Introduction

Welcome to the Open Source Computerized Adaptive Testing System! OSCATS aims to be a fully featured engine for writing computerized adaptive testing (CAT) simulations and can be used as a back-end for operational CAT. OSCATS supports both the Item Response Theory (IRT) and cognitively diagnosis/latent classification psychometric frameworks, including algorithms for dual-purpose measurement. The core CAT administration routine includes eight points for algorithmic customization, and OSCATS comes ready to be extended with new psychometric models, item selection algorithms, and statistical calculations.

An OSCATS test is formed by supplying an item bank and a set of algorithms for item selection, item administration or simulation, and statistical calculations, such as estimating an examinee's latent ability or class and tracking item exposure. These algorithms are OscatsAlgorithm objects that attach callback functions to one or more of eight signals emitted by the OscatsTest test administration loop. Items in the bank have a psychometric model in either the IRT or latent classification framework (or both). These models are objects subclassed from the abstract OscatsContModel or OscatsDiscrModel class. Using new models or algorithms in OSCATS is as simple as writing a new subclass with the desired functionality.

To get started with OSCATS, it's best to read the documentation for the test administration loop in oscats_test_administer() and the OscatsTest signals. Then, take a look at a few of the example programs provided with OSCATS. For the specific syntax and semantics of each object, refer to the complete API documentation.

If you have questions about OSCATS, please do not hesitate to ask them on the OSCATS mailing list. And, if you find OSCATS useful in developing your CAT or conducting CAT research, let us know: We would love to include a reference to your work on the OSCATS website.