Abstract:
The main purposes of this study were 1) to analyze the items biased against sexes and geographical regions of the English Entrance Examinations, Form AB's and ABC's in 1988-1990 by Delta-Plot Method. Chi-square Method and Three-parameter Logistic Method, and 2) to compare the numbers of biased item results from the 3 mentioned methods. The populations of this study were 6 groups of approximately 30,000-80,000 actual testes and each was subdivided according to sexes and their geographical regions. They were then sampled by a simple sampling technique with 2 conditions: equal numbers of males and females and not more than 3,000 in each subgroup. The subjects in each subgroup were then ranging from 424 t0 3,000 which were more than optimal sizes. The instruments used were 600 items of the 1988-1990 English Entrance Examinations, Form AB's and ABC's. The data were analyzed by means of Classical Item Analysis, Logistic Item Analysis., ICC test, Factor Analysis, Chi-square test, Bias test and F-test. The findings can be summarized as follows: 1.On average, there were approximately 5-35 items biased against sexes in each form of the examination (approx. 5%-35%) depending on methods of analyses. 2. On average, there were approximately 8-36 items biased against geographical regions in each form of the examinations (approx. 8%-36%) depending on methods of analyses. 3. Generally speaking, the examination in 1988-1990 biased against the testees from 4 geographical regions more than those from the central region approximately 3-4 times. There was a tendency that they biased against males more than females. 4. The numbers of biased items results from the 3 methods were significantly different and correlated each other insignificantly. Three-parameter Logistic Method when excluded low biases was the most sensitive method. It detected the highest numbers of biases items and the next one was Chi-Square Method. Delta-Plot Method was a very conservative method. It detected biased items the least