Psychopathy and Violent Crime

Method

The methodology of the study was a mixed-methods assessment utilizing the psychopathy checklist-revised (PCL-R) found in Hare (1991) as cited in Camp, Skeem, Barchard, Lilienfeld, and Poythress (2013). The PCL-R is a self-reporting checklist, offering insight into how a violent person perceives themselves. While the checklist provided the qualitative data, researchers used semi-structured interviews to inform their qualitative analysis (Camp et al., 2013). To ensure quantifiable reliability a secondary research team was brought in to independently analyze the lifetime patterns of the participant’s history of violence in order to more accurately understand the motivation. Together these research steps provided the baseline data. In order to quantify this data over time, participants (violent offenders in prison serving lengthy prison sentences) were followed over one year to test their involvement in the three different forms of violence outlined in the PCL-R. The method of the follow-up assessment included a detailed observance of official records and secondary interviews after one year (Camp et al., 2013).

Restatement of Hypotheses

The following hypotheses were specified in the study.

Hypothesis 1:  There is a significant correlation between psychopathy, anti-social personality disorder, and the propensity towards violent crimes and aggressive behavior.  Individuals who have a serious mental illness are significantly more likely to engage in aggressive criminal behavior than people without mental disorders.  

Hypothesis 2: When individuals with serious mental illnesses lack adequate and effective treatment, the likelihood they will engage in violent crimes and aggressive criminal behavior is significantly higher than when they are receiving treatment.

Hypothesis 3: Criminal aggression is not, itself, indicative of mental illness.  However, criminals with serious mental illness are significantly more likely to be involved in reoffending and recidivism.

Participants

Snowball sampling was utilized to recruit participants for the study. Using Facebook, the research disseminated a broadly worded study recruitment message targeted at American adults between the ages of 21 and 55. Individuals who agreed to complete the questionnaire for the study were given a $10 Amazon gift certificate and were asked to invite others of their acquaintance to participate in the survey. Online sampling expanded the researcher’s geographic reach, and the use of snowball and incentive methods increased the pool of potential study participants.

Measures and Procedures

The sequential procedures for the study were as follows:

1. Participants viewed the recruitment message on Facebook.

2. Participants who clicked on the recruitment message were given a link to an informed consent form, which was hosted by Survey Monkey.

3. Participants who affixed their signatures to the online informed consent form were taken to the study page, also hosted by Survey Monkey.  

4. The study page contained online versions of the measures utilized in the study that participants filled out.

5. Upon completion of the survey, participants were issued their $10 gift certificate and asked to send the survey link to others of their acquaintance. 

6. After the conclusion of the 2-week data collection period, data were downloaded from Survey Monkey into Excel format and subjected to the data analyses described subsequently.

7. All online forms were deleted from Survey Monkey’s servers in order to maintain the privacy and confidentiality of study participants. The raw data of the study were retained only in Excel format and stored on the password-protected, encrypted hard drive of the researcher.  

8. All data were analyzed in Stata / 15.2 SE statistical software.

The measures of the study can be described separately for each of the hypotheses. In Hypothesis 1, measures were needed for the predictor variables of psychopathy and antisocial personality disorder as well as for the outcome variable of aggressive behavior. Psychopathy was measured as a global score on the Levenson Self-Report Psychopathy Scale (LSRPS) (Levenson, Kiehl, & Fitzpatrick, 1995), which measures psychopathy on a scale ranging from 26 to 130, with higher values indicating higher levels of psychopathy. Antisocial personality disorder was measured through global score on the Antisocial Personality Questionnaire (APQ) (Blackburn & Fawcett, 1999), which measures antisocial tendencies on a scale ranging from 24 to 120, with higher scores representing higher levels of antisocial personality and behavior. Aggressive behavior was dichotomized into a self-reported measure of whether the study participant had ever been charged with assault, sexual assault, homicide, or battery.  

In Hypothesis 2, measures were needed for the predictor variable of receiving mental health treatment and the outcome variable of engaging in violent crimes and aggressive criminal behavior. The outcome variable was taken from Hypothesis 1 and consisted of whether the study participant had ever been charged with assault, sexual assault, homicide, or battery. The predictor variable was whether the participant had ever (a) been an inpatient in a mental health facility, (b) spent more than 12 continuous months in psychotherapy, or (c) received at least 12 continuous months of pharmacological treatment for a mental health issue.   

