A comparison of predictive validity of selected components by using accumulative average score equating and previous components of central university admission system
Abstract:
The purpose of this study were 1) to study the correlation between Cumulative Grade Point Average, Grade Point Average, ONET, ANET and Undergraduate Achievement. 2) to equate Cumulative Grade Point Average and Grade Point Average by using IRT method and 3) analyze and compare predictive validities of the selected components which were equated the accumulative average score and the current components which were not equated of the accumulative average score. The sample of this study was 2,200 freshmen, 2006 academic year of the public universities in Bangkok metropolitan area, which were Chulalongkorn University, Kasetsart University, Thammasat University, Srinakharinwirot University, and Silpakorn University. The sample used in this research were from 5 fields of study, ie. Health science, Physical science, Engineering. Architecture, and Accountancy. The data were the components of central university admission system and undergraduate achievement. The results were as follow: 1. The components of central university admission system. ie. Cumulative Grade Point Average, Grade Point Average, ONET and ANET were correlated with the undergraduate achievement at .05 level of significance in all fields of studies except Accountancy, whose ANET was not correlated with the undergraduate achievement at .05 level of statistical significance. 2. The result of accumulative average score equating was lower than previous accumulative average score all field of studies. 3. According to the result of the analysis and comparison of predictive validities, it was found that on the field of Health Science, Physical science and Engineering, the selected components 6-10 could be used instead of previous components. In the field of Architect,the selected components 6,8,10 could be used instead of previous components .As for Accountancy, the predictive validities of the selected components 1-15 were not different at .05 level of statistical significance.