004; Sellbom et al., 2012)

The ABQ is a 16-item self-report checklist that assesses engagement in various antisocial behaviors. Scores on the ABQ are summed to obtain a total score, where higher scores on the ABQ reflect greater engagement in antisocial behaviors. Scores on the ABQ have been associated with scores on other measures assessing a variety of antisocial behaviors (i.e., substance use, history of criminal behaviors, etc.; Kastner & Sellbom, 2012; Wall et al., 2013).

Narcissistic Grandiosity Scale (NGS; Rosenthal et al., 2007)

The NGS is a 16-item Likert-type self-report measure used to assess grandiose narcissism. Scores on the NGS can be summed to obtain a total score, where higher scores are indicative of higher levels of grandiose narcissistic personality traits. Scores on the NGS have demonstrated evidence of construct validity by demonstrating associations with scores on other measures of narcissism (Miller et al., 2013).

Hypersensitive Narcissism Scale (HSNS; Hendin & Cheek, 1997)

The HSNS is a 20-item Likert-type self-report measure that assesses core aspects of vulnerable narcissism (i.e., hypersensitivity, anxious self-preoccupation, emotional distress, etc.). Scores on the HSNS are summed to obtain a total score, where higher scores on the HSNS reflect greater levels of vulnerable narcissistic personality traits. Scores on the HSNS have demonstrated evidence of construct validity through associations with scores on other measures of narcissism, as well as scores on measures of theoretically relevant personality traits (i.e., hypersensitivity, anxious self-preoccupation, emotional distress, etc.; Arble, 2008).

Inventory of Depression and Anxiety Symptoms–Second Version (IDAS-II; Watson et al., 2012)

The IDAS-II is a 99-item Likert-type self-report inventory that assess a variety of internalizing symptomology. While the IDAS-II is composed of 19 subscales, only a subset of 81 of the measure’s items were administered in the current study. As such, scores on the following 15 subscales were examined in this study: dysphoria, well-being, insomnia, lassitude, panic, ill-temper, suicidality, traumatic intrusions, traumatic avoidance, appetite loss, appetite gain, ordering, mania, checking, and euphoria. Item responses were summed for each scale. Scores on the IDAS-II have demonstrated evidence of construct validity through their associations with scores on other measures assessing a variety of symptoms related to depression and anxiety (Stasik-O’Brien et al., 2018; Watson et al., 2007).

Data Analyses

To examine the reliability of scores on the E-LSRP, McDonald’s omega estimates were calculated for E-LSRP Total, Egocentric, Callous, and Antisocial scores. To evaluate the E-LSRP’s correlated three-factor structure, a CFA with mean and variance adjusted weighted least squares (WLSMV) estimation was used, as the item distributions were ordered categorical (Brown, 2015; Muthén & Muthén, 2010). All latent factor variances were fixed to 1.00 for scale setting purposes. There was no missing data pattern discernable (in light of extremely low rates of actually missingness), and therefore, mean-imputation was implemented for five items for which 1 to 2 participants did not respond. Global model fit was evaluated using standard fit indices. Specifically, root mean square error of approximation (RMSEA) values less than or equal to.06 indicated good model fit, RMSEA values between .061 and .080 indicated acceptable model fit, and RMSEA values greater than .080 indicated poor model fit (Browne & Cudeck, 1992; Browne & Cudeck, 1993; Hu & Bentler, 1999; Marsh et al., 2004). Additionally, standardized root mean square residual (SRMR) values of .08 or lower indicated good model fit (Hu & Bentler, 1999). Furthermore, given the sample size of the current data set, Chi-square tests of significance are reported but not interpreted, as they were likely not good indicators of model fit (Brown, 2015). Comparative fit index and Tucker–Lewis index values were not reported, as a calculation of the RMSEA of the null model demonstrated that this value was less than .158, meaning that these values would mathematically not be able to reach an adequate level of fit even with an RMSEA value that demonstrated acceptable fit (Kenny, 2014). We also compared the three-factor model with a one- and two-factor model to determine if a more parsimomious alternative should be considered. In addition, if model fit failed to meet a priori thresholds, we considered the application of conceptually defensible correlated residuals and respecified the model accordingly (see e.g., Sellbom & Tellegen, 2019). All nested models were compared using the DIFFTEST function in Mplus. Local model fit was evaluated by examining both the interfactor correlations and the magnitude of the item loadings onto their specified factors. Item loadings of .30 or higher were considered meaningful (Christian & Sellbom, 2016). Finally, all CFA analyses were conducted using Mplus version 7.31 (Muthén & Muthén, 1998-2015).

A series of Pearson’s r correlations and multiple regression analyses were conducted to examine the associations between scores on the E-LSRP and scores on external criterion measures. Multiple regression analyses were only conducted if a moderate to large significant relation was demonstrated between E-LSRP subscale scores and scores on external criterion measures. Due to the large number of analyses conducted, a Bonferroni correction was applied to both the correlation and multiple regression analyses. This resulted in a threshold of p < .002 (.05/21 criterion measures) for statistical significance in correlation coefficient interpretations and a threshold of p < .007 (.05/7 criterion measures) for statistical significance in multiple regression interpretations. Additionally, in an effort to account for shared method variance between scores across self-report measures (Campbell & Fiske, 1959), only statistically significant and meaningful results (i.e., results of a moderate effect size: r ≥ .30; R² ≥ .13; Cohen, 1988) were interpreted. Furthermore, comparisons of the E-LSRP subscale scores’ magnitudes of prediction in the multiple regression analyses were evaluated first by comparing the standardized regression coefficients (β) for scores on each subscale in each model. Within these models, higher standardized regression coefficients equated to stronger prediction magnitudes. Second, to determine whether differences in magnitude between standardized regression coefficients were significant, comparisons between hypothesized regression coefficients were conducted using the “linearHypothesis” function from the “car” package in RStudio (RStudio Team, 2019). This function calculates a Wald Test (i.e., F-statistic and p value) reflecting a statistical comparison of a regression coefficient to a specified value. In this case, we compared coefficients of two predictors from the regression models, using one of the observed coefficients as the specified value and the second as the coefficient to be contrasted. Due to the large number of significance tests conducted, a Bonferroni correction was calculated and applied to interpretations of significance. This resulted in a significance value of p < .004 (.05/14 criterion measures) for significance in differences between regression coefficients. All correlation and multiple regression analyses were conducted using RStudio (RStudio Team, 2019).

