1989 Volume 12 Issue 4 Pages 129-137
Desiring to clarify the distribution of computerized test response times, the authors first equalized, as response curve data, student response times collected individually during the administration of computerized tests, then attempted a statistical analysis using a Gamma distribution. They found that applying a Gamma distribution, which has fewer parameters, not only adequately but also more clearly describes the distribution of these response times than does the here-to-for more commonly used Weibull distribution model. Using a Gamma distribution model, the authors propose a method for using response time data in relation to test item analysis. They also show several examples involving the prediction and use of test response time distribution data. Furthermore, in support of item bank utilization in the construction of tests, the authors developed a method for using estimated curves resulting from their methods to determine the percentage of students who would not be able to complete specific items within pre-designated time limits.
