Session: Measurement Issues and Analyses for Public Health Research and Evaluation

5081.0: Wednesday, November 10, 2004: 8:30 AM-10:00 AM

Oral

Measurement Issues and Analyses for Public Health Research and Evaluation

Quantifying constructs required to evaluate the health of a population involves going beyond direct measurement. While mortality can be measured directly, constructs such as physical, psychological and social functioning often require knowledge of more sophisticated statistical techniques and psychometrics. Moreover, advanced techniques such as structural equation modeling, item response theory, logistic regression and confirmatory factor analysis can be extremely useful for discerning causal pathways and identifying problems in measurement across social conditions and other predictors of health disparities. This session will highlight techniques which are useful to individuals involved in measuring health and identifying causal pathways for health disparities and health outcome differences.

Learning Objectives: At the end of this session, the participant will be able to: 1. Understand the role of 1) measures of clustering in logistic regression to investigate contextual effects and 2)robust variance estimates for cluster-correlated survival data from surveys. 2. Identify a a new tool for evaluation and policy in primary care research. 3. Desribe the contributions of confirmatory factor analysis for assessing measurement bias in public health research and methods for estimating the size of eligible populations in the study of women, infants and children (WIC).

Moderator(s):

Stuart A. Gansky, DrPH

8:30 AM

Using measures of clustering in logistic regression to investigate contextual effects