Results

Internal Consistency

Results from coefficient alpha internal consistency reliability estimates are presented in Table 1. Overall, results from these analyses supported our hypothesis that internal consistency reliability estimates in this sample would demonstrate a similar pattern of results to those presented in previous research (Christian & Sellbom, 2016; Maheux-Caron et al., 2018; Sellbom et al., 2019). Specifically, as seen in Table 1, McDonald’s omega internal consistency reliability estimates were highest for E-LSRP Total and Egocentric scores (.88 and .83, respectively) and were lowest for Callous and Antisocial scores (.75 and .80, respectively).

Factor Structure

Results from the confirmatory factor analyses are presented in Tables 2 and 3. We initially tested three models: a one-factor model, a two-factor model (based on the original LSRP “primary” and “secondary” scale configuration with Egocentricity and Callous items on the same factor), and the hypothesized three-factor model. Table 2 shows the model fit statistics for these three models. As indicated by the global model fit indices, the one-factor model did not meet benchmarks for adequate model fit. The two-factor model was marginal at best (based on RMSEA), with the three-factor model (albeit also marginal) demonstrating the best fit; chi square difference tests supported the three-factor model. Furthermore, given the marginal global fit of the three-factor model, modification indices were consulted to determine whether correlations between item residuals would improve model fit, as they did for the CFA analyses conducted by both Christian and Sellbom (2016) and Maheux-Caron et al. (2018). Item residuals were allowed to correlate according to two criteria: (1) if they belonged to the same factor and (2) if they were conceptually related beyond that explained by their shared latent factor (as set forth by Christian & Sellbom, 2016; Maheaux-Caron et al., 2018). According to the latter criteria, the residuals for item Pairs 7 and 6 and Pairs 5 and 8 on the Egocentric factor were allowed to correlate. Additionally, the residuals for item Pairs 16 and 12, Pairs 17 and 15, Pairs 19 and 18, and Pairs 21 and 16 on the Callous factor were allowed to correlate. Finally, the residuals for item Pairs 28 and 27, Pairs 33 and 32, Pairs 35 and 30, Pairs 32 and 30, Pairs 34 and 32, and Pairs 24 and 26 on the Antisocial factor were allowed to correlate. These modifications resulted in a modified three-factor model (see Model 3a in Table 2), which just barely reached threshold for adequate global model fit statistics. Although this modified model was a statistically significant improvement on the standard three-factor model, because of the large number of additional parameters estimated (which could result in overfitting), and no substantive conclusions were ultimately changed, we retained the standard three-factor model (without modifications) for further evaluation.

Table 2.

Model Fit Statistics for E-LSRP One- Two- and Three-Factor Models.

Models	χ²	RMSEA [90% CI]	SRMR	DIFFTEST
Model 1	2369.09^***	.087 [.084, .091]	.106	—
Model 2	1649.54^***	.067 [.063, .071]	.090	180.19^***
Model 3	1471.80^***	.062 [.058, .066]	.084	85.10^***
Model 3a	1333.79^***	.058 [.054, .062]	.080	234.00^***

Note. Model 3a is compared to Model 3. E-LSRP = Expanded–Levenson Self-Report Psychopathy Scale; Model 1 = one-factor model, Model 2 = two-factor model; Model 3 = three-factor model, Model 3a = three-factor model with post hoc modifications; RMSEA = root mean square error of approximation; CI = confidence interval; SRMR = standardized root mean square residual; DIFFTEST = χ² difference test for nested models.

***

p < .001.

Table 3.

Factor Loadings From Three-Factor Model.

Factors	λ	SE
Egocentric
Success is based on survival of the fittest; I am not concerned about the losers.	.51	.05
People who are stupid enough to get ripped off usually deserve it.	.61	.04
I tell other people what they want to hear so that they will do what I want them to do.	.65	.03
I often admire a really clever scam.	.72	.03
In today’s world, I feel justified in doing anything I can get away with to succeed.	.77	.03
Making a lot of money is my most important goal.	.59	.04
My main purpose in life is getting as many goodies as I can.	.69	.03
For me, what’s right is whatever I can get away with.	.76	.03
I enjoy manipulating other people’s feelings.	.69	.03
I let others worry about higher values; my main concern is with the bottom line.	.52	.04
Looking out for myself is my top priority.	.48	.04
Callous
I make a point of trying not to hurt others in pursuit of my goals.	.49	.05
Cheating is not justifiable because it is unfair to others.	.65	.04
Even if I were trying very hard to sell something, I wouldn’t lie about it.	.58	.04
I feel bad if my words or actions cause someone else to feel emotional pain.	.69	.04
I tend not to think about other people’s feelings.	.52	.05
People are too emotional at funerals.	.51	.05
When people are sad around me, I feel sad myself.	.33	.05
I’m not a very emotional person.	.39	.05
I tend to cry in sad movies.	.25	.05
I feel bad when I do something wrong.	.69	.04
My friends consider me a warm person.	.35	.05
I would be upset if my success came at someone else’s expense.	.61	.04
Antisocial
I have been in a lot of shouting matches with other people.	.49	.04
I find myself in the same kinds of trouble, time after time.	.66	.04
When I get frustrated, I often “let off steam” by blowing my top.	.47	.05
I am often bored.	.51	.04
I quickly lose interest in tasks I start.	.47	.04
I am a bit of a rebel.	.47	.05
I like planning things out.	.34	.05
I usually can’t keep out of trouble for too long.	.70	.03
Before I do anything, I carefully consider the possible consequences.	.50	.04
I know rules are there, but I don’t tend to follow them.	.75	.03
I often do things before thinking them through.	.64	.03
I don’t plan anything very far in advance.	.47	.04
Getting into trouble doesn’t bother me.	.55	.05

Note. Parameters calculated using weighted least squared means and variance adjusted conﬁrmatory factor analysis. All factor loadings were significant p < .001. E-LSRP = Expanded–Levenson Self-Report Psychopathy Scale.; λ = standardized factor loading; SE = standard drror.

