ANNOUNCEMENT: January 26, 2004 - StatItem is now part of Simstat 2.5. It is no longer sold separately.
Designed and written by Normand Péladeau
SIMSTAT already provides several useful tools for scale development and testing such as inter-rater agreement analysis, scale reliability and sensitivity (ROC curves) analysis. STATITEM is a module for scale development and testing that allows one to perform classical item analysis on multiple-choice item questionnaires.
The program can compute, for each item in the scale, various statistics to assess their discrimination value and their contribution to the overall internal consistency the scale. STATITEM can also be used to improve the reliability and validity of a test by identifying items that should be eliminated, substituted or revised. In doing so, item analysis makes it possible to increase the overall quality of a scale while shortening it, either by eliminating unsatisfactory items or by removing redundant ones.
- Can perform item analysis on questionnaires with up to 2000 items.
- Exclude cases containing missing values or to treat those responses as incorrect.
- Item statistics (frequency & percentage of endorsement, item-total biserial & point-biserial correlation, Cronbach's alpha if item deleted, user defined discrimination index).
- Item-total endorsement rate matrix and item characteristic curves (based on original scores or smoothed data)
- Item-total response breakdown to compare the percentage of correct responses among several groups of equal size (between 2 and 10 groups)
- Scale total statistics (mean, median, minimum and maximum, standard deviation, standard error, skewness, kurtosis, Cronbach's alpha, Ferguson discrimination index, etc)
- Summary statistics on percent correct, biserial and point-biserial correlations.
- Frequency distribution of total scores.
- Option to save scale totals on disk.