(tylko po angielsku) |
In the first part of the course we deal with confirmatory factor analysis. This includes specification, identification, estimation and model modification of CFA models. Then we deal with ordinal confirmatory factor analysis and the treatment of missing values. In a next step we discuss simultaneous confirmatory factor analysis (SCFA) for many items and factors and multigroup factor analysis (MGCFA and MGSCFA) to study differences between groups, organizations and countries. This includes the estimation of latent means instead of observed means and new techniques for testing measurement invariance like alignment and Bayesian CFA. Finally we discuss the use of multitrait-multimethod approaches and higher-order or hierarchical confirmatory factor analysis.
In the second part the specification, identification, estimation and model modification of full structural equation models are presented. In a next step the computation of direct, indirect and total causal effects, and their standard errors and full and partial mediation are dealt with. The following topic is then multigroup structural equation modeling for comparative and cross-cultural research and moderation. Finally we deal with models employing formative and reflective indicators (so called Mimic Models) and Feedback models. If time allows we additionally present models to handle longitudinal data like autoregressive models and latent growth curves.
All procedures in CFA and SEM are illustrated using data from the European Social Survey employing the scales for values and attitudes toward immigration.
Classes will be held on the following dates:
22, 23 and 24th May 2017; 2 pm - 6 pm