This article investigates the relationship between IQ scores and criminal thinking as measured by the Criminal Sentiments Scale–Modified (CSS-M). We argue the CSS-M may not capture criminal thinking, but reflects the test-takers’ cognitive ability to interpret the test’s intentions. Based on our analysis of inmates admitted to Pennsylvania prisons in 2013, we find that highly intelligent inmates receive lower scores on the CSS-M, controlling for other measures of risk. In our analysis, the CSS-M’s ability to predict institutional misconduct is greatest for inmates whose IQ scores fall in the middle of the distribution, with a weaker relationship between CSS-M and misconduct found among low- or high-IQ inmates. We propose a need for reevaluation of actuarial assessment tools for cognitively low- or high-functioning inmates. Taking into account the social-environmental factors of the testing situation and being aware of the test-takers’ interpretative processes may be crucial for generating valid results.
All Science Journal Classification (ASJC) codes
- Pathology and Forensic Medicine