Eight members of the ReSEP group and researchers from the Department of Economics at the University of Stellenbosch attended the Rasch Workshop that was led by Professor Alan Tennant and David Andrich at UCT from 21 to 25 January 2013.
The Rasch model is considered the simplest model in item response theory (IRT) and is also known as the 1 parameter logistic model. According to the Rasch model, an individual’s response to a binary item (i.e., true/false) is determined by the individual’s trait level and the difficulty of the item. Another way of expressing the Rasch model is in terms of the probability that an individual with a particular trait level will correctly answer an item that has a particular difficulty. In the education context an individual’s trait level would reflect his or her ability.
One particularly useful application of Rasch is to equate the difficulty level of tests. If some test items are common across two tests, this can be used to calibrate the difficulty level of items in both tests, and adjustments can be made to test performance levels to make them comparable. This technique is widely used in testing systems throughout the world, and is, for instance, very common in Australia.
Apart from its use in educational measurement, RASCH models are frequently used for psychological measurement and more recently also for market research.