WinMira software description
WINMIRA is a standalone software for estimating and testing a large number of discrete mixture models for categorical variables. Models with a nominal as well as continuous latent variables, and combinations of both, can be estimated with the software. WINMIRA 2001 can be used for analyses with the Latent Class Analysis (LCA), with the Rasch model (RM), with the Mixed Rasch model (MRM) and with Hybrid models (HYBRID) for dichotomous and polytomous data.
WinMira new features
- Improved estimation algorithms Estimation of the polytomous Rasch models is speeded up and improved. A completely rewritten algorithm ensures increased convergence speed.
- User friendly help: the online help system was completely rewritten and includes a detailed description of all new features. See the following example sections for an impression of some new features and the corresponding help system files.
- Full SPSS support, it reads and writes data directly in SPSS file format. Person parameter estimates and standard errors, posterior probabilities of class-membership (in the case of mixture models) as well as person-fit statistics can now be appended directly to the SPSS data-file.
- In addition, EXCEL and other spreadsheet data files can be imported and exported by means of using the 'save as' option to generate tab-delimited files. parameter constraints have been improved and extended. Logistic parameter constraints (i.e., item parameters (for dichotomous models), or item locations as well as threshold distances (for polytomous models) can be imposed both within and between classes. Alternatively, category probabilities can also be constrained.
- More documentation All features of the software are outlined in a 140 page manual that is included in the registered version of WINMIRA 2001. Depending on the license purchased, the latest versio of the manual is delivered in high quality printable electronic form (PDF and PostScript) and is optionally accompanied by a printed manual.
For polytomous data, WINMIRA is capable of estimating four different submodels, i.e., the partial credit model, the rating scale model, the equidistance model and the dispersion model. Moreover, it is possible to estimate models for data with a different number of categories for each item. WINMIRA can be used for scale construction, i.e., item selection, and/or classification of subjects into homogenous populations, e.g. for analyzing different solution styles in assessments, or for identifying aberrant response patterns (e.g. as a result of cheating or pre-existing knowledge about the examination).
Ordering
For price information and ordering, please visit the Prices and Ordering Page
