In the calculation of total item scores for the matching criterion, not-reached, off-task, and omitted items were considered to be wrong responses. Polytomous items were weighted more heavily in the formation of the matching criterion, proportional to the number of score categories. For each item, calculation of the Mantel-Haenszel statistic did not include data from examinees who did not reach the item in question.
Each DIF analysis was a two-step process. In the initial phase, total item scores were formed and the calculation of DIF indices was completed. Before the second phase, the matching criterion was refined by removing all identified C or CC items, if any, from the total item score. The revised score was used in the final calculation of all DIF indices. Note that when analyzing an item classified as C or CC in the initial phase, that item score is added back into the total score for the analysis of that item only.