Version 1.5 of the MiscPsycho package had been uploaded to CRAN (should hit
mirrors in a day or so). This package has a set of functions that may be useful
for psychometric applications.
The package has been updated to include the following:
1) All functions (where appropriate) now use standard formula arguments
2) All functions now use S3 print and summary methods
3) Help files have been substantially improved
4) Where appropriate, functions now use the standard extractor functions, coef()
4) The vignette 'MP' now provides very comprehensive descriptions of the
statistical methods and substantive examples for
all functions
5) alpha.Summary has been deprecated. The alpha function and its summary method
now includes the "conditional alpha" that was
previously provided by alpha.Summary.
6) The following new methods have been added and tested with operational data:
* Cheat function now provides the user with choice of newton-raphson iterations,
bisection method, or Rs internal uniroot function
* A new function called "SSI" implements a K nearest neighbor
algorithm to derive conditional norms for student achievement data. This was
designed primarily to construct conditional norms for student achievement growth
models, but is flexible such that conditional norms for any score can be
constructed
* Objects of class jml can now be plotted to visually examine IRT data fit by
the jml function
* the function "classical" now returns standard errors of the item
p-values as well as design-consistent standard errors to reflect clustered
samples
All functions have been rigorously unit tested by comparing output to known
answers, results from other software programs, and through monte carlo
simulations. However, should an error or bug arise, please let me know. Comments
regarding R structure and general implementation are always appreciated.
Sincerely,
Harold Doran
[[alternative HTML version deleted]]
