Citation

Paternoster R. J. Res. Crime Delinq. 1986; 23(2): 128-168.

Copyright

(Copyright © 1986, SAGE Publishing)

DOI

10.1177/0022427886023002003

PMID

unavailable

Abstract

Frequently, deterrence researchers have combined into composite scales items measuring attitudinal states (perceived risk, moral beliefs) and self-reported delinquency. These items have generally been offense-specific items that refer to behaviors as diverse as drug use, theft, and violence. The justification for the construction of composite scales has been based upon either the face validity of the item's content or simple factor analyses and reliability checks. This article more carefully examines the measurement properties of common scales employed in perceptual deterrence research. It is found that (1) the pooling of items across offenses and (2) the pooling of subsamples cannot be done indiscriminately. The effect of such indiscriminate pooling of items and subjects on a structural equation model of the deterrence process is examined.