[R] recommended package/docs for analyzing multiple choice tests

Doug Holton doug00 at gmail.com
Wed Sep 19 02:36:14 CEST 2007


Hi,
  What package would you recommend for analyzing the
validity/reliability of multiple choice tests.  Doing things such as
classical test analysis, factor analysis, item response theory.

I've used psychometric (item.exam), MiscPsycho (alpha.Summary), and ltm
(rcor.test).  MiscPsycho reported the numbers most similar to what I get
in SPSS: corrected point biserial correlations, cronbach's alpha.  I
didn't understand what the psychometric package meant by its
"discrimination" and "item reliability" numbers output by the item.exam
function.  Perhaps the former is uncorrected point biserial
correlations?  They were higher values.  I downloaded and inspected
the source code for both packages.  It was hard to understand what the
functions were doing without some comments in the code.

Also, would you recommend a book or resource with examples of using R
for test analysis.

Thank you, it's been great learning R,
-Doug



More information about the R-help mailing list