APHA
Back to Annual Meeting Page
 
American Public Health Association
133rd Annual Meeting & Exposition
December 10-14, 2005
Philadelphia, PA
APHA 2005
 
5141.0: Wednesday, December 14, 2005 - 12:45 PM

Abstract #107075

Effect of correlated errors on the test-retest reliability of self-reported age of first drug use

Scott L. Hershberger, PhD1, Dennis G. Fisher, PhD1, Grace L. Reynolds, DPA1, Charles L. Henry, JD2, Mario J. Perez2, and Adi Jaffe, BA1. (1) Center for Behavioral Research & Services, California State University, Long Beach, 1090 Atlantic Avenue, Long Beach, CA 90813, 562-495-2330, scotth@csulb.edu, (2) Office of AIDS Program and Policy (OAPP), Los Angeles County Department of Health Services, 600 S. Commonwealth Avenue, 6th Floor, Los Angeles, CA 90005

Introduction: Self-reported age of first drug use is frequently measured in longitudinal studies examining alcohol and illicit drug use. Although it is well known that memory can bias estimates of test-retest reliability, less well known are the effects of measurement error correlations. Test-retest correlations as estimates of reliability are only valid to the extent the assumptions of the true score model of reliability are met, among which are (1) that measurement errors are uncorrelated across occasions and that (2) measurement errors are uncorrelated with true score estimates of age of first drug use. True-score models are proposed that can account for correlations between errors and correlations between errors and true scores. These models allow random measurement error on earlier occasions to affect directly or indirectly responses on later occasions. Methods: Data were obtained from street-recruited men and women on four occasions approximately three months apart using the Risk Behavior Assessment (RBA). The initial sample size was 1600. Respondents were asked the age at which they first used alcohol, marijuana, cocaine, heroin, nonprescription methadone, and amphetamines. The test-retest reliability of responses was examined using structural equation modeling. Results: The test-retest reliability of age of first drug use is seriously overestimated if measurement error correlations and measurement error-true score correlations are not taken into account. Conclusions: In order for between-occasions correlations to be correctly interpreted as reliability estimates, investigators should first verify that the assumptions of the true-score model of reliability are met.

Learning Objectives:

Presenting author's disclosure statement:

I wish to disclose that I have NO financial interests or other relationship with the manufactures of commercial products, suppliers of commercial services or commercial supporters.

Measurement Issues: Validity and Reliability

The 133rd Annual Meeting & Exposition (December 10-14, 2005) of APHA