Interaction of psychiatric factors and comorbid psychopathology: does measurement method matter? (2023)

The editor's final edited version of this article is available atpsychological assessment

Participants, Settings and Procedures

The sample was drawn from an anonymized dataset of 1,534 offenders convicted or serving sentences in community-based or inpatient drug treatment programs in Oregon, Utah, Nevada, and Florida¹Participants were also recruited from a (prison) drug treatment program in Texas. There were 727 individuals from community drug rehabilitation programs (47.4% of the sample) and 807 individuals from prisons (53.6% of the sample). Overall, participants were predominantly male (83.3%) and Caucasian (65.8%). One-third (29.6%) did not complete high school, 43.2% completed high school or earned a GED, and 26.7% attended at least college. The procedure was as follows: participants were given a detailed explanation of the procedure, asked for written informed consent, and completed a series of self-report questionnaires and interviews.

Count

regression-based methods

A series of linear and logistic regressions of the PCL-R and PPI were conducted on F1, F2, and their interactions, respectively, to examine the independent and interactive effects of each factor and gender in predicting psychopathology, child maltreatment, and rearrest. Correlations between psychopathic factors and outcome variables are presented inTable 1.Covariates (Sex, Education, Race, Location), F1 and F2 in the first step, followed by the interaction F1xF2 in the second step, the three-way interaction of each factor with gender (F1xSex, F2xSex) and in the third step F1xF2xSex.

On the PCL-R, F1 and F2 had significant effects, with F1 negatively related to psychopathology and F2 positively related to psychopathology (seeTable 2). However, it is noteworthy that the PCL-R F1 showed significant correlations only with anxiety, depression, drug and alcohol abuse, and childhood abuse. The gender interactions between F1xF2 and F1xF2x were not significant, suggesting that F1 has no impact and there are no gender differences. For PPIs, F1 and F2 had main effects, with F1 negatively related to psychopathology and childhood maltreatment and F2 positively related (seeTable 2). F1xF2 interaction is significant for depression and suicidality (Second= −.09 andSecond= −.07 orPis < .01). Further investigation of the interaction revealed that F1 was protective. Furthermore, the three-way interaction of F1xF2xgender is significant for suicidality (Second= −.17, p <.01), so the interaction is significant for men but not for women.³Finally, the unnormalized beta weights and standard errors of the PCL-R and PPI interaction terms were converted to z-scores and compared using a cut-off table to determine if there were differences between measures. +/- 2.58 cutoff forPi< .01, a conservative level of significance controlling for Type I errors. Significant differences were found for depression (z = 4.29) and suicidality (z = 2.70), but not for BPD (z = −.72 ), anxiety (z = 2.29), drug use (z = .84), and alcohol use (z = .36), child abuse (z = −1.79), and rearrest (z = −.37). Although the PPI showed a stronger association with psychopathology, there was little evidence for significant differences between measures.

possible profiling⁴

One- to five-class solutions were obtained by fitting two sets of latent profile models to the PCL-R and PPI F1 and F2 scores using Mplus version 6.1 (Muthen & Muthen, 2006). Latent profile analysis was preferred over latent class analysis as it was more appropriate due to the continuity of the variables. Estimates for each model were performed using 1,500 random seed sets and 500 final-stage optimizations to ensure that the resulting estimates were based on global rather than local maximum likelihoods. Model selection was based on the interpretability of parameter estimates and the fit of the comparative data models in terms of information criteria and hypothesis testing. For each model, the Akaike information criterion (AIC), Bayesian information criterion with sample size adjustment (SS–BIC), andLilac, 1987), relative entropy and the Lo Mendell Rubin (LMR) test. are the preferred criteria for choosing the number of classes to commit to latent data structure profile models (Hansen et al., 2007;). AIC and SS-BIC are based on the log-likelihood of the model, with the penalty term accounting for model complexity. Relative entropy measures the uncertainty in classifying a subject into a latent class. Values ​​range from 0 to 1, with values ​​close to 1 indicating a reasonable level of classification accuracy between observed and predicted class participation. The LMR test compares the considered model (K) with a classless model (K-1). A non-significant value indicates that the number of classes in the model under study has no better fit than a model with one less class (). Therefore, the best-fitting model will have smaller AIC and SS-BIC values, higher entropy values, and a significant Lo Mendell Ruby test. The fit index is used in conjunction with theory as a guide to model selection, as the most appropriate model must not only be statistically significant, but also interpretable and provide a clear and plausible explanation for the data.

