A Statistical Study of Crime in Ghana

Understanding the dynamics, patterns, and probabilities associated with the correlates of crime is a promising way to managing crime. In this study, a multinomial logistic regression was used to predict the propensity of individuals for committing particular crimes. The secondary data of 6702 prisoners was collated from Ghana Prisons Service for the purpose of the study. ANOVA and Brown-Forsythe robust tests of equality of means were employed, where the assumptions for homogeneity of variance were sustained and violated respectively. Pearson’s correlation matrix was also used in the analysis. Our findings showed that religious affiliation and educational level of convicts significantly affected the odds that they would commit a particular crime. Multinomial logistic regression analysis indicated that illiteracy significantly affected the odds that one would commit the crimes of manslaughter, rape, theft, causing harm, and issuing death threats. On the other hand, religious affiliation of an offender significantly affected the odds to commit the crime of murder. Educational level (r= -0.25; p< 0.05) and religious affiliation (r= -0.26; p<0.05) correlated negatively with crime. There were no significant differences in the mean score of crime across educational and religious levels. However, there were significant differences in the mean score of crime across age and gender. The mean difference from the post-hoc analysis showed a pattern of an initial rise in crime among the younger age group (8-25 years), a subsequent decline in the age group of 26-35, and a final surge in individuals beyond 35 years that did not surpass the initial peak. Females (M: 6.89, SD: 1.253) were found to have lower crime incidence than males (M: 7.43, SD: 3.008) for all crimes considered in this study. We recommend that Ghana’s Prison Service consider incorporating further demographic information of inmates in order to support research; which could help identify avenues for the amelioration of crime locally.