Our overarching goal is to develop, evaluate and disseminate methods of latent variable modelling that are designed to extract maximum reliable and valid information from cross-national surveys that collect categorical data. Within this broad remit, there are six specific objectives:

Research

  • To examine the utility of different methods of statistical model selection for categorical data when applied to latent class and latent trait modelling of cross-national surveys.
  • To explore conceptually and empirically the pragmatics of assessing equivalence of measurement in cross-national analyses. The aim is to explain and compare the courses of action available to a survey analyst under different circumstances of measurement equivalence, and to assess their impact on substantive conclusions. These, as well as the methods of model selection discussed above, will be compared using analytical and simulation techniques as well as empirical analyses of current survey data in three areas of application: on confidence in criminal justice systems, public attitudes towards science and technology, and characteristics of civil society.
  • To explore the implications of the conclusions of such latent variable approaches for future cross-national survey design. For example, existing scales might be improved by removing indicators with poor measurement properties, or developing items that do not perform well in cross-national contexts.

Teaching and dissemination

  • To provide accessible descriptions and examples of how latent class and latent trait models are used to analyse cross- national survey data. Such modelling is used both to answer substantively interesting comparative research questions and to examine cross-national equivalence of survey measurements.
  • To build capacity in comparative cross-national research methods by providing training in latent class and latent trait modelling for cross-national survey analysis. Communities to be targeted include doctoral researchers, early career and experienced social scientists, academics who teach methodology at the undergraduate and graduate level, and researchers who engage in comparative survey research for governmental and non-governmental organisations. The training has been delivered via hands-on training workshops. The project website also provides¬† user’s guides and examples implemented in the Mplus and R computer software.
  • To disseminate the results of the research in conference and seminar presentations and academic articles.