02120nas a2200193 4500008004100000245007900041210006900120260001200189520146800201653002801669653000801697653001801705100002001723700001701743700002301760700002301783700002301806856009701829 2011 eng d00aThe Use of Decision Trees for Adaptive Item Selection and Score Estimation0 aUse of Decision Trees for Adaptive Item Selection and Score Esti c10/20113 a
Conducted post-hoc simulations comparing the relative efficiency, and precision of decision trees (using CHAID and CART) vs. IRT-based CAT.
Conclusions
Decision tree methods were more efficient than CAT
But,...
Conclusions
CAT selects items based on two criteria: Item location relative to current estimate of theta, Item discrimination
Decision Trees select items that best discriminate between groups defined by the total score.
CAT is optimal only when trait level is well estimated.
Findings suggest that combining decision tree followed by CAT item selection may be advantageous.
This simulation study sought to compare four different computerized adaptive testing (CAT) content-balancing procedures designed for use in a multidimensional assessment with respect to measurement precision, symptom severity classification, validity of clinical diagnostic recommendations, and sensitivity to atypical responding. The four content-balancing procedures were (a) no content balancing, (b) screener-based, (c) mixed (screener plus content balancing), and (d) full content balancing. In full content balancing and in mixed content balancing following administration of the screener items, item selection was based on (a) whether the target number of items for the item’s subscale was reached and (b) the item’s information function. Mixed and full content balancing provided the best representation of items from each of the main subscales of the Internal Mental Distress Scale. These procedures also resulted in higher CAT to full-scale correlations for the Trauma and Homicidal/Suicidal Thought subscales and improved detection of atypical responding.
1 aRiley, Barth, B1 aDennis, Michael, L1 aConrad, Kendon, J uhttp://apm.sagepub.com/content/34/6/410.abstract