In terms of local fit, correlations between the latent factors were as follows: Callous with Egocentric: r = .73, Antisocial with Egocentric: r = .58, Antisocial with Callous r = .34. Additionally, as seen in Table 3, with the exception of one item (Item 20), all Egocentric, Callous, and Antisocial items loaded meaningfully onto their respective factor, providing good evidence of local model fit.

To examine the possibility of method artifacts (e.g., reverse-worded statements) contributing to marginally fitting models, especially in light of larger than previously reported latent factor correlations, we specified a post hoc model (WLSMV estimation) in which the three-factor model was specified along with orthogonal (residual) group factors reflecting wording direction (i.e., standard vs. reverse-coded) of E-LSRP items. However, E-LSRP items in this model (RMSEA = .068 [.063, .072]) demonstrated no coherent pattern of factor loadings that would have provided support for the presence of a method-variance factor(s) (Johnson et al., 2011). Moreover, a post hoc bifactor model (RMSEA = .061 [.057, .066]) that estimated (using WLSMV) one general factor and three group factors reflecting the E-LSRP factors did not serve as an adequate explanation either, as a calculation of the Omega Hierarchical (OmegaH) value for this model’s general factor (.76) did not meet the recommended cut-off value of .80 or higher that would have supported the presence of an overarching first factor (Rodriguez et al., 2016a, 2016b). As such, we retained the standard three-factor model as the most parsimonious alternative, as its parameters were generally consistent with hypotheses.

Convergent Validity

Results from Pearson’s r correlations and multiple regression analyses between scores on the E-LSRP and scores on external criterion measures are presented in Tables 4, 5, and 6. Results for Total E-LSRP scale scores generally supported our hypothesis that Total E-LSRP scores would demonstrate significant and meaningful positive associations with theoretically relevant external criterion measures. Specifically, as seen in Table 4, there were significant moderate to large positive associations demonstrated between Total E-LSRP scores and meanness, disinhibition, vulnerable narcissism, and anger related to negative emotionality. However, unexpectedly, there were no significant or meaningful associations demonstrated between E-LSRP Total scores and scores on a measure of grandiose narcissism or dysphoria.

Table 4.

Pearson’s r Correlations Between E-LSRP Scale Scores and External Criterion Scale Scores.

External criterion scales	Hypotheses	Total	E	C	A
TriPM Boldness	Discriminant	.03	.09	.05	−.09
TriPM Meanness	(+)Total, (+)E, & (+)C	.65*	.55 ^*	.55*	.44*
TriPM Disinhibition	(+)Total & (+)A	.44*	.31 ^*	.17	.56*
TriPM Total	(+)Total, (+)E, (+)C, & (+)A	.60*	.50 ^*	.40*	.51*
ABQ Total	(+)Total & (+)A	.40*	.29^*	.20*	.46*
NGS Total	(+)Total & (+)E	.16	.30*	.14	−.01
HSNS Total	(+)Total & (+)A	.35*	.23*	.12	.40*
IDAS-II Dysphoria	(+)Total & (+)A	.19^*	.08	.01	.36*
IDAS-II Wellbeing	Discriminant	−.11	.02	−.10	−.20^*
IDAS-II Insomnia	Discriminant	.21^*	.13	.06	.29^*
IDAS-II Lassitude	Discriminant	.16	.09	.06	.23^*
IDAS-II Panic	Discriminant	.25^*	.17	.14	.30 ^*
IDAS-II Ill Temper	(+)Total & (+)A	.38 ^*	.26^*	.20^*	.43 ^*
IDAS-II Suicidality	Discriminant	.17	.11	.07	.23^*
IDAS-II Traumatic Intrusions	Discriminant	.16	.12	−.02	.28^*
IDAS-II Traumatic Avoidance	Discriminant	.10	.09	−.03	.16
IDAS-II Appetite Loss	Discriminant	.18^*	.16	.11	.17
IDAS-II Appetite Gain	Discriminant	.11	.12	.01	.13
IDAS-II Ordering	Discriminant	.13	.12	.03	.16
IDAS-II Mania	Discriminant	.24*	.17	.08	.32 ^*
IDAS-II Checking	Discriminant	.16	.15	.02	.19^*
IDAS-II Euphoria	Discriminant	.19*	.20*	.10	.15

Note. n (TriPM) = 368; n (ABQ) = 391; n (NaGS) = 343; n (HSNS) = 343; n (IDAS-II) = 380. Bold value indicates moderate to large effect size. Bonferroni Correction applied: .05/21 = .002. E-LSRP = Expanded–Levenson Self-Report Psychopathy Scale; E = E-LSRP Egocentric Scale; C = E-LSRP Callous Scale; A = E-LSRP Antisocial Scale; (+) = hypothesized positive association; TriPM = Triarchic Psychopathy Measure; ABQ = Antisocial Behavior Questionnaire; NGS = Narcissistic Grandiosity Scale; HSNS = Hypersensitive Narcissism Scale; IDAS-II = Inventory of Depression and Anxiety Symptoms–Second Edition.

p < .002.

Table 5.

Multiple Regression Analyses for E-LSRP Scales Predicting Hypothesized External Criterion Measures.

