[R-pkgs] rockchalk_1.5.4 posted

Paul Johnson pauljohn at ku.edu
Mon Apr 9 23:31:42 CEST 2012


Greetings:

rockchalk is a collection of functions to facilitate presentation of regression models.
It includes some functions that I have been circulating for quite some time (such as
"outreg") as well as several others. The main aim is to allow people who do not
understand very much R to survive a course in intermediate regression analysis.
The examples included with the functions include more than the usual amount of detail.

This version features

1) a full vignette called "rockchalk" that illustrates many of the
functions in the package.  The vignette includes an explanation of why mean-centering
and residual-centering do not help with the "inessential multicollinearity" problem
that concerns users of regression models in which there are interactions or
squared terms.

2) a function "summarize" that is intended to remedy some of the shortcomings I
perceive in R's summary function.

The package is not coordinated with any particular textbook and should help in
any class in which students are expected to estimate regression models, summarize
them in professional-looking tables, and conduct various follow up diagnostics.

Highlights:

plotSlopes & testSlopes: draw "simple slopes" graphs and conduct hypothesis tests
for moderator variables.

plotCurves: an attempt to provide a "termplot" replacement that allows plots that
incorporate mediators.

plotPlane: a wrapper for persp to plot regressions in 3d (similar to scatterplot3d)

standardize, meanCenter, residualCenter: after fitting a non-centered model, use these
to experiment with the benefits (or lack thereof) that flow from centering.

-- 
Paul E. Johnson			email: pauljohn at ku.edu
http://pj.freefaculty.org	Assoc. Director
Professor, Political Science 	Center for Research Methods and Data Analysis
1541 Lilac Lane, Rm 504		1425 Jayhawk Blvd.	
University of Kansas		Watson Library, Rm. 470  	
Lawrence, Kansas 66045-3129	Lawrence, Kansas 66045-7555
Ph: (785) 864-3523		Ph: (785) 864-3353



More information about the R-packages mailing list