Conferences

ESRA2009:

Conference main page

Overview of sessions

Time table

Warsaw 2009: Presentations and short courses

Marginal models and measurement issues

Session: Marginal models for dependent data

Author:

Marcel Croon; Tilburg University, Netherlands

Abstract:

Marginal models for dependent observations offer opportunities to solve some questions pertaining to the quality of the measurement procedures in the social sciences.

It will first be made clear how marginal models can be applied to test hypotheses about various scalability and reliability coefficients for measurement scales. Up till now only a limited number of such tests based on normality assumptions and approximations were available. Marginal models enlarge the spectrum of statistical tests for hypothesis about scalability and reliability coefficients.

Marginal models also provide means to analyse data from multiple response items in which respondents may choose an arbitrary number of alternatives from a given set of response options. By means of marginal modelling, the relationship between a multiple response item and other variables (including eventually other multiple responses items) can be studied.