Tisch 3Contains general adjustment statistics for latent profile analysis. Four categories of models were identified as the best fitting models for PCL-R and PPI. In the PCL-R, four model categories stand out as the best fit on most fit indices. Although the AIC value for the five-level model is lower, all other metrics for the five-level model worsen, suggesting that the four-level model fits the data better. Furthermore, the four-level model provided the best category explanation based on psychiatric factors. The results for PPI are slightly different as the three-level model has a lower SS-BIC value while the five-level model has a lower AIC value and slightly higher entropy compared to the other class solutions. However, the LMR statistics for the three- and five-category models were not significant and uninterpretable. The other fit indices indicated that the four-category model was the best fit.

Four levels vary depending on the combination of F1 and F2 values: level 1 (low F1-low F2), level 2 (low F1-high F2), level 3 (high F1-low F2) and level (high F1-high Level ). F2). PCL-R and PPI class members are thereTable 2.When using PCL-R most fall into the low F1 - high F2 category, while when using PPI most fall into the low F1 - low F2 category. However, while the distribution of individuals between classes is different, the PCL-R and PPI form similar classes based on F1 and F2 scores. The categories of different genders are also consistent.⁵In addition, mean conditional probabilities (all >.70) indicated good agreement between each category and the category in which the individual had been placed (Table 4).

Next, a four-category PCL-R and best-fit PPI model was used to test whether F1 has a protective or enhancing effect on F2 in the context of psychopathology, child maltreatment, and arrest. A series of analyzes of covariance (ANCOVA) were performed on continuous variables (eg, BPD, anxiety, depression, drug use, alcoholism, suicide, and child abuse) and chi-square analyzes were used to examine between-class differences in categorical variables (eg ., eg gender, race, education and conception). A significance level of 0.01 was used for each ANCOVA and Chi-Square analysis to test for Type I error rates.

class difference

PCL-R

as the picture showsTable 5Participants of the four classes were compared on several demographic characteristics. Chi-square analysis revealed a significant association between class and gender.X²(3) = 82,32,Pi<.001; Rasse χ²(3) = 40,86,Pi<.001; education x²(6) = 39,56,Pi<.001; and location x²(3) = 52,09,Pi<.001. Because all demographic variables showed significant differences, they were included as covariates in subsequent analyses. A series of univariate ANCOVAs were conducted to determine whether the four categories differed on psychopathology, child maltreatment, or re-incarceration. Teaching has a significant impact on BPD [eat(3) = 29,62,Pi< .001, Cohen'sHey= .50], anxiety[eat(3) = 6,90,Pi< .01,Hey= .24] Depression [eat(3) = 5,64,Pi< .01,Hey= 0.22], drugs [eat(3) = 21,14,Pi<.001,Hey= .42], drink [eat(3) = 3,76,Pi< .05,Hey= .18], suicide [eat(3) = 6,73,Pi<.001,Hey= .24] and child abuse [eat(3) = 16,47,Pi<.001,Hey= .38]. as the picture showsFigure 1.1-1.7, both high F2 classes showed higher scores on BPD, anxiety, depression, drug and alcohol abuse, suicidality, and child abuse than the two low F2 classes. Chi-square analysis revealed significant differences between class and rear seat [x²(3) = 14,02,Pi< .01,Hey= .24] and two individuals with a high F2 score were more likely to be rearrested than two individuals with a lower F2 score.

open in a separate window

open in a separate window

Figure 1.1-1.7

Psychiatric Checklist-Revised (PCL-R) and Psychiatric Personality Inventory (PPI) including mean BPS scores, anxiety, depression, substance use, alcoholism, suicide, child abuse, and arrest rate one year after exit. CoensHeyIndividual contrasts between classes appear with a high coefficient of 2. The essentialz– Also shown are the measured contrast values ​​between classes with a high factor of 2. Abbreviations are as follows: LL, low F1-low F2; HL, high F1-low F2; left hand, low F1-high F2; HH, high F1-high F2;

