Harpp et al. developed a test statistic having a Gaussian distribution, and suggested that those student pairs with a test statistic at or greater than five standard deviations could be said to significantly deviate from the norm. (The probability of obtaining a z-score of 5.0 or more in a normal distribution is about 0.0000003.) In other words, Harpp and Hogan (1993) devised a procedure for testing the hypothesis that the item responses given by a pair of students were statistically unrelated, or independent of each other; if they were, the value of their test statistic would be less than 5.0, and the hypothesis would be accepted. A test statistic of 5.0 or more would result in the hypothesis being rejected; in this case, we would have strong statistical evidence to suspect that the similarities found in the students’ item responses were beyond what we’d expect by chance, pointing to the possibility of cheating.
Harpp et al. developed a test statistic having a Gaussian distribution, and suggested that those student pairs with a test statistic at or greater than five standard deviations could be said to significantly deviate from the norm. (The probability of obtaining a z-score of 5.0 or more in a normal distribution is about 0.0000003.) In other words, Harpp and Hogan (1993) devised a procedure for testing the hypothesis that the item responses given by a pair of students were statistically unrelated, or independent of each other; if they were, the value of their test statistic would be less than 5.0, and the hypothesis would be accepted. A test statistic of 5.0 or more would result in the hypothesis being rejected; in this case, we would have strong statistical evidence to suspect that the similarities found in the students’ item responses were beyond what we’d expect by chance, pointing to the possibility of cheating.
การแปล กรุณารอสักครู่..