Description of Terminology Used in the Test Statistics and Item Analysis Section of the Exam Scoring Report TEST STATISTICS Position Header ROW 1 ROW 2 ROW 3 Test Name Test Date Number of Examinees Number of Items Maximum Possible Points Highest Score Lowest Score Median Mean ROW 4 ROW 5 ROW 6 ROW 7 ROW 8 ROW 9 ROW 10 ROW 11 ROW 12 Page 1 of 2 Description The number of students taking the exam. The number of items included in the test. The maximum raw score value a student can receive. This value may exceed the number of item values if some items are ‘weighted’ more than 1. The highest score on the exam (percent). The lowest score on the exam (percent). The median of the scores earned by students taking the exam (percent). The mathematical average of the scores earned by students taking the exam (percent). Standard The standard deviation of the scores earned by students taking the Deviation exam. Test A measure of internal consistency of the exam. Can be taken as a lower Reliability bound estimate of the reliability of scores of the exam. Standard Provides an estimate of how far the true score may lie from the Error of observed score for an average examinee in the group. The standard Measurement error of measurement (SEM) is a function of the standard deviation of the score and the test reliability. In practice SEM is used to calculate the upper and lower limits of student’s score where his/her true score is assumed to be located. Revised 12/2004 ITEM STATISTICS Position Header Column 1 -----Item Column 2 – including headers which will read across the rows Overall Admins p rpb Description Item number. If you scan across this row, the correct option statistics are highlighted in bold. Total number answering item. Commonly known as “p-value”, which is the proportion of examinees answering the item correctly. The “p-value” is used as a difficulty index of the item. The higher the “p-value” the easier the item. The lowest possible “p-value” is 0 (very difficult; no examinees answer the item correctly) and the highest possible “p-value” is 1 (very easy item; everyone has it correct). A measure of correlation between a given item and the total score. It is a special case of the Pearson product moment correlation when one of the variables is dichotomous (takes a value either 0 or 1). Point-biserial is used as an index of item discrimination which is a characteristic of an item that addresses the ability of an item to measure sensitively individual differences in the domