Assessing the validity of police integrity scale in a comparative context

ABSTRACT

The current study examines the validity of the police integrity measure created by Klockars et al. (1997, The measurement of police integrity. Washington, DC: National Institute of Justice. pp. 65–70; 2006, The Measurement of Police Integrity. Research in Brief. U.S. Department of Justice, Office of Justice Programs, National Institute of Justice: Washington, DC: Government Printing Office. Available from: https://www.ncjrs.gov/pdffiles1/nij/181465.pdf). Whereas this theoretical and methodological approach has been used by many scholars across the world, there are no systematic assessments of its validity. We primarily investigate the criterion validity by assessing whether the data on police integrity are appropriately correlated with other independent indicators of integrity. We rely on the descriptive statistics from many prior police integrity studies conducted over the last 20 years and on other nation/state level available sources of data (Transparency International Corruption Perceptions Index, the International Crime Victimization Survey, the World Value Survey, and the World Governance Indicators). We find relatively robust evidence of criterion validity of the measures, although the strength of this relationship varies across criteria. The theoretical, methodological, and policy implications are discussed.

KEYWORDS:

• Police integrity
• validity
• scales
• comparative

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 One of the scenarios they include in the analysis is Scenario 10 that presents a case of excessive force and not police corruption (the behaviour described in the scenario does not have the for-gain element).

2 The initial measures used a five point Likert scale, but some of the studies only reported the % of respondents who indicated that no punishment would/should follow. Therefore, only estimates from studies that could be converted into this measurement scheme were included here.

3 For both the predictive and postdictive validity analyses, there were a total of 28 coefficients for each.