From unsolved crime numbers to other forensic statistics: Opole Police Department’s observations using Bayesian modeling

Abstract

The non-detectability of crimes is a big problem for maintaining public order, and by reviewing scientific reports on this subject it seems that this problem can even be considered ‘shameful’. For this reason, we undertook in this study an assessment of this forensic phenomenon adapting modern geostatistical methods. This study uses all reported incidences of crimes to the Opole Provincial Police Headquarters committed in the Opole region over a five year period 2015–2019. Spatio-temporal patterns of undetected crime rates were built. Their annual five-year growth rates are based on the reference population using hierarchical Bayesian modeling. The combinations of these levels were compared with the odds of detected to undetected crime numbers. The statistical analysis was extended with square area and crude rates of the reported crime statistics per reference population. We found that the degree of crime detection is directly related to the reference population size which may also suggest the validity of a detailed analysis in terms of adjusting the appropriate staffing of police personnel in administrative units. Furthermore, the influence of the area of administrative territories on the number of undetected crimes was also evident. Additionally, overall reported crime rates and the fluctuations in time within territorial units themselves implicated more important changes in criminal statistics than the differences between them and were established using unsolved crime statistics only. The use of Bayesian methods (despite the fact that they are statistically advanced, is simple in a computer application) allows for the unfolding of geostatistical data (regardless of the type) and new possibilities of scientific inference. All these statistical results are quite easy to achieve with the current programming technique, nevertheless they require the cooperation of interested professional forensic and statistical bodies and further scientific explanations. Finally, it is worth emphasizing that Bayesian methods allow brand-new cognitive horizons to be opened in forensic research.