ABSTRACT
Advances in psychometric theory over the last 30 years have introduced many new tools and techniques to researchers interested in measuring psychological constructs. The revolution of item response theory (IRT) has raised questions about the relevance of its predecessor, classical test theory (CTT). In fact, some writers have suggested that CTT has been made obsolete by its successor. For example, Rojas Tejada and Lozano Rojas (2005) discussed how recent research has been used to “displace the CTT in favour of the use of Item Response Theory-based models” (p. 370), and Harvey and Hammer (1999) predicted that “IRT-based methods . . . will largely replace CTT-based methods over the coming years” (p. 354). Samejima, in critiquing CTT, describes its “ fatal deficiency [italics added],” which relates to how CTT models measurement precision (Samejima, 1977, p. 196). Borsboom argues that “few, if any, researchers in psychology conceive of psychological constructs in a way that would justify the use of classical test theory as an appropriate measurement model” (Borsboom, 2005, p. 47). We have heard people dismiss CTT as irrelevant and antiquated, more worthy of history books than contemporary psychometric classes. Often these same individuals treat IRT as a panacea for all psychometric woes. In short, CTT is treated as an old racehorse that is nice to have around, though everyone is expecting it to perish soon. According to this argument, IRT is the new steed that has won a few races and is expected to abolish its predecessor’s triumphs. We believe that this urban legend is just plain myth and
that CTT still has uses in modern psychometrics. Having said that, like most urban legends, there is a kernel of truth to the reported obsolescence of CTT. In some cases and applications, CTT has been supplanted by IRT. In this chapter, we will sort out fact from fiction and provide a psychometric road map for people trying to navigate this confusing literature.