Table 5

Demographic variables and group differences between PCL-R and PPI categories

PCL-R	low F1 low F2	low F1 Hoch F2	Hoch F1 low F2	Hoch F1 Hoch F2	statistical data
Gender (% of men)	67,8	84,6	91,3	92.1	H2(3) = 82,32, p <.001
Race (% Caucasian)	74.1	71.1	54.3	55,8	H2(3) = 40,86, p <.001
raise					H2(6) = 39,56, p < 0,001
less than HS	19.4	27.2	13.0	25.9
HS or equivalent	44.2	48,9	26.1	43,7
some universities and more than	36.4	23.9	60,9	30.4
Ort					H2(3) = 52,09, p < 0,001
Prison	39,9	47.2	67,4	63.2
drug abuse	60.1	52,8	32.6	36.8
producer price index
Gender (% of men)	84.1	60	84,2	89,7	H2(3) = 12,09, p <,01
Race (% Caucasian)	66,5	75	60,8	83,9	H2(3) = 13,14, p < .01
raise					H2(6) = 39,91, p < 0,001
less than HS	26.3	24	20.2	36.2
HS or equivalent	47,6	36	37,9	48.3
some universities and more than	26.1	40	41,9	15.5
Ort					H2(3) = 36,77, p < 0,001
Prison	48,7	44	65,7	37,9
drug abuse	51.3	56	34.3	62.1

open in a separate window

Observations.PCL-R = Psychiatric Checklist - Revised. PPI = Psychopathic Personality Inventory; F1 = factor 1; F2 = factor 2.

To determine whether F1 has an enhancing or protective effect, a significant comparison between the high F2 class and between high F1 and low F1 is needed to isolate the effect of F1 on high F2. Post hoc comparison (Figure 1.1-1.7) showed that the Low F1-High F2 and High F1-High F2 scales differed significantly on the magnitude of anxiety but not on other types of psychopathology. Anxiety levels were higher in the Low F1-High F2 scale than in the High F1-High F2 scale, suggesting a possible protective effect of F1.⁶A second set of ANCOVAs was performed to determine the presence of gender differences. Overall, gender had no effect on the association of all four categories with externalizing variables: BPD [eat(3) = .18,Pi= .913,Hey= 0], anxiety[eat(3) = .77,Pi= .510,Hey= 0.09], depression [eat(3) = 1,2,Pi= .320,Hey= 0.09], drugs [eat(3) = 0,17,Pi= .920,Hey= 0], alcohol consumption [eat(3) = 0,25,Pi= .247,Hey= 0.06], suicide[eat(3) = 1,41,Pi= .238,Hey= .11], child abuse[eat(3) = 1,58, p = 0,193,Hey= .11] and recaptured [Wald'sX²(1) = 3.42, p = 0.064, Exp(B) = 1.37 (0.98-1.9)]. Furthermore, a comparison of the "low F1-high F2" and "high F1-high F2" categories showed no significant effect of gender in: BPD [eat(1) = 0,02,Pi= .877,Hey= 0], anxiety[eat(1) = 0,31,Pi= .575,Hey= 0], Depressioneat(1) = .10,Pi= .758,Hey= 0], drugs [eat(1) = 0,22,Pi= .641,Hey= 0], alcohol consumption [eat(1) = 0,14,Pi=.708,Hey= 0], suicide[eat(1) = .79,Pi= .373,Hey= .06], child abuse[eat(1) = 0,43,Pi= .511,Hey= 0] and recapture [Wald'sX²(1) = 2,43,Pi= 0.119, End (B) = 1.47 (.91-2.39)].

difference between measurements

To determine whether measures of F1 influence the association between F2 and psychopathology, a formal comparison of regression weights was used. In this approach, the low F1-high F2 and high F1-high F2 categories are selected from the PCL-R and localized to psychopathology. The same procedure was repeated for the PPI class. Compare the unnormalized beta weights and standard errors of PCL-R regression and PPI regression using a cutoff table. +/- 2.58 critical z value forPi< .01. Results showed a significant difference (z=3.35) in depression, indicating a greater protective effect of PPI-I than PCL-R F1. However, there were no significant differences in BPD (z=0.89), anxiety (z=2.39), drug use (z=0.90), alcohol use (z=0.99), suicidality (z=1.07 ) and in child abuse (z = −.19) or rearrest (z = .02).⁸Thus, PCL-R F1 and PPI-I had the same effect at F2 on child maltreatment, psychopathology, and readmission. However, PPI-I showed a stronger effect on these outcomes.

