Document Type

Article

Publication Date

7-2-2012

Abstract/ Summary

There are two approaches to psychometrics. Classical test theory is the traditional approach, focusing on test-retest reliability, internal consistency, various forms of validity, and normative data and standardization. Modern test theory or item response theory (IRT) focuses on how specific test items function in assessing constructs. IRT makes it possible to scale test items for difficulty, to design parallel forms of tests, and to provide for adaptive computerized testing (DeMars, 2010). “(T)he basic concepts of item response theory rest upon the individual items of a test rather than upon some aggregate of the item responses such as a test score” (Baker, 1985/2001, p. 6). Using IRT methodology in data analysis can be challenging because “IRT programs are still much more traditional and ‘user-unfriendly’ than many commercially-available statistical packages” (Kline, 2005, p. 107). This paper outlines some of the basic procedures involved in using two representative programs: MULTILOG (du Toit, 2003; Thissen, Chen, & Bock, 2003) and PARSCALE (du Toit, 2003; Muraki & Bock, 2003). A third program, WINSTEPS (Bond & Fox, 2007; Linacre, 2006), is also noted briefly. Also provided is some of the essential background material on IRT theory and rationale, and on its requirements and assumptions. Readers are encouraged to consult the software manuals, books, chapters, and articles in the reference list for more detailed information on technique and for authoritative and definitive theoretical coverage.

Version

pre-print (i.e. pre-refereeing)