Back to Annual Meeting

Back to Annual Meeting

APHA Scientific Session and Event Listing

Ruth L. Eudy, MSW, PhD, College of Public Health, University of Arkansas for Medical Sciences, 4301 West Markham, Slot 820, Little Rock, AR 72205, 501-526-6656, eudyruth@uams.edu

One of the key issues for public health research and evaluation is separation of effect size from background noise in the form of measurement error. Structural equation modeling (SEM), a statistical technique that developed in the disciplines of sociology, econometrics and psychology, can provide public health researchers with a powerful tool for addressing measurement error. SEM is particularly effective when working with multi-item scales in data sets with fairly large samples. Many of the nationally available data sets, such as the NLS, NHANES, NHIS and others contain several multi-item scales measuring attitudes, health behaviors, developmental levels and mental health. Utilizing summative scales to represent these concepts can lead to underestimation of actual effect sizes in structural equations such as regression, path analysis or time series analysis. This presentation uses data from the NHIS (National Health Interview Survey) to illustrate SEM measurement models for adult health status and adult mental health twelve months later. The focus is on the steps in model development, testing for reliability and validity, sample size requirements and interpretation of fit measures. Next, a structural model testing the relationship of health status to mental health, controlling for demographic factors, is used to demonstrate the technique of model estimation using both observed and latent variables. Results obtained using SEM are compared to results of traditional regression analysis with summative scales. Discussion includes the implications and applications of this technique for public health research and evaluation in diverse settings.

Learning Objectives:

At the conclusion of this session, participants will be able to
• Describe the steps in developing latent variable and structural models and interpreting output.
• Identify the strengths of structural equation modeling as a statistical tool for public health research and evaluation.
• Acquire handouts with instructions and programming language and a bibliography for further study of SEM.

Keywords: Public Health Research, Evaluation

Presenting author's disclosure statement:

Any relevant financial relationships? No