improvement v. PROTECTION

In the analyses, there were few instances where F1 scores influenced the associations of F2 with psychopathology, childhood maltreatment, and re-encapsulation in either a protective or an ameliorative direction. There are a few exceptions: F1 was found to be protective against stress on the PCL-R and the PPI for anxiety, depression, and suicidality, meaning that individuals who scored high by 2 on each reading experienced these maladaptive outcomes increases “Low " is also rated with factor 1 as "high". However, we are not sure how much confidence there is in these results, as F1 neither protected nor enhanced in all other cases. these effects contrast with results showing a reinforcing effect (Code et al., 1993;Sprague et al., 2012;Verona et al., 2012). Our comparison results may be due to the control of covariates in the regression analysis, the use of latent profile analysis, and the small number of individuals in the main category comparisons. Furthermore, we used only the PPI, whereas previous studies examining reinforcement effects used a composite score of the PPI with another psychiatric measure and used a screening version of the PCL-R instead of the original measure. Despite these differences, F2 is likely to determine the relationship between measures of psychosis and psychopathology and other maladaptive outcomes, with little or no impact on F1, consistent with other studies finding that F2 alone has stronger associations and has yes ratios to F1 or its association with psychopathology and other external correlates (eg, recidivism, violence;Keneally et al., 2010;).

Measurement

Several notable findings emerged from the various measurements. First, the PCL-R and PPI still appear to function in similar ways related to psychopathology and other maladaptive behaviors. Formal comparisons further supported this conclusion and showed that in most cases there were no significant differences between the interventions. This study shows for the first time that the PCL-R and the PPI differ slightly in their external correlates, despite previous concerns about their different perceptions of F1.

Although there was little difference in the correlation between the two measures, the LPA models differed in sample size composition, and the PPI showed a stronger relationship with child maltreatment, psychopathology, and rearrest. For example, depression and suicidality were the only outcomes that differed and had a protective effect on PPI. Although methodological differences cannot be ruled out, these results are consistent with the nature of the PCL-R F1 and PPI-I (Benning et al., 2003;Skim et al., 2007;Warren et al., 2003) and opposed the parallelism of the two measures, at least in their conceptualization for F1. That is, PPI-I, representing fearless mastery, is largely an adaptive trait that has been consistently shown to be inversely related to psychopathological stress disorder (Benning et al., 2003;Patrick et al., 2006).

Therefore, in this case, F1 may act as a protective factor against some forms of mental impairment. This result is consistent with the fact that only 3% of samples had a high F1 value and a low F2 value in PPI. In general, we want people in this group to be less likely to be incarcerated than others. Further support comes from the fact that individuals with a high F1 score in this sample are more likely to have a higher level of education. In contrast, PCL-R F1, which mainly includes lack of empathy, callousness, and arrogance, is rather maladaptive and generally unrelated or weakly related to psychopathology.Warren et al., 2003). The potential protective effect of PPI-I may have important implications for therapeutic approaches, particularly in populations that respond poorly to treatment due to characteristics traditionally associated with F1 (eg, numbness, lack of empathy) (seeSalekin et al., 2010for review). An important difference, therefore, may be the adaptive value of PPI-I-related traits and their relationship to risk for psychopathology and other F2-related maladaptive behaviors.

