Exploring the factor structure of measures of confidence in procedural justice and performance of the criminal justice system by actor

auteurs Anjuli Van Damme
  Lieven Pauwels
  Nicole Haas
tijdschrift GofS (ISSN: )
jaargang 2012
aflevering Social conflicts, citizens and policing
onderdeel Artikelen
publicatie datum 14 september 2012
taal English
pagina 119
keywords latent variable approach, measurement error, Reliability, correlational validity, attitudes towards the criminal justice system, procedural justice and performance of the criminal justice system
samenvatting

It is commonly acknowledged that many survey measures of confidence in the criminal justice system suffer from a number of measurement problems. A major criticism is the use of single-item questions to evaluate the criminal justice system in general. In this article we contribute to an existing knowledge of measures by assessing two theoretical dimensions of confidence in the criminal justice system: (1) procedural justice or fairness and (2) performance of the criminal justice system. The innovative part of this study is that it explicitly tests to what extent we can measure these dimensions by type of actor (the police, the public prosecutor, judges and the criminal justice system in general). Items that are highly similar in content, but different by type of actor, were presented to a sample of students enrolled at Ghent University. A latent variable modelling approach was used to test the reliability and correlational validity of our multiple-component instrument. Confirmatory factor analyses were conducted on a battery of items that measure procedural justice and performance of three actors within the criminal justice system and the criminal justice system in general. Special attention is paid to the existence of non-random error between indicators that result from question wording. Our analyses suggest that it is possible to distinguish between procedural justice and performance by actor. We tested and found that items that refer to the same actor have correlated error and we propose a way of dealing with it.