As raw scores obtained from the instruments used in assessments are not amenable to statistical analysis or the provision of valid and reliable comparisons across students, schools, states or countries and over time, most LSAs commonly use item response models in the scaling of cognitive data. Raymond Adams and Alla Berezner describe and illustrate three reasons for using IRT in this chapter. These include that IRT models (i) support the process of test development and construct validation, (ii) facilitate the usage of the tests consisting of a number of rotated test forms within one assessment to increase content coverage and (iii) enable the maintenance of scales that are comparable across countries and over time when used in conjunction with multiple imputation methodology.