Abstract
Most studies of prisoners, the nature of their offending and any related characteristics have been conducted in rich Western countries. In Nigeria, prison conditions differ in many important respects, key among them that prisoners share large communal spaces much of the time—up to 50 men sleeping in the same space as well as spending the day together. Our aim was to compare levels of impulsivity, aggression and suicide‐related behaviours between prisoners in one prison in Nigeria who had committed a homicide and those who had not, allowing for socio‐demographic factors. A case–control study design was employed with 102 homicide and an equal number of non‐homicide offenders. Each participant was interviewed using the Abbreviated Barratt Impulsiveness Scale for impulsivity, the Modified Overt Aggression Scale for aggression, the MINI International Neuropsychiatric Interview (Module C) for suicide‐related behaviours, and a questionnaire for ascertaining socio‐demographic characteristics. On bivariate analysis, motor impulsivity was higher among homicide offenders (p = .014) while non‐planning was higher among non‐homicide offenders (p = .006), but this relationship was affected by demographic variables. Physical aggression levels did not distinguish the two groups, but homicide offenders were less likely to record property‐directed and auto‐aggressive behaviours (p < .05). By contrast, on average, scores for suicide‐related behaviours were lower among the homicide offenders (p = .001), with non‐homicide offenders showing a mean score in the high‐risk category (13.25; SD, 1.25). As motor impulsivity significantly differentiated the groups, this may be an important measure to add to any risk assessment battery when there are concerns about homicidal behaviours. In this sample, history of interpersonal aggression did not distinguish the groups. People with indications of self‐harm or suicide‐related behaviours may be at less risk of violence to others, but care should be taken in interpreting this finding as it is not entirely consistent with other findings.