Predicting Police Integrity: An Application of Support Vector Machines (SVM) to the Police Integrity Instrument

Research using the 11-scenario police integrity instrument designed by Klockars et al. document a range of factors influencing the willingness to report a fellow officer for police crime and police misconduct. A consistent quandary within this scholarship is that while some findings are consistent, when disaggregated by scenario type, there are wide variations obscuring patterns that may allow for targeted interventions improving police integrity. This study applies support vector machines (SVMs) to construct predictors for 608 responses to the Police Integrity Instrument from police officers enrolled in a police university for in-service training in China. Results confirm that while perceptions of seriousness remain the most successful predictors of the self-reported willingness to report a fellow officer, perceptions of seriousness associated with ethical dilemmas display high survivability suggesting targeted interventions may be an effective pathway towards improving police integrity.