Methodenzentrum

Keynotes

Kenneth Bollen

Model Implied Instrumental Variables (MIIVS): A New Orientation to Structural Equation Models (SEMS)

Kenneth A. Bollen
University of North Carolina
at Chapel Hill

You can find the video of the keynote here.

Abstract

It is hardly controversial to say that our models are approximations to reality.  Yet when it comes to estimating structural equation models (SEMs), we use estimators that assume true models (e.g., ML) and that readily spread bias through estimated parameters when the model is approximate.  This talk presents the Model Implied Instrumental Variable (MIIV) approach to SEMs initially proposed in Bollen (1996).  It has greater robustness to structural misspecifications and the conditions for robustness are well understood.  In addition, the MIIV-2SLS estimator is asymptotically distribution free.  Furthermore, MIIV-2SLS has equation based overidentification tests that can help pinpoint errors in specification.   Beyond these features, the MIIV approach has other desirable qualities.  It permits new tests of dimensionality and tests of causal vs. reflective indicators.  MIIV methods apply to higher order factor analyses, categorical measures, growth curve models, and nonlinear latent variables.  Finally, it permits researchers to estimate and test only the latent variable model or any other subset of equations.   This presentation will provide an overview of this new orientation to SEMs and illustrate MIIVsem, an R package that implements it.

Xinyuan Song

Joint Modeling Approach for Analyzing Complex Data with Latent Variables

Xinyuan Song The Chinese University of Hong Kong

You can find the video of the keynote here.

Abstract

This talk introduces several joint modeling approaches for analyzing complex data with latent variables. Several statistical models, including hidden Markov model, additive hazards model, transformation model, and regularized regression model, are considered to analyze multivariate longitudinal data, time-to-event data, and other non-normal data in the presence of latent variables. The estimating equation method, EM algorithm, and Bayesian methods are used to conduct statistical inference. Applications to real-life studies are presented.

Carolin Strobl

Model-based recursive partitioning of psychometric models: A data-driven
approach for detecting heterogeneity in model parameters

Carolin Strobl
Universität Zürich

You can find the video of the keynote here.

Abstract

Model-based recursive partitioning is a flexible framework for detecting differences in model parameters between two or more groups of subjects. Its origins lie in machine learning, where its predecessor methods, classification and regression trees, had been introduced around the 1980s as a nonparametric regression technique. Today, after the statistical flaws of the early algorithms have been overcome, their extension to detecting heterogeneity in parametric models makes recursive partitioning methods a valuable addition to the statistical “toolbox” in various areas of application, including econometrics and psychometrics. This talk gives an overview about the rationale and statistical background of model-based recursive partitioning in general and in particular with extensions to psychometric models for paired comparisons as well as item response models. In this context, the data-driven approach of model-based recursive partitioning proves to be particularly suited for detecting violations of homogeneity or invariance, such as differential item functioning, where we usually have no a priori hypotheses about the underlying group structure.

Michel Regenwetter

The geometry of probabilistic choice induced by heterogeneous hypothetical constructs and/or error-prone responses

Michel Regenwetter
University of Illinois
at Urbana-Champaign

You can find the video of the keynote here.

Abstract

Heterogeneity of choice behavior has many potential sources: Different people may vary in the underlying hypothetical construct (e.g., preferences in decision making), a given individual may fluctuate in the hypothetical construct or be uncertain about it (e.g., uncertain preferences). Even for a given, fixed, latent state of a construct, overt behavior may vary due to probabilistic errors in responses. It is plausible that much behavior inside and outside the lab combines all of these sources of heterogeneity. This talk reviews a general geometric framework through which we can compare the parameter spaces induced by different types and sources of heterogeneity. This framework makes it possible to diagnose whether heterogeneity in observed behavior is due to heterogeneity in hypothetical constructs or in overt response processes.