Measure	E-LSRP scales	F (df)	R ²	B	SE	t	p	β
NGS		37.01 (3, 387)*	.22
	Egocentric			0.40	.09	4.41^*	<.001	.31
	Callous			0.01	.11	0.12	.902	.01
	Antisocial			−0.25	.09	2.77^*	.006	−.17
TriPM Meanness		93.77 (3, 364)*	.44
	Egocentric			3.60	.77	4.66^*	<.001	.24
	Callous			6.71	.89	7.51^*	<.001	.36
	Antisocial			4.26	.79	5.40^*	<.001	.24
TriPM Disinhibition		54.88 (3, 364)*	.31
	Egocentric			0.97	1.00	0.96	.337	.06
	Callous			0.19	1.16	0.16	.873	.01
	Antisocial			10.98	1.03	10.71^*	<.001	.53
ABQ		37.01 (3, 387)*	.22
	Egocentric			0.90	0.68	1.32	.186	.08
	Callous			0.73	0.78	0.94	.349	.05
	Antisocial			5.55	0.68	8.20*	<.001	.41
IDAS-II Dysphoria		21.16 (3, 376)*	.14
	Egocentric			−1.04	.69	−1.52	.130	–.09
	Callous			−0.66	.80	−0.83	.406	–.05
	Antisocial			5.41	.70	7.77^*	<.001	.42
IDAS-II Ill Temper		30.69 (3, 376)*	.20
	Egocentric			0.22	.43	0.51	.613	.03
	Callous			0.73	.50	1.48	.139	.08
	Antisocial			3.28	.43	7.58^*	<.001	.40
HSNS		24.08 (3, 339)*	.18
	Egocentric			0.17	.07	2.63	.009	.17
	Callous			−0.09	.08	−1.15	.251	−.07
	Antisocial			0.41	.07	5.92^*	<.001	.33

Note. n (TriPM) = 368; n (ABQ) = 391; n (HSNS) = 343; n (IDAS-II) = 380. Bolded values indicate the E-LSRP scale and β with the highest magnitude of association with the external criterion measure. Bonferroni Correction applied: .05/7 = .007. E-LSRP = Expanded–Levenson Self-Report Psychopathy Scale; TriPM = Triarchic Psychopathy Measure; ABQ = Antisocial Behavior Questionnaire; NGS = Narcissistic Grandiosity Scale; HSNS = Hypersensitive Narcissism Scale; IDAS-II = Inventory of Depression and Anxiety Symptoms–Second Edition.

p < .007.

Results supported our hypothesis that E-LSRP Egocentricity scale scores would demonstrate a meaningful positive association with scores on theoretically relevant external criterion measures and that Egocentric scores would predict scores on these measures with a higher magnitude than Callous or Antisocial subscale scores. Specifically, as seen in Table 4, there was a significant moderate to large positive associations demonstrated between E-LSRP Egocentric subscale scores and scores on both meanness and grandiose narcissism. Furthermore, results from a multiple regression analysis demonstrated that scores on the Egocentric subscale significantly predicted scores on grandiose narcissism with a higher magnitude than scores on either the Callous or Antisocial subscales; it also incremented Callous in the predictor of meanness scores, though Callous was a significantly better predictor (see Tables 5 and 6). In addition to our hypothesized relations, scores on the Egocentric subscale also demonstrated a significant moderate positive association with scores on a measure of disinhibition (see Table 4). However, results of multiple regression analyses suggested that Antisocial scores, not Egocentric scores, were the strongest predictor of disinhibition scores, as hypothesized (see Tables 5 and 6).

Table 6.

Comparisons Between Regression Coefficients.

Criterion measure	E-LSRP regression coefficients under comparison	F (df)	p
NGS
	Egocentric vs. Antisocial	44.00 (1)	.004
	Egocentric vs. Callous	45.49 (1)	.004
TriPM Meanness
	Callous vs. Egocentric	109.90 (1)	<.001
	Callous vs. Antisocial	69.23 (1)	<.001
TriPM Disinhibition
	Antisocial vs. Egocentric	49.25 (1)	<.001
	Antisocial vs. Callous	19.66 (1)	<.001
ABQ
	Antisocial vs. Egocentric	33.91 (1)	<.001
	Antisocial vs. Callous	16.93 (1)	<.001
IDAS-II Dysphoria
	Antisocial vs. Egocentric	16.88 (1)	.003
	Antisocial vs. Callous	15.55 (1)	.002
IDAS-II Ill Temper
	Antisocial vs. Egocentric	22.22 (1)	<.001
	Antisocial vs. Callous	19.41 (1)	<.001
HSNS
	Antisocial vs. Egocentric	38.17 (1)	<.001
	Antisocial vs. Callous	11.71 (1)	.002

Note. n (TriPM) = 368; n (ABQ) = 391; n (HSNS) = 343; n (IDAS-II) = 380. Bonferroni Correction applied: α = .05/14 = .004. E-LSRP = Expanded–Levenson Self-Report Psychopathy Scale; TriPM = Triarchic Psychopathy Measure; ABQ = Antisocial Behavior Questionnaire; NGS = Narcissistic Grandiosity Scale; HSNS = Hypersensitive Narcissism Scale; IDAS-II = Inventory of Depression and Anxiety Symptoms–Second Edition.

Overall, results for E-LSRP Callous subscale scores supported our hypothesis. Specifically, these results demonstrated a large positive correlation between E-LSRP Callous scale scores and meanness from the triarchic psychopathy model. Furthermore, results from a multiple regression analysis demonstrated that scores on the Callous scale significantly predicted scores on this scale with a higher magnitude than scores on the Egocentric or Antisocial scales (see Tables 5 and 6).

Furthermore, results for E-LSRP Antisocial scale scores also supported our hypotheses. Specifically, as seen in Table 4, there were significant moderate to large associations demonstrated between E-LSRP Antisocial scale scores and scores on measures assessing a history of antisocial behaviors, emotional distress, anger related to negative emotionality, vulnerable narcissism, and disinhibition. Furthermore, results from multiple regression analyses demonstrated that the Antisocial subscale significantly predicted scores on these scales with a higher magnitude than scores on the Egocentric or Callous scales (see Tables 5 and 6). In addition to our hypothesized relations, scores on the Antisocial scale demonstrated a significant moderate positive association with scores on meanness. However, results of the multiple regression analysis indicated that Egocentricity and Callous scores were the strongest predictors of meanness scores, as hypothesized (see Tables 5 and 6).

Discriminant Validity

The results regarding discriminant validity were generally supported for E-LSRP Total, Egocentric, Callous, and Antisocial scores. Specifically, as seen in Table 4, E-LSRP Total, Egocentric, and Callous subscale scores did not demonstrate significant or meaningful relations with scores on boldness from the triarchic psychopathy model or measures assessing common symptoms of internalizing disorders. There were also no significant and meaningful associations demonstrated between internalizing scale scores and Antisocial scores, with two exceptions. Specifically, as seen in Table 4, Antisocial scale scores did demonstrate unexpected moderate positive associations with scores on measures of mania and panic symptoms.

