| Rascal - User-Friendly Rasch Calibration |
Rascal is also available as part of the ASC Item and Test Analysis Package.
Rascal analyzes data files (ASCII format) of test item
responses produced by optical mark readers (scanners) or by
manual data entry — the same format used by Iteman 4 and Xcalibre
— to estimate the item difficulty and person (ability)
parameters based on the one-parameter (Rasch) logistic IRT model
for dichotomous data. Rascal can center the scale of the
parameter estimates on difficulty (i.e., a true Rasch scale) or
on ability (a three-parameter IRT model with fixed discrimination
and zero guessing). Rascal automatically transforms the IRT score
scale (in a linear fashion) to make your score reports more
interpretable. The software can also “fix” certain item
parameters to specified values and automatically calibrate the
remaining items onto that scale. This feature can be used for
linking tests and item pools.
Rascal Features
For each item, Rascal estimates an item difficulty parameter,
a Pearson chi-square fit statistic, and the standard error
associated with the difficulty estimate. Rascal also generates a
table for converting number-correct (raw) scores into IRT
(ability) scores. The score frequency table shows, for each
maximum-likelihood estimate of ability, its standard error, the
frequency and cumulative frequency, percentile, and the
transformed scaled score. Rascal can also record
maximum-likelihood (IRT) scores for each examinee to an external
text file that can be used by other statistical packages.
Rascal provides a graphical “items-by-persons” map,
displaying a side-by-side comparison of the distributions of item
difficulties and person abilities. Also provided are graphical
displays of the IRT-based test response function (test
characteristic curve) and test information function.
Specifications
Rascal can analyze up to 750 items with no limit on the number of examinees.
Sample Output
The following links are to sample files used and produced by
Rascal . The input data sets required by any of the Assessment
Systems analysis programs is a very straightforward ASCII (text)
format. The item responses may be either alphabetic or numeric.
Data files may be created either through the use of data entry or
produced from an optical mark reader (scanner). Note that the analysis program does not
“drive” the optical scanner - rather it works with
the output produced by the scanner. We have worked with
individuals who are using scanners from most of the major vendors
without any real difficulties.
This example is typical of a multiple-choice test with all of
the items on a single scale. The data set consists of 400
examinees responding to a test containing 30 items. Options for
this run were set to center the resulting item response theory
(IRT) scale on the difficulty parameters (i.e., a true Rasch
scale). Click on the following for the , , and
the optional file. The optional statistics file is simply a listing of the item parameter for each item, one per line.
Books on Rasch Models and Related Methods
System Requirements
Windows 2000, XP, Vista, Windows 7
Downloadable Manual
Demonstration Version
- This demo version of Rascal is designed to run only the sample input data files that are provided with it. You can vary the options available in Rascal with these data files, but cannot analyze other data files.
- Note: If you attempt to use the demo version of Rascal to analyze other data
files, it will give you incorrect results.
Products
Your Account
Your cart