From a broader perspective, the results of the current study raise questions about the utility of F1 in the conceptualization of psychosis, given growing evidence of its limited clinical relevance. In fact, a combination of F1 and F2 features is often used to identify psychosis, as shown by the total score of the psychosis measure. However, in the case of the PCL-R, F1 consistently lacked correlation with external correlates and failed to add additional value to F2, which was well captured by the ASPD (Keneally et al., 2010;). This raises the question of whether F1 is psychopathological in nature or should be considered part of a psychosis. Furthermore, the PPI-I tended to show stronger correlations with correlates considered orthogonal to psychopathy (e.g., psychological distress) than with those considered central to psychopathy (e.g., aggression, empathy, antisocial behavior), which contrasts with factors detected by the PPI-I PCL-R and other measures of psychopathy (Blonigen et al., 2010;;Differential correlations also hold for personality dimensions, as the PPI-I tends to be negatively related to neuroticism, positively related to extraversion, and unpleasantness, while the PCL-R F1 shows no correlation with negative and positive affect or weakly and negatively correlated with reconciliation (). Also, the degree of overlap between F1 and F2 varies considerably depending on the conceptualization and measurement used. Specifically, the PCL-R factors were generally correlated, but the PPI factors were not, obscuring the similarities between these two factors and highlighting the ongoing controversy regarding the utility of F1 in the conceptualization of psychosis. Furthermore, PPI-I subscales often overlap with PPI-II subscales, raising questions about optimal factor solutions (Newman et al., 2008).

Strengths, limitations and future directions

The current study has several strengths. The first is a comparison of two analytical methods for studying interaction effects. Although both methods have been used in psychosis research, they have not been used together, nor have latent profile analyzes been used to examine the statistical influence of F1 on F2 correlations. Furthermore, this is one of the first studies to specifically examine F1 and its impact on F2 across different measurement methods, with particular emphasis on examining measurement differences between PCL-R and PPI. Third, the sample included both men and women to examine potential gender differences in interaction effects.

Despite these advantages, there are still some limitations. The sample is fairly uniform in that it consists primarily of white male offenders and should be replicated in a more diverse sample. Second, the proportion of women in the sample compared to men is relatively small - although very high (>250). Third, several groups were small, which may have affected our ability to detect interactions. Fourth, given that both the PAI and PPI are self-reported measures, the slightly stronger effect on PPI (and the apparent protective effect) could be due to methodological differences. However, the potential implications of methodological differences are in stark contrast to the fact that the results overall are consistent with psychopathic theory and the results of previous studies. Fourth, despite empirical support for the two-factor structure of psychopathy, there is still debate about the factor structure of psychopathy. Instead of three faces () and the four-facet model of psychosis (NeumannWait.2013) is suggested. In further work with larger samples, it may be useful to examine statistical interactions between these facets in exploratory analyses, although in this sample (eg, comparing all pairwise combinations or facets or all together) this increased significantly the risk of a bergen type I' It is wrong given the large number of analyses.

In fact, analyzes using different factor structures/conceptualizations can lead to different results. For example, the three-way model analyzed traits related to F1, and validation studies showed that lack of emotional factors (factor 2 in the three-way model) was associated with criminal behavior and social withdrawal, while interpersonal traits (arrogant-deceptive interpersonal behavior) style factor) were associated with indicators of adaptive functioning (eg, positive mood, low stress reactivity) that were more strongly associated with potential protective effects (;). Thus, the lack of association in the current F1 study could be due to the two-factor model combining all F1 traits, negating the effect of the finer-grained analysis. Therefore, choosing a two-factor model of psychosis in this case could lead to a type II error. However, given the limited clinical relevance of this model and the two-factor model, the utility of the deficient affective component of F1 (and the three-factor model) in conceptualizing psychosis may also be questionable.

Future directions may include examining other components of F1 that may statistically influence F2. For example, it might be worth considering the 'passive' element of the PPI (which expresses a lack of emotionality, a lack of response to the suffering of others, a lack of guilt and empathy) or the related concepts of 'meanness', which include arrogance and aggression. include competition and aggressive aggression recently proposed as part of psychosis () have a protective or enhancing effect against psychopathology. Furthermore, examination of other models of psychopathy, such as the three- and four-facet models, may help further analyze the structure of psychosis and clarify the role of F1 traits in associations with external correlates. Next, examination of other well-validated measures of self-reported psychopathy could provide further insight into the reinforcing or protective effects of different psychopathy concepts. Finally, F1 and F2 appear to be heterogeneous combinations of traits derived from existing models of normal personality, such as the affective five-factor model. In particular, the PCL-R F1 mainly reflected low agreeableness, while the PPI-I reflected low neuroticism/negative affectivity. For example in a literature search:Lilienfeld, Smith, Watts, Berg, and Latzman (in press)The PPI-I, reflecting fearless dominance or boldness, was found to be strongly negatively correlated with Big Five personality model neuroticism and Big Three personality model negative emotionality. In contrast, the F2 of these two measures largely reflected low agreeableness and low conscientiousness, with little contribution from high neuroticism/negative affect (). Statistical interactions can emerge at the level of these seemingly more homogeneous dimensions. The answers to these and other unexplored questions can serve as a basis for prevention and intervention efforts for particularly difficult and burdensome populations.