Discussion

The goal of the current investigation was to further build on the E-LSRP construct validity literature (Christian & Sellbom, 2016; Lee & Sellbom, 2021; Maheux-Caron et al., 2018; Sellbom et al., 2019) by replicating and extending reliability and construct validity research to a correctional setting in which it was likely that there would be more sample variability at the higher end of psychopathic personality trait dimensions. Overall, our findings were generally supportive of our hypotheses. We observed acceptable reliability estimates for E-LSRP scale scores, the internal structure was generally upheld, and convergent and discriminant validity findings were similar to and expanded upon existing nomological networks. These findings are discussed in greater detail in the subsequent sections.

Internal Psychometric Properties

In support of this conclusion, results from our examination of the reliability of E-LSRP scores aligned with our hypothesis that reliability estimates in the current study’s sample would demonstrate a pattern of findings similar to those presented previously (Christian & Sellbom, 2016; Lee & Sellbom, 2021; Maheux-Caron et al., 2018; Sellbom et al., 2019). Specifically, reliability estimates in this sample were highest for E-LSRP Total and Egocentric scores and were lowest for Antisocial and Callous scores. As such, these findings suggest that reliability estimates demonstrated by these scores appear to be relatively stable across a variety of samples. Moreover, results from the current study also supported the conclusions drawn by previous research (Christian & Sellbom, 2016; Maheux-Caron et al., 2018) that reliability estimates for scores on the E-LSRP’s Antisocial and Callous subscales were an improvement upon those demonstrated by these subscales on the 19-item LSRP. Specifically, reliability estimates for scores on the Callous and Antisocial subscales in this sample were .75 and .80, respectively, which are higher than those previously described for scores demonstrated by the 19-item LSRP in correctional settings (see Sellbom et al., 2018, for a review). As such, these results suggest that reliability estimates for E-LSRP scores appear to be an improvement on those demonstrated by 19-item LSRP scores across a variety of samples.

In addition to findings supporting the improved reliability of E-LSRP scores, results from our investigation into the internal structure of the E-LSRP also support the conclusion that E-LSRP scores demonstrate fidelity in assessing psychopathy in correctional settings. Much like the factor analyses presented by Christian and Sellbom (2016) and Maheux-Caron et al. (2018), global fit statistics from the first factor analysis model were marginal to adequate. The absolute fit indices diverged somewhat in that the well-established RMSEA index was acceptable for the three-factor model, whereas the SRMR value was slightly above conventional threshold for acceptable fit. Respecification attempts through examining correlated residuals significantly improved overall model fit, but SRMR values remained marginal. Because RMSEA has a more voluminous research base supporting its use relative to SRMR, and parameters were generally consistent with expectations, we tentatively accepted the three-factor model with these caveats.

One potential explanation that may help understand why our model was only able to reach an adequate level of model fit could be the large factor correlations demonstrated in this sample masking important sources of covariation, but as reported earlier, a post hoc examination of a dominant general factor or existence of method variance factors did not yield plausible explanations. The most reasonable explanation might therefore be variability. Previous studies (Christian & Sellbom, 2016; Maheux-Caron et al., 2018) examined community or university samples in which there was likely less variability in psychopathy traits. Restricted variability would attenuate correlations among latent factors relative what is observed in a sample with greater variability in psychopathy traits, such as a correctional sample.

Despite potential concerns about global model fit and factor correlations, other aspects of our internal structure results provided good support for the construct validity of the E-LSRP. Specifically, an examination of other aspects of local model fit, such as standardized item loadings, suggested that items were loading onto the factors at a level that was comparable to that seen in Christian and Sellbom’s (2016) original study.³ The AIL for items on the Egocentric scale in the current study was .63, whereas the AIL for this scale was .62 in Christian and Sellbom’s (2016) study. Additionally, the AIL for items on the Callous scale in the current study was .51, whereas the AIL for this scale was .55 in Christian and Sellbom’s (2016) study. Last, the AIL for items on the Antisocial scale in the current study was .54, whereas the AIL for this scale was .46 in Christian and Sellbom’s (2016) study. Thus, these indications of local model fit suggest that the high factor correlations demonstrated in our sample are likely not affecting the E-LSRP items’ ability to assess their respective latent constructs.

Construct Validity

In terms of convergent validity, results demonstrated by the associations between scores on each of the E-LSRP subscales and scores on external criterion measures provide convincing evidence to suggest that these subscales are accurately operationalizing the latent constructs that they are meant to assess. In line with previous research (Christian & Sellbom, 2016; Maheux-Caron et al., 2018; Lee & Sellbom, 2021; Sellbom et al., 2019), the association demonstrated between Egocentric scores and scores on a measure assessing grandiose narcissism suggest that Egocentric scores may be adequately capturing the interpersonal component of Cooke and Michie’s (2001) interpersonal, affective, and behavioral model of psychopathy. This conclusion is supported by associations with both the meanness psychopathy construct (that includes interpersonal psychopathy features; Patrick et al., 2009) and grandiose narcissism, which is consistent with Cooke and Michie’s (2001) formulation of grandiosity and superficiality that describes the subtype of narcissism highlighted by the construct of grandiose narcissism (Pincus & Lukowitsky, 2010).

Additionally, in line with previous research, the large association demonstrated between Callous scores and meanness indicates that Callous scores may be adequately capturing the affective component of Cooke and Michie’s (2001) model. Indeed, core features of meanness involves callousness, exploitativeness, unemotionality, and lack of empathy (Patrick et al., 2009), or the “deficient affective experience” posited by Cooke and Michie (2001). Furthermore, this finding is consistent with E-LSRP Callous scores being associated with CAPP-SR symptom scores of unempathic, lacks emotional depth, lacks remorse, uncaring, uncommitted, and detached (Sellbom et al., 2019) and EPA-SF trait scores of callous, coldness, manipulation, self-centeredess (Lee & Sellbom, 2021).

