In grading my examinations, I use analytic scoring schemes (i.e., assigning points for individual steps). Moreover, scoring rubrics are used (Wilson & Onwuegbuzie, 1999). Here, the lowest score that students can achieve on an examination is 1 x the number of items on the examination form. For example, if an examination form contains 10 items and each item is measured on a 5-point Likert-type scale (e.g., 1 = strongly disagree that item answered adequately, 2 = disagree that item answered adequately, 3 = neutral, 4 = agree that item answered adequately, 5 = strongly agree that item answered adequately), then students' scores would range from 10 to 50, which could then be converted to a percentage by dividing the score by 50 and multiplying by 100. For this example, students would be expected to receive a rating of "4" or "5" for each item. I never "curve" examination scores because this is only justified if the score distribution pertaining to the class is normal, which is rarely the case. Indeed, at the graduate level, examination scores typically are negatively skewed. Rather than "curving," I conduct a classical test theory (CLT) item analysis, using CLT indices (e.g., item difficulty, item discrimination, point-biserial correlation) to determine whether any items reduce score reliability and/or score validity. Any misfitting items are then thrown out and do not form part of the final score. I use the mean of the weighted aggregate scores for each assignment to determine final student grades. As noted above, I never "curve" student scores.

See all answers to this question

See all of Tony's answers