Comparison of Applicability of Bayesian and Frequentist Statistics in Forensic Science Through a Fictitious Investigation

• Authors: Orbán, József

Abstracts

EN

Shrinking budgets and resources are a challenge for forensic scientists as increasing efficiency is expected from forensic investigators. The growing number of certain crimes caused by the recent economic crisis has accelerated the development of efficient and quick methods. The easiest way to improve efficiency is the introduction of the philosophy of knowledge-transplantation, based on acquired results in other sciences. This study provides an insight based on applicable mathematical results. The recommended frequentist and Bayesian statistics can serve as affordable tools for increasing the efficiency of forensic investigators’. The required mathematical method is selected according to the objective targeted at the beginning of the process. By means of a hypothetical case study, the opportunities offered by frequentist and Bayesian statistics are used to demonstrate their applicability in investigation and decision-making. The demonstration centres on a special situation of a stolen computer containing sensitive information.In the given situation the first and most urgent action is to get back the stolen equipment and then to catch the perpetrator. The suggested steps based on statistical methods are different. In the case of the suspect being the target statistical occurrences may lead closer to the result and the estimation uses the frequency of residence; therefore a frequentist calculation is carried out.Taking into consideration if the subject of the crime shall be targeted it is suggested that a Bayesian inference be used. This is based on a priori hypothesis of the suspect’s expected behaviour and his or her assumed relation to, as well as his follow on actions with, the computer. The paper encourages decision makers to use both mathematical methods as an aid while special attention has to be given to their proper use.

Keywords

EN

Forensic science; investigation; bayesian statistical model; frequentist statistical model

• Publisher: Wyższa Szkoła Policji w Szczytnie
• Journal: Internal Security
• Year: 2013
• Volume: 5(1)
• Pages: 197-214

Contributors

author

Orbán, József

Identifiers

DOI

10.5604/20805268.1059485