A comparison of examinees' abilities, answer-changing characteristics and time spent in computerized adaptive testing under varying testing conditions and examinees' ability levels
Abstract:
The purpose of this research were to develop the computer program of adaptive test of the comprehension of fraction of Prathomsuksa 6 students. The testing contained two conditions; i.e., review allowed within successive 5-item blocks and no item review allowed and to compare the difference of examinees' ability levels, answer changing characteristics and the time spent in computerized adaptive testing under the various levels of examinees' abilities. The samples consisted of 200 Prathomsuksa 6 students in 2006 academic year from Wat Plubplachai School. The computerized adaptive testing program was used to collect the data. The one-way analysis of variance and the Scheffe or Dunnetts T3 method for post hoc analysis were employed to analyzed the collected data. The findings were as follows: 1. There was a significance at .05 level on examinees ability levels among various ability levels of examinees. After changing the test condition from non-review test to review test, the low-ability examinees received the most different ability more than the medium-ability examinees and the high-ability examinees by order. 2. There was no significance at .05 level on retaining the same answer in answer-changing characteristic among various ability levels of examinees under the review test condition. 3. There was a significance at .05 level on changing the answer; wrong to right, wrong to wrong, right to wrong, in answer-changing characteristic among various ability levels of examinees under the review test condition. The medium-ability examinees and the high-ability examinees had he number of changing answer from wrong to right more than the low-ability examinees. The low-ability examinees had the number of changing answer from wrong to wrong and from right to wring more than the high-ability examinees and the medium-ability examinees. 4. There was no significance at .05 level on the different time spent among various ability levels of examinees under the review test condition.