1

Win Show:Dr. Lilienfeld is a co-developer of the Psychopathic Personality Inventory-Revised and receives royalties from the sale of this inventory published by Psychological Assessment Resources.

¹The study used data obtained as part of a grant from the National Institutes of Mental Health to examine personality traits in social deviance. Previous work has described the data collection process, other uses of the data, and other variables recorded in the dataset (;).

²It should be noted that the Cold Heart factor of the PPI was not used because it would not allow for a parallel comparison of two measures of psychopathy or to examine the nature of the interaction of psychopathy factors among externally correlated factors.

³Due to space limitations, the full regression model is not included and can be accessed assupplementary material.

⁴Categories were also derived from the mean split of the F1 and F2 scores of the PCL-R and the PPI and a subsequent analysis was performed. Despite the more even distribution of individuals across classes, the results for interaction effects, gender differences, and measurement differences speak to the current results.

⁵Due to the focus of this document on evaluation and space limitations, this information is not included in the table, but is available assupplementary material.

⁶Individual post hoc comparisons, which for reasons of space do not include all possible class comparisons, can be taken assupplementary material.

⁷To ensure that the inclusion of education as a covariate did not control for PPI-I-related traits (e.g., fearless dominance), the analysis was rerun without education as a covariate and showed no change in the observed results. In most cases, however, the PCL-R effect is weak and the PPI effect is strong.

⁸Due to space limitations, z-scores for each possible combination of class comparisons are not included and can be determined as followssupplementary material.

There are no conflicts of interest for any of the authors.

