Evaluating current practices in measuring and modeling adolescent alcohol frequency data

Evaluating current practices in measuring and modeling adolescent alcohol frequency data Public Deposited

Last Modified

Creator

McGinley, James S.
- Affiliation: College of Arts and Sciences, Department of Psychology and Neuroscience

Abstract

Substance use is a significant health risk behavior from both developmental and public health perspectives. In recent years, there has been substantial growth in the theoretical conceptualization of pathways to substance use during adolescence. However, in order to test these developmental theories researchers must be able to validly measure and model substance use. This project evaluated the current standard practices in measuring and modeling adolescent alcohol frequency data. Using a simulation study and empirical demonstration, I investigated the degree to which the quantitative characteristics of ordinal measures and ordinal scoring approaches impact researchers' ability to draw valid inferences from standard linear models. My results showed that ordinal alcohol frequency measures interacted with scoring approaches to substantially reduce statistical power and led to different patterns of effects. There was no clearly superior ordinal scoring approach and, in some conditions, the performance of scoring approaches depended on which measure was used.

Date of publication