Abstract
This paper develops and illustrates a comprehensive method for analyzing the endorsements to test items (called CATIV). It (1) accounts for whatever kind of variance is present; (2) is analogous to analysis of variance, but is applied to test-item responses; (3) can analyze many more variables than analysis of variance; (4) shows which test items are additive and which are configural; (5) includes a criterion for identifying test-item answers which belong to a universe of answers other than the one dominant in the study; (6) selects patterns of answers which yield substantive types; (7) assesses the relative dependability of the types which are isolated; (8) isolates all substantive types present in the data; and (9) can be applied by pencil and paper to a column matrix which shows the number of endorsements of every answer to every item of a test, provided these are arranged in order from the most popular to the least popular answer in terms of number of endorsements.
Get full access to this article
View all access options for this article.