Title: R-Package cmm: Categorical Marginal Models

Author: Andries van der Ark

Affiliation: Tilburg University

Abstract: Categorical marginal models (CMMs) are a class of robust and
flexible statistical models that can handle complex dependency structures in
the data without the requirement of additional distributional assumptions.
CMMs are so general that they can solve a sheer infinite number of
data-analysis problems. Despite these attractive properties, CMMs are seldom
used because they are difficult to understand, because there is no
user-friendly software available, and because-until recently-CMMs could be
applied to small data sets only.

We contributed the R-package cmm (Bergsma & Van der Ark, 2009), which provides
ML estimates, standard errors, and fit statistics for CMMs. The package is
flexible in the sense that-as far as we know-any CMM can be estimated. Due to
its flexibility the package is not very user-friendly, which has resulted in
only four citations (including two self-citations) in the last five years on
Google Scholar. Our main challenge is to redevelop the R package into a
user-friendly tool that will boost the use of CMMs. 

In the workshop, I will give a short introduction to CMMs, demonstrate the R
package, and I discuss the ideas for improving the package.