Sunday, January 01, 2006

st: SSC > New module -hcavar-

Thanks to Kit Baum, there is now a new module -hcavar- available for download from SSC.

This module perfoms a Hierarchical Clusters Analysis (HCA) on a list of variables (and not on the observations !!). If the variables are numerous or ordinal variables, a transformation of the correlation matrix is used as distance matrix. If the variables are binary variables, this is possible to use other coefficients of dissimilarity : Jaccard, Dice, Russel or Matching (or pearson). In the field of the Item Response Theory (IRT), this is possible to realize with this module a HCA on conditional measures of proximity (HCACCPROX method): conditional covariances, conditional correlations or Mantel Hanszel Measures of dissimilarity. The *matrix* option allows a specific matrix of distances between the variables to be given.

The module produces a dendrogram and displays the compositions of the clusters indicated by the *partition* option. In the field of IRT, the *detect* option allows computing the DETECT, Iss and R indexes.

The module requieres the -detect- and -genscore- modules be installed (available on SCC).

The -hcavar- module replaces and corrects the old -hcaccprox- module. This new module is more powerfull because it allows ordinal or numerous variables to be used.

More details about this module (and my other modules) on my website:

Happy new year 2006 to all the members of statalist !

Jean-Benoit Hardouin

-- *************************************************************************** Jean-Benoit Hardouin PhD in Biostatistics Regional Health Observatory - Orléans - France Email : **************************************************************************

* * For searches and help try: * * *


Links to this post:

Create a Link

<< Home

This page is powered by Blogger. Isn't yours?