In Hypothesis 3, measures were needed for the predictor variable of having a serious mental illness as well as for the outcome variable of recidivism. Recidivism was defined as whether a participant had ever been re-incarcerated. Having a serious mental illness was also dichotomized as whether a participant had ever been diagnosed with schizophrenia, bipolar disorder, or any psychotic disorder (such as brief psychotic disorder or substance-induced psychotic disorder). 

Data Analysis

The data analyses utilized in the study have been described separately for each hypothesis. For Hypothesis 1, a logistic regression model was applied in which the two predictors were (a) score on LSRPS and (b) score on APQ and the outcome variable was (a) having been charged with assault, sexual assault, homicide, or battery (coded as 1); or (b) not charged with assault, sexual assault, homicide, or battery (coded as 0). The predictors for Hypothesis were dichotomized in the manner described below. 

It would have been possible to allow variables (a) and (b) to be coded as continuous variables, but, based on the existing literature, it seemed more likely that, for instance, individuals who surpassed the critical value for psychopathy would be more likely to be aggressive, The LSRPS and APQ each have critical values; an individual who surpasses the critical value of the LSRPS can be described as exhibiting psychopathic tendencies, whereas an individual who surpasses the critical value of the APQ can be described as exhibiting an antisocial personality. Therefore, for Hypothesis 1, raw scores on the LSRPS and APQ were dichotomized so as to identify individuals who were (a) psychopathic, (b) not psychopathic, (c) antisocial, and (d) not antisocial.  

An odds ratio (OR) was generated so that (a) being psychopathic and (b) being antisocial could be associated with changes in the odds of having displayed aggressive or violent behavior. The null hypothesis was that neither being psychopathic nor being antisocial would have ORs that were significant (at p < .05) predictors of having displayed aggressive or violent behavior. The null hypothesis for Hypothesis 1 would, therefore, be rejected if the ORs for either LSRPS or APQ score had p values below .05.   

For Hypothesis 1, having committed an aggressive crime was coded as 0, whereas not having committed an aggressive crime was coded as 1. Similarly, psychopaths were coded as 1 and non-psychopaths as 0. In the same manner, antisocial people were coded as 1 and non-antisocial people as 0. The 0 / 1 coding choices for the logistic regression structure in Hypothesis 1 indicate that, in an OR significantly greater than 1, having a certain condition (psychopathy or antisocial personality) would be associated with increased odds of having committed a violent or aggressive crime. For example, serial killer Ted Bundy has deemed a psychopath and he committed serious acts of heinous crime. 

Hypothesis 2 was also tested by means of a logistic regression in which the predictor was ever having received mental health treatment. Because Hypothesis 2 applied to study participants who had a mental illness, the first step in the analysis was to delimit the sample to only those participant who described themselves as ever having been diagnosed with schizophrenia, bipolar disorder, or any psychotic disorder (such as brief psychotic disorder or substance-induced psychotic disorder). Then, for this sub-sample only, mental health treatment was defined as ever having (a) been an inpatient in a mental health facility, (b) spent more than 12 continuous months in psychotherapy, or (c) received at least 12 continuous months of pharmacological treatment for a mental health issue. For analytical purposes, any participant who satisfied conditions (a), (b), or (c) was coded as 0, whereas any participant who failed to meet all of these conditions was coded as 1.  As in Hypothesis 1, the outcome variable for Hypothesis 2 was (a) having been charged with assault, sexual assault, homicide, or battery (coded as 1); or (b) not charged with assault, sexual assault, homicide, or battery (coded as 0). Because of the coding structure of the logistic regression for Hypothesis 2, an OR significantly over 1 would mean that not having received treatment would be associated with a higher likelihood of having committed a violent crime.

For Hypothesis 3, logistic regression was also used. Any participant had ever been diagnosed with schizophrenia, bipolar disorder, or any psychotic disorder (such as brief psychotic disorder or substance-induced psychotic disorder) was coded as 1, whereas any participant who had not been diagnosed with any of these disorders or diagnoses was coded as 0. Any participant who had been incarcerated more than once was then coded as a 1, whereas participants who had been incarcerated only once were coded as 0. Because Hypothesis 3 was based on the outcome of recidivism, it was delimited only to those individuals who had been incarcerated at least once. 

Conclusion

Chapter 3 contains a restatement of the hypotheses of the study and related them to specific approaches to data collection, instrumentation, and analysis. For each of the hypotheses, details were provided on logistic regression approaches capable of hypothesis testing. The fourth chapter of the study contains the findings associated with each hypothesis.  