• Alterman A, McDermott PA, Cacciola J, Rutherford M, Boardman CR, McKay J, Cook TG. Antisoziale Typologie bei Methadonpatienten.Journal of Abnormal Psychology.1998;107: 412-422.[PubMed][Google Scholar]
• Ball DJ, Shanahan MJ. Modeling complex interactions: Human-centered and variable-centered approaches. In: Little TD, Bovaird JA, Card NA, eds.Modeling context effects in longitudinal studies.Lawrence Erlbaum Associates, Inc; Mahwah, NJ: 2007. S. 255-283.[Google Scholar]
• Benning SD, Patrick CJ, Hicks BM, Bronigan DM, Kruger RF. Factor structure of the psychiatric personality inventory: validity and implications for clinical assessment.Psychological assessment.year 2003?15: 340-350.[PubMed][Google Scholar]
• Blonigen DM, Patrick CJ, Douglas KS, Poythress N, Skeem JL, Lilienfeld SO, Edens JF. Multimethod assessment of psychoses in relation to internalizing and externalizing factors of a personality assessment inventory: methodological differences and the implications of inhibitory effects.Psychological assessment.2010;22: 96-107. [Free PMC items][PubMed][Google Scholar]
• CorderJW. Mood syndrome in psychopaths with borderline personality disorder?British Journal of Psychiatry.1993;162: 641-650.[PubMed][Google Scholar]
• Cooke DJ, Michie C. Refining the structure of psychosis: Towards a hierarchical model.Psychological assessment.2001;13: 171-188.[PubMed][Google Scholar]
• District Falls. Contemporary scientific opinion on the subject of psychosis.Psychological Science in the Public Interest.2011;12: 93-94.[PubMed][Google Scholar]
• Grove WM, Tellegen A. Problems in the classification of personality disorders.Journal of Personality Disorders.1991;5: 31-41. [Google Scholar]
• Haapasalo J, Pulkkinen L. The Psychiatric Checklist and the Nonviolent Offender Population.Criminal behavior and mental health.1992;2: 315-328. [Google Scholar]
• Hall JR, Benning SD. The "Successful" Psychopath: Adjustment and Subclinical Manifestations of Psychosis in Sex-Specific Populations. In: Patrick CJ, editor.Textbook of Psychiatry.Guilford Press; New York: 2006. S. 459–478.[Google Scholar]
• Hall JR, Benning SD, Patrick CJ. Criterion-based validity of a three-factor model of psychosis: personality, behavior, and adaptive functioning.Evaluate.200411: 4-16.[PubMed][Google Scholar]
• rabbit trail.The Hare Psychiatric Checklist – Revised Edition.Multiple Health Systems. Toronto: 1991.[Google Scholar]
• Hare RD, Harpur TJ, Hakstian AR, Forth AE, Hart SD, Newman JP. The psychosis inventory-revised: reliability and factor structure.Psychological assessment.1990;2: 338-341. [Google Scholar]
• Harpur TJ, Hare RD, Hakstian AR. Two-factor conceptualization of psychosis: construct validity and impact assessment.Psychological Assessment: A Journal of Counselling and Clinical Psychology.1989;1(1): 6–17. [Google Scholar]
• Henson JM, Reise SP, Kim KH. Detecting admixture based on differences in structural models using latent variable admixture modeling: a comparison of relative model fit statistics.Structural Equation Modeling: A Multidisciplinary Journal.2007;14:202-226. [Google Scholar]
• Herve H, Yong Hui Ling J, Hare RD. Criminal psychosis and its subtypes. Paper presented at the meeting of the American Psychological-Legal Association, New Orleans. March 2000.[Google Scholar]
• Hicks BM, Markon KE, Patrick CJ, Krueger RF, Newman JP. Identifying psychopathic subtypes based on personality structure.Psychological assessment.200416:276-288.[PubMed][Google Scholar]
• Hicks BM, Patrick CJ. Psychopathy and negative affect: Analysis of inhibitory effects shows differential associations with affective stress, anxiety, and angry hostility.Journal of Abnormal Psychology.year 2006?115:276-287. [Free PMC items][PubMed][Google Scholar]
• Jacobo MC, Blais MA, Baity MR, Harley R. Concurrent validity of personality assessment checklist thresholds in patients seeking dialectical behavior therapy.Personality Assessment Diary.2007;8874-80.[PubMed][Google Scholar]
• Jung T, Wickrama KAS. Introduction to latent class growth analysis and growth mix modeling.Compass for social and personality psychology.2008;2: 302-317. [Google Scholar]
• Karpman B. On the need to classify psychoses into two distinct clinical types: symptomatic and idiopathic.Journal of Criminal Psychopathology.1941;3: 112-137. [Google Scholar]
• Kennealy PJ, Skeem JL, Walters GD, Camp J. Do PCL-R psychopathy central interpersonal and affective traits interact with antisocial behavior and disinhibition to predict violence?Psychological assessment.2010;22: 569-580.[PubMed][Google Scholar]
• Lanza ST, Rhoades BL, Greenberg MT, Cox M. Multiple risk modeling in infancy to predict the quality of the caregiving environment: Contributions of a person-centered approach.Infant behavior and development.2011;34: 390-406. [Free PMC items][PubMed][Google Scholar]