Further in line with previous E-LSRP research, the associations demonstrated between Antisocial scores and scores on measures assessing disinhibition and antisocial behaviors indicate that Antisocial scores may be adequately capturing the behavioral domain of Cooke and Michie’s (2001) model; the domain describing the difficulties with impulse control and reckless behaviors captured by the constructs of disinhibition and antisociality (Patrick et al., 2009). Moreover, in addition to capturing these elements of psychopathy, Antisocial scores also appear to be capturing the anger-prone and personally distressed features of psychopathy (Lilienfeld, 1994; Sellbom, 2011), as they demonstrated associations with scores on measures assessing vulnerable narcissism, distress, and anger related to negative emotionality. Furthermore, the magnitude of associations observed for Antisocial scores with scores on measures assessing vulnerable narcissism, distress, and anger related to negative emotionality were smaller than those observed between this scale and scores on measures assessing disinhibition and antisociality. These findings indicate that, whereas scores on this subscale may be capturing anger-prone and personally distressed components of psychopathy, there is not an oversaturation of these elements being assessed by these subscale scores. Thus, these findings support the Christian and Sellbom’s (2016) and Maheux-Caron et al.’s (2018) conclusions that the convergent and discriminant validity of Antisocial scale scores represent a substantial improvement over the 19-item LSRP counterpart.

Results that speak to discriminant validity of E-LSRP scale scores also provide evidence to support construct validity. Specifically, the lack of significant and meaningful associations demonstrated between E-LSRP subscale scores and scores on scales assessing a variety of internalizing symptoms generally provides good evidence for the discriminant validity of these subscale scores.

One exception to this general finding of support for discriminant validity was the significant and meaningful association demonstrated between Antisocial scores and scores on the IDAS-II Panic subscale. This finding warrants further investigation, as it is inconsistent with psychophysiology research suggesting that as an individual’s levels of disinhibition increase, their startle response habituation rate increases as well (LaRowe et al., 2006), meaning that individuals with high levels of disinhibition often experience less feelings related to panic. It is also possible that due to the high number of correlations conducted, this finding may have been an anomaly. Nonetheless, future research should continue to investigate the nature of this relation as well as its ramifications for E-LSRP Antisocial scores’ discriminant validity.

There were a few additional unexpected findings in the current study that speak to possible questions about discriminant validity. The Antisocial subscale was moderately and uniquely associated with meanness from the triarchic psychopathy model. Although this was not hypothesized, there is substantial literature to indicate that meanness correlates at a large magnitude with measures of disinhibition (i.e., Antisocial); for instance, the meta-analytic association between TriPM Meanness and Disinhibition was .53 (Sleep et al., 2019). So, this finding it not overly surprising. The moderate correlation between Egocentricity and disinhibition from the triarchic model likely has a similar explanation. The moderate correlation between Antisocial and manic symptoms maybe be due to a shared tendency toward externalizing behavior. Previous research has suggested that scores on measures assessing impulsivity have demonstrated relations with scores on measures assessing mania (Watson et al., 2012). Given that antisociality is also a construct enveloped within the broader externalizing liability, it makes conceptual sense that Antisocial scores would demonstrate some overlapping variance with mania scores.

Limitations, Conclusions, and Future Directions

There are two limitations associated with the current study that warrant discussion. First, the current study was limited by its reliance on self-report measures. As such, one potential direction for future research could be to continue to investigate the construct validity of scores on the E-LSRP using a wide variety of assessment instruments as external criterion measures (e.g., clinical interviews, cognitive and behavioral tasks, etc.). Second, the current study was also limited by its entirely male sample, and thus, do not generalize to female offenders. As such, another potential direction for future research could be to investigate the construct validity of E-LSRP scores in a female correctional sample.

Overall, results from the current study provide support for the psychometric properties of scores on the E-LSRP in a correctional sample. Specifically, results from our investigation into the reliability of E-LSRP scores in a correctional sample were comparable to those demonstrated by Christian and Sellbom (2016), Maheux-Caron et al. (2018), Lee and Sellbom (2021), and Sellbom et al. (2019) in university and community samples. Additionally, results from our investigation into the construct validity of E-LSRP scores suggested that despite only demonstrating an adequately fitting factor structure, scores on the Egocentric, Callous, and Antisocial subscales appear to be accurately operationalizing the constructs that they were intended to assess. As such, results from this study suggest that the E-LSRP may be a useful tool for the detection and assessment of psychopathy in male correctional settings, and that serves as an improvement over the original LSRP. More generally, in addition to improving upon specific aspects of the current study, future research needs to examine the applied utility of the E-LSRP in offender contexts, including the prediction of violent and other forms of reoffending, adherence to community supervision, treatment, and other issues that concerns the criminal justice system.

Supplemental Material

sj-pdf-1-asm-110.1177_10731911211038619 – Supplemental material for Construct Validity of the E-LSRP in a Correctional Sample

Supplemental material, sj-pdf-1-asm-110.1177_10731911211038619 for Construct Validity of the E-LSRP in a Correctional Sample by Martin Sellbom, Jaiden S. Butler, Tayla T. C. Lee, Andrea M. Loucaides, Tracy L. Masterson and Dustin B. Wygant in Assessment

Footnotes

Acknowledgements

The authors thank a team of research assistants (Connor Lynch, Ashley Creasap, Keefe Maccarone, Alexis Bifro, Olivia Kastelic, Megan Whitman, Jacob Jarvis, Sarah Kline, Yanu Sou, Jessica Neundorf, Elizabeth Fonte, who were under the supervision of Dr. Tracy Masterson) for their assistance in data collection. The authors also thank the Ohio Department of Rehabilitation and Corrections and the staff at Lorain Correctional Institute for allowing the research to take place in their facility. None of the opinions or conclusions expressed in this article reflects any official policy or position of the Ohio Department of Rehabilitation and Corrections.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The University of Minnesota Press provided funding for the prison data collection.

ORCID iDs

Martin Sellbom

Tayla T.C. Lee

Dustin B. Wygant

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Arble

E. P.

(2008). Evaluating the psychometric properties of the hypersensitive narcissism scale: Implications for the distinction of covert and overt narcissism [Masters Theses and Doctoral Dissertations, Eastern Michigan University, Ypsilanti, MI]. https://commons.emich.edu/theses/236/

Ben-Porath

Y. S.

Tellegen

(2008). The Minnesota Multiphasic Personality Inventory-2 Restructured Form: Manual for administration, scoring, and interpretation. University of Minnesota Press.