Chapter 4: Results

Introduction

The purpose of Chapter 4 is to present the results of the study. The results have been presented in order of the hypothesis. Each hypothesis has been tested and, in appropriate cases, is accompanied by supporting graphics. 

Hypothesis 1 Results

The first hypothesis of the study was as follows: There is a significant correlation between psychopathy, anti-social personality disorder, and the propensity towards violent crimes and aggressive behavior.  Individuals who have a serious mental illness are significantly more likely to engage in aggressive criminal behavior than people without mental disorders.  The confidence interval plot in Figure 1 below indicates that the proportional likelihoods of being a violent criminal, being a psychopath, and being antisocial.  

(Figure 1 omitted for preview. Available via download)

The logistic regression of the odds of being a violent criminal on psychopathy and antisocial status was significant, χ2 = 36.15, p < .001. It was found that being a psychopath made a participant 1.88 times likely to have committed a violent crime than someone who was not a psychopath, p = .034. Next, it was found that being antisocial made a participant 2.66 times likely to have committed a violent crime than someone who was not antisocial, p = .001. Therefore, the null hypothesis for Hypothesis 1 was rejected, as the odds of having committed a violent crime were, at p < .05, positively correlated to either being a psychopath or being antisocial.    

Hypothesis 2 Results

The second hypothesis of the study was as follows: When individuals with serious mental illnesses lack the adequate and effective treatment, the likelihood they will engage in violent crimes and aggressive criminal behavior is significantly higher than when they are receiving treatment. 

(Figure 2 omitted for preview. Available via download)

For purposes of analysis, the 200 individuals who responded were sub-sampled into the 30 individuals who reported having a serious mental illness. Within this sample, a logistic regression model was run in order to relate the odds of engaging in violent crimes and aggressive criminal behavior to receiving mental health treatment.  The logistic regression of the odds of being a violent criminal on treatment status, delimited to those individuals with serious mental illnesses, was significant, χ2 = 5.33, p = .02. It was found that receiving treatment made someone with a serious mental illness 1.91 times less likely to have committed a violent crime than someone with serious mental illness who did not receive treatment, p = .035. Therefore, the null hypothesis for Hypothesis 2 was rejected, as the odds of those with serious mental illness having committed a violent crime were, at p < .05, significantly reduced by having been exposed to treatment.      

Hypothesis 3 Results

The third hypothesis of the study was as follows: Criminal aggression is not, itself, indicative of mental illness.  However, criminals with a serious mental illness are significantly more likely to be involved in reoffending and recidivism. Hypothesis 3 was delimited to those individuals in the sample (n = 40) who had been incarcerated at least once. Once the sample was delimited in this manner, it was possible to run a logistic regression related to the odds of having recidivated to variation in serious mental illness status. 

The logistic regression of the odds recidivating on serious mental illness status, delimited to those individuals who had been incarcerated at least once, was significant, χ2 = 4.00, p = .04. It was found that having a serious mental illness made someone who had been incarcerated at least once 1.35 times more likely to have recidivated, p = .05. Therefore, the null hypothesis for Hypothesis 3 was rejected, as the recidivism odds of those who had been incarcerated at least once were, at p < .05, significantly increased by having a serious mental illness.

(Figure 3 omitted for preview. Available via download)

Conclusion

Several statistically significant findings were generated by the logistic regression models applied in Chapter 4. It was found that:

• Being a psychopath made a participant 1.88 times likely to have committed a violent crime than someone who was not a psychopath.

• Being antisocial made a participant 2.66 times likely to have committed a violent crime than someone who was not antisocial. 

• Receiving treatment made someone with a serious mental illness 1.91 times less likely to have committed a violent crime than someone with serious mental illness who did not receive treatment.

• Having a serious mental illness made someone who had been incarcerated at least once 1.35 times more likely to have recidivated.  

References

Blackburn, R., & Fawcett, D. (1999). The Antisocial Personality Questionnaire: An inventory for assessing personality deviation in offender populations. European Journal of Psychological Assessment, 15(1), 14-24. 

Camp, J. P., Skeem, J. L., Barchard, K., Lilienfeld, S. O., & Poythress, N. G. (2013). Psychopathic predators? Getting specific about the relation between psychopathy and violence. Journal of Consulting and Clinical Psychology, 81(3), 467-480. doi:10.1037/a0031349

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.