RMF2012 Abstracts

Session: Tuesday 3rd July PM (16.00 - 17.30)

Title: Comparative Methods for Cross-National Surveys Using Latent Variable Models

Name: Jouni Kuha

Affiliation: London School of Economics

Abstract Details

Latent variable models are used to analyse relationships among multiple survey items and to derive 'measurements' of latent concepts corresponding to the items. In cross-national surveys, a key aim is to compare levels of the latent constructs across countries. An additional challenge is then the comparability or equivalence of measurement across countries. There is a nascent literature on examining measurement equivalence with linear factor analysis models, but relatively little on latent variable models for categorical items (latent trait and latent class models). This session presents findings from an ESRC-sponsored project, which aims to develop methods for such analyses.

The level of the session is: Accessible

Presentation details

Presenter 1

Start time: 16:00

Presentation title: An overview: The problem of measurement equivalence in latent trait and latent class models, and the tools available to address it

Author: Dr Jouni Kuha

Affiliation: London School of Economics and Political Science

Presenter 2

Start time: 16:30

Presentation title: A methodological study: Sensitivity analysis of the consequences of ignoring measurement inequivalence in latent variable modelling of cross-national survey data

Author: Dr Irini Moustaki

Affiliation: London School of Economics and Political Science

Presenter 3

Start time: 17:00

Presentation title: An applied study: A latent trait model of scientific knowledge and a latent class model of attitudes towards biotechnology from the Eurobarometer survey

Author: Dr Sally Stares

Affiliation: London School of Economics and Political Science