Brinkley

C. A.

Diamond

P. M.

Magaletta

P. R.

Heigel

C. P.

(2008). Cross-validation of levenson's psychopathy scale in a sample of federal female inmates. Assessment, 15(4), 464-482. https://doi.org/10.1177/1073191108319043

Brown

T. A.

(2015). Confirmatory factor analysis for applied research. Guilford Press.

Browne

M. W.

Cudeck

(1992). Alternative ways of assessing model fit. Sociological Methods & Research, 21(2), 230-258. https://doi.org/10.1177/0049124192021002005

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 111-135). Sage.

Campbell

D. T.

Fiske

D. W.

(1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56(2), 81-105. https://doi.org/10.1037/h0046016

Christian

Sellbom

(2016). Development and validation of an expanded version of the Three-Factor Levenson Self-Report Psychopathy Scale. Journal of Personality Assessment, 98(2), 155-168. https://doi.org/10.1080/00223891.2015.1068176

Cleckley

(1941). The mask of sanity. Mosby.

10.

Cleckley

(1988). The mask of sanity (5th ed.). Emily S. Cleckley.

11.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum.

12.

Cooke

D. J.

Michie

(2001). Refining the construct of psychopathy: Towards a hierarchical model. Psychological Assessment, 13(2), 171-188. https://doi.org/10.1037/1040-3590.13.2.171

13.

Few

L. R.

Miller

J. D.

Lynam

D. R.

(2013). An examination of the factor structure of the elemental psychopathy assessment. Personality Disorders, 4(3), 247-253. https://doi.org/10.1037/per0000016

14.

Hare

R. D.

Neumann

C. S.

(2008). Psychopathy as a clinical and empirical construct. Annual Review of Clinical Psychology, 4(1), 217-246. https://doi.org/10.1146/annurev.clinpsy.3.022806.091452

15.

Hendin

H. M.

Cheek

J. M.

(1997). Assessing hypersensitive narcissism: A reexamination of Murray's Narcism Scale. Journal of research in personality, 31(4), 588-599. https://doi.org/10.1006/jrpe.1997.2204

16.

L. T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1-55. https://doi.org/10.1080/10705519909540118

17.

Johnson

R. E.

Rosen

C. C.

Djurdjevic

(2011). Assessing the impact of common method variance on higher order multidimensional constructs. Journal of Applied Psychology, 96(4), 744. https://doi.org/10.1037/a0021504

18.

Karpman

(1948). The myth of the psychopathic personality. American Journal of Psychiatry, 104(9), 523-534. https://doi.org/10.1176/appi.ajp.104.9.523

19.

Kenny

D. A.

(2014, October 6). Measuring model fit. http://davidakenny.net/cm/fit.htm

20.

Kiehl

K. A.

Hoffman

M. B.

(2011). The criminal psychopath: History, neuroscience, treatment, and economics. Jurimetrics, 51, 355-397. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4059069/

21.

LaRowe

S. D.

Patrick

C. J.

Curtin

J. J.

Kline

J. P.

(2006). Personality correlates of startle habituation. Biological Psychology, 72(3), 257-264. https://doi.org/10.1016/j.biopsycho.2005.11.008

22.

Lee

K. Y.

Sellbom

(2021). Further validation of the elemental psychopathy assessment–Short Form (EPA-SF) in a large university sample. Jo urnal of Personality Assessment, 103(3), 289-299. https://doi.org/10.1080/00223891.2020.1779734

23.

Levenson

M. R.

Kiehl

K. A.

Fitzpatrick

C. M.

(1995). Assessing psychopathic attributes in a noninstitutionalized population. Journal of Personality and Social Psychology, 68(1), 151-158. https://doi.org/10.1037/00223514.68.1.151

24.

Lilienfeld

(1994). Conceptual problems in the assessment of psychopathy. Clinical Psychology Review, 14(1), 17-38. https://doi.org/10.1016/0272-7358(94)90046-9

25.

Lynam

D. R.

Gaughan

E. T.

Miller

J. D.

Miller

D. J.

Mullins-Sweatt

Widiger

T. A.

(2011). Assessing the basic traits associated with psychopathy: Development and validation of the Elemental Psychopathy Assessment. Psychological Assessment, 23(1), 108-124. https://doi.org/10.1037/a0021146

26.

Lynam

D. R.

Whiteside

Jones

(1999). Self-reported psychopathy: A validation study. Journal of Personality Assessment, 73(1), 110-132. https://doi.org/10.1207/S15327752JPA730108

27.

Maheux-Caron

Gamache

Sellbom

Christian

Lussier

Savard

(2018). French adaptation and validation of the expanded version of the three-factor Levenson Self-Report Psychopathy Scale. Assessment, 27(7), 1448-1462. https://doi.org/10.1177/1073191118811607

28.

Marsh

H. W.

Hau

K. T.

Wen

(2004). In search of golden rules: Comment on hypothesis-testing approaches to setting cutoff values for fit indexes and dangers in overgeneralizing Hu and Bentler's (1999) findings. Structural Equation Modeling, 11(3), 320-341. https://doi.org/10.1207/s15328007sem1103_2

29.

Miller

J. D.

Gentile

Wilson

Campbell

W. K.

(2013). Grandiose and vulnerable narcissism and the DSM-5 pathological personality trait model. Journal of Personality Assessment, 95(3), 284-290. https://doi.org/10.1080/00223891.2012.685907

30.

Muthén

L. K.

Muthén

B. O.

(1998-2015) Mplus user's guide. Muthén & Muthén.

31.

Muthén

L. K.

Muthén

B. O.

(2010). Mplus user's guide: Statistical Analysis with latent variables. Muthén & Muthén.

32.

Patrick

C. J.

(2010). Operationalizing the triarchic conceptualization of psychopathy: Preliminary description of brief scales for assessment of boldness, meanness, and disinhibition (Unpublished test manual, Florida State University, Tallahassee, FL).

33.

