You are correct CJ, but that is ok for my application as it is returning the data I need. I'm building a test item analysis db (analyzing results from individual questions); multiple exams with multiple versions and revisions for multiple courses (over a hundred courses) with different amount of participants. The value I need for statistical calculation is simply the amount of questions on an exam, for which I reference the examination result data (a comparison between individual participant responses and the answer key - returns a 1 for correct and 0 for incorrect). The examinations can and will have different amount of questions (from 10 to 50).
Once an exam answer key is entered (by course, version, and revision), the data entry clerk can select the course, exam version, exam revision, course date, and instructor followed by the participant results. A report can then be generated to view the statistical analysis of individual examination items (questions) and the selected answers (mean, standard deviation, KR-20, standard error of measurement, min/max discriminating power, etc). Exam date can be used as filters to analyze a particular exam for one class or many (increasing statistical accuracy). Same with the instructor.