• Lilienfeld SO. Conceptual problems in psychiatric assessment.Clinical Psychology Review.1994;14: 17-38. [Google Scholar]
• Lilienfeld SO, Andrews BP. Development and initial validation of a self-report measure of psychopathic personality traits in a noncriminal population.Personality Assessment Diary.1996;66: 488-524.[PubMed][Google Scholar]
• Lilienfeld SO, Watts AL, Smith SF, Berg JM, Latzman RD. Deconstructing and reconstructing psychopathy: identifying and assembling the personality building blocks of Cleckley's chimeras.personality magazine.In the news[PubMed][Google Scholar]
• Dr Lynam, Derefinko KJ. mental illness and personality. In: Patrick C, editor.Textbook of Psychiatry.Guilford? New York: 2006. 133–155.[Google Scholar]
• Malterer MB, Lilienfeld SO, Neumann CS. Concurrent validity of a psychiatric personality inventory with batterer and community samples.Evaluate.2010;17: 3-15. [Free PMC items][PubMed][Google Scholar]
• Marcus DK, Fulton JJ, Edens JF. A two-factor model of psychopathic personality: Evidence from the Psychopathic Personality Scale.Personality disorders: theory, research, and treatment.2013;4: 67-76.[PubMed][Google Scholar]
• Miller JD, Lynam DR. Examining the legal network of psychopathic personality recording: a meta-analytic review.Personality disorders: theory, research, and treatment.2012;3: 305-326.[PubMed][Google Scholar]
• Morey LC.Personality Assessment Inventories: A Professional Handbook.psychological assessment resources; Odessa, FL: 1991.[Google Scholar]
• Muthen LK, Muthen BO.Plus.Muthen & Muthen; Los Angeles, CA: 1998–2010.[Google Scholar]
• Neumann CS, Malterer MB, Newman JP. Psychiatric Personality Inventory (PPI) factor structure: results from a large sample of inmates.Psychological assessment.2008;20: 169-174. [Free PMC items][PubMed][Google Scholar]
• Patrick CJ, Edens JF, Poythress N, Lilienfeld SO, Benning SD. Construct validity of a two-factor model of the psychopathic personality inventory and the perpetrator.Psychological assessment.year 2006?18:204-208. [Free PMC items][PubMed][Google Scholar]
• Patrick CJ, Fowles DC, Kruger RF. Triadic Conceptualization of Psychosis: The Developmental Origins of Disinhibition, Audacity, and Meanness.development and psychopathology.2009;21: 913-938.[PubMed][Google Scholar]
• Poythress NG, Lilienfeld SO, Skeem JL, Douglas KS, Edens JF, Epstein M, Patrick CJ. Use of the PCL-R to assess the validity of two self-report measures of offender psychopathy.Evaluate.2010;17: 206-219.[PubMed][Google Scholar]
• Salekin RT, Worley C, Grimes RD. Treatment of psychosis: An overview and introduction to the mental model approach to psychosis.behavioral sciences and law.2010;28:235-266.[PubMed][Google Scholar]
• Sanders B, Becker-Lausen E. Measuring psychological maltreatment: early data from the Child Abuse and Trauma Scale.child abuse and neglect.1995;19: 315-323.[PubMed][Google Scholar]
• Saunders B, Giolas MH. Dissociation and childhood trauma in adolescents with mental disorders.American Journal of Psychiatry.1991;148: 50-54.[PubMed][Google Scholar]
• Sclove L. Application of model selection criteria to some problems in multivariate analysis.Psychometry.1987;52: 333-343. [Google Scholar]
• Skeem JL, Johansson P, Andersshed H, Kerr M, Louden JE. Two subtypes of psychopathic violent offenders alongside major and minor variants.Journal of Abnormal Psychology.2007;116: 395-409.[PubMed][Google Scholar]
• Skeem JL, Poythress N, Edens JF, Lilienfeld SO, Cale EM. Psychopathic Personality or Personality? Investigating potential variants of psychosis and their impact on risk assessment.aggression and violence.year 2003?8: 513-546. [Google Scholar]
• Sprague J, Javdani S, Sadeh N, Newman JP, Verona E. Is Borderline Personality Disorder Expressed as a Female Psychosis Phenotype?Personality disorders: theory, research, and treatment.2012;3127-139. [Free PMC items][PubMed][Google Scholar]
• Vassileva J, Kosson DS, Abramowitz C, Conrod P. Classification of psychopaths and psychopaths in offenders.Law and Criminal Psychology.2005;10: 27-43. [Google Scholar]
• Verona E, Sprague J, Javdani S. Interactions between gender and factor level in psychosis.Personality disorders: theory, research, and treatment.2012;3:247-262.[PubMed][Google Scholar]
• Vitale JE, Newman JP. Using a revised psychiatric checklist for a sample of women: reliability, validity, and implications for clinical utility.Clinical Psychology: Science and Practice.2001;8: 117-132. [Google Scholar]
• Walters GD, Duncan SA. Predicting release outcomes of inmates undergoing forensic examination using the PCL-R and PAI.Journal of Forensic Psychiatry and Psychology.2005;16: 459-476. [Google Scholar]
• Warren JI, Burnette ML, South SC, Chauhan P, Bale R, Friend R, Van Patten I. Female psychosis: structural models and comorbidities.International Journal of Law and Psychiatry.year 2003?26:223-242.[PubMed][Google Scholar]