ABSTRACT
Item response theory (IRT) is sometimes presented as an alternative to the traditional true score theory (some of which was described in the previ ous chapter). However, it is more realistic in applications involving large scale test development and scoring to consider the traditional theory and IRT to be complementary: Practical test construction using IRT typically includes reference to traditional statistics as well as the item parameters and goodness-of-fit statistics of IRT. Both the traditional theory and IRT provide means to evaluate and report the precision of test scores, usually in the form of standard errors of measurement. In addition, although IRT need not make use of the summed score that is the central feature of the traditional theory, IRT's treatment of summed scores can be informative. In subsequent chapters, we mix and match aspects of the traditional the ory and IRT rather freely.