Patrick

C. J.

Fowles

D. C.

Krueger

R. F.

(2009). Triarchic conceptualization of psychopathy: Developmental origins of disinhibition, boldness, and meanness. Development and Psychopathology, 21(3), 913-938. https://doi.org/10.1017/S0954579409000492

34.

Pincus

A. L.

Lukowitsky

M. R.

(2010). Pathological narcissism and narcissistic personality disorder. Annual Review of Clinical Psychology, 6, 421-446. https://doi.org/10.1146/annurev.clinpsy.121208.131215

35.

Rodriguez

Reise

S. P.

Haviland

M. G.

(2016a). Applying bifactor statistical indices in the evaluation of psychological measures. Journal of Personality Assessment, 98(3), 223-237. https://doi.org/10.1080/00223891.2015.1089249

36.

Rodriguez

Reise

S. P.

Haviland

M. G.

(2016b). Evaluating bifactor models: calculating and interpreting statistical indices. Psychological Methods, 21(2), 137-150. https://doi.org/10.1037/met0000045

37.

Rosenthal

S. A.

Hooley

J. M.

Steshenko

(2007). Distinguishing grandiosity from self-esteem: Development of the Narcissistic Grandiosity Scale. Manuscript in preparation.

38.

RStudio Team. (2019). RStudio: Integrated development for R. RStudio. http://www.rstudio.com/

39.

Salekin

R. T.

Chen

D. R.

Sellbom

Lester

W. S.

MacDougall

(2014). Examining the factor structure and convergent and discriminant validity of the Levenson Self-Report Psychopathy Scale: Is the two-factor model the best fitting model? Personality Disorders, 5(3), 289-304. https://doi.org/10.1037/per0000073

40.

Sellbom

(2011). Elaborating on the construct validity of the Levenson Self-Report Psychopathy Scale in incarcerated and non-incarcerated samples. Law and Human Behavior, 35(6), 440-451. https://doi.org/10.1007/s10979-010-9249-x

41.

Sellbom

Ben-Porath

Y. S.

Patrick

C. J.

Wygant

D. B.

Gartland

D. M.

Stafford

K. P.

(2012). Development and construct validation of MMPI-2-RF indices of global psychopathy, fearless-dominance, and impulsive-antisocial. Personality Disorders, 3(1), 17-38. https://doi.org/10.1037/a0023888

42.

Sellbom

Cooke

D. J.

Shou

(2019). Development and initial validation of the Comprehensive Assessment of Psychopathic Personality–Self-Report (CAPP-SR). Psycho-logical assessment, 31(7), 878-894. https://doi.org/10.1037/pas0000714

43.

Sellbom

Lilienfeld

Fowler

McCrary

(2018). The self-report assessment of psychopathy, challenges, pitfalls, and promises. In Patrick

(Ed), The handbook of psychopathy (pp. 211-258). The Guilford Press.

44.

Sellbom

Verona

(2004). Differential neuropsychological mechanisms for two psychopathy factors. Poster presented at the 19th annual meeting for the Society for Research in Psychopathology, St. Louis, MO.

45.

Shou

Sellbom

Han

(2017). Evaluating the construct validity of the Levenson Self-Report Psychopathy Scale in China. Assessment, 24(8), 1008-1023. https://doi.org/10.1177/1073191116637421

46.

Skeem

J. L.

Polaschek

D. L.

Patrick

C. J.

Lilienfeld

S. O.

(2011). Psychopathic personality: Bridging the gap between scientific evidence and public policy. Psychological Science in the Public Interest, 12(3), 95-162. https://doi.org/10.1177/1529100611426706

47.

Sleep

C. E.

Weiss

Lynam

D. R.

Miller

J. D.

(2019). An examination of the Triarchic Model of psychopathy's nomological network: A meta-analytic review. Clinical Psychology Review, 71(July), 1-26. https://doi.org/10.1016/j.cpr.2019.04.005

48.

Smith

S. F.

Lilienfeld

S. O.

(2013). Psychopathy in the workplace: The knowns and unknowns. Aggression and Violent Behavior, 18(2), 204-218. https://doi.org/10.1016/j.avb.2012.11.007

49.

Somma

Fossati

Patrick

Maffei

Borroni

(2014). The three-factor structure of the Levenson Self-Report Psychopathy Scale: Fool's gold or true gold? A study in a sample of Italian adult non-clinical participants. Personality and Mental Health, 8(4), 337-347. https://doi.org/10.1002/pmh.1267

50.

Stasik-O’Brien

S. M.

Brock

R. L.

Chmielewski

Naragon-Gainey

Koffel

McDade-Montez

O’Hara

M. W.

Watson

(2018). Clinical utility of the Inventory of Depression and Anxiety Symptoms (IDAS). Assessment, 26(5), 944-960. https://doi.org/10.1177/1073191118790036

51.

Wall

T. D.

Sellbom

Goodwin

B. E.

(2013). Examination of intelligence as a compensatory factor in non-criminal psychopathy in a non-incarcerated sample. Journal of Psychopathology and Behavioral Assessment, 35(4), 450-459. https://doi.org/10.1007/s10862-013-9358-1

52.

Watson

O’Hara

M. W.

Naragon-Gainey

Koffel

Chmielewski

Kotov

Stasik

S. M.

Ruggero

C. J.

(2012). Development and validation of new anxiety and bipolar symptom scales for an expanded version of the IDAS (the IDAS-II). Assessment, 19(4), 399-420. https://doi.org/10.1177/1073191112449857

53.

Watson

O’Hara

M. W.

Naragon-Gainey

Koffel

Chmielewski

Kotov

Stasik

S. M.

Ruggero

C. J.

(2012). Development and validation of new anxiety and bipolar symptom scales for an expanded version of the IDAS (the IDAS-II). Assessment, 19, 399-420.

54.

Watson

O’Hara

M. W.

Simms

L. J.

Kotov

Chmielewski

McDade-Montez

E. A.

Gamez

Stuart

(2007). Development and validation of the Inventory of Depression and Anxiety Symptoms (IDAS). Psychological Assessment, 19(3), 253-268. https://doi.org/10.1037/1040-3590.19.3.253

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.07 MB