Martin Jelínek, Petr Květon, Dalibor Vobořil

The concepts of item response theory and computer adaptive testing have been mentioned very often in recent years in connection with modern trends in psychological (and other) diagnostics. The ambition of this book is to provide the reader with a comprehensive description of the basic principles of Item Response Theory (IRT) and Computerized Adaptive Testing (CAT), which is based on it. Item response theory represents a complex mathematical apparatus that tries to capture the situation when an individual responds to a test item. Thanks to this apparatus, we can predict how an individual with a certain level of ability will respond to a specific item. The basis of IRT is a mathematical model that indicates the probability of a certain answer depending on the level of the individual's latent trait and the characteristics of a specific item. Each item is described by the so-called characteristic functions of the Item Characteristic Function (ICF), which expresses the relationship between a latent trait and the probability of a given answer. Based on the defined relationship between the latent trait and the probability of a given response, the concepts of conditional measurement error and information gain are derived, which are applied in a wide range of psychometric applications.

Item response theory is an alternative to classical test theory (CTT). Model-grounded item response theory brings a number of advantages to the measurement process. IRT considers a test as a set of separate items, and the basic unit for which different models are created is the individual item. In contrast, in the CTT, an item is not conceptually separable from the whole of a particular test. Especially thanks to the isolation of the test item, IRT offers a suitable apparatus for applying the general principles of the so-called adaptive testing. The basic idea of ​​adaptive testing is simple: ask the tested person to solve only those items that allow an ideal assessment of the level of the measured trait. In practice, its application requires either an experienced administrator who is able to estimate the level of the tested person's traits, choose a reasonably difficult item and repeat the entire process after answering it, or the use of a mathematical apparatus that would do the same in real time - usually IRT in combination with computer technology. Adaptive testing is nowadays mostly understood as computer adaptive testing based on the principles of IRT.

The reader will gradually be introduced to the basic IRT models for items of dichotomous and polytomous format, the assumptions of these models, further to the ways in which it is possible to arrive at estimates of the parameters of items and persons on the latent trait scale, with the possibilities of their transformation and the assessment of the suitability of IRT models with respect to dates. Attention is still devoted to practical applications of IRT, such as recognizing the different functioning of items in different groups of respondents and using the concepts of informational contribution for the construction of test methods. The theoretical description of IRT aspects is preliminarily supplemented by practical examples on real data, including specific tips for working with the renowned IRT software Bilog-MG, Parscale and Multilog. In addition, this part of the book is accompanied by online material, which contains calculation templates implemented in a spreadsheet, which will help a better understanding of calculation problems and represent a useful aid for the reader to perform his own analyses. In the text, the templates are marked with a serial number and a QR code, in the appendix we present the corresponding list of Internet addresses.

The second part of the book is focused on a thorough introduction to the principles of adaptive testing in a historical perspective and on the current concept of adaptive testing, where computer adaptive testing based on IRT is synonymous with adaptive testing. The chapters on computer adaptive testing are based on the authors' experiences with creating their own CAT software. This software is also used to demonstrate the advantages and problems that adaptive testing brings to diagnostic practice.

Size: 14×21 cm 160 pages ISBN: 978-80-247-3515-3 Catalogue. no. books: 2189