Comparing the efficiency of composite scores constructing methods in the Test Battery by Utilization of conditional standard error of measurement
Subject Areas :
مجتبی جهانی فر
1
,
حجت اله درفش
2
1 - عضو هیات علمی دانشگاه شهید چمران اهواز، اهواز، ایران
2 - عضو هیات علمی دانشگاه شهید چمران اهواز، اهواز، ایران
Keywords: Test Battery, Composite score, Scale score, Weighting scheme, CSEM,
Abstract :
The test is used for decision-making in education. Decision that is taken by results of a test raises sensitivity. Admission decision is taken according to score that is acquired in test. If the test is composed of several sub-tests with different content is named Test Battery and the resulting score is called composite score. Different methods can be designed to construct composite scores, and based on the composite scores, admission decision is made toward accepting or not accepting inviduals. But which constructing technique can be design with more precision and less error? So, it is more efficient than other composite score constructing techniques. This study aimed to compare the effectiveness of ways to make the composite score that is used in the Test Battery was conducted. 10,000 random real samples from Iran universities entrance exam in seven sub-tests have used to compare efficiency of 6 methods that was designed to construct composite scores. Raw scores obtained from correct answers, so Normalizing and Arcsine methods was used to construct scale scores. To reduce irregularity in the distribution of scores, kernel smoothing methods and to make composite scores, nominal and effective weighting schemes were used. To evaluate the effectiveness of the composite scores constructing methods, the conditional standard error of measurement was used. Results showed that composite scores that used Arcsine scale scores or normalization scale score with pre-smoothing had less average conditional standard error of measurement and more efficiency in compare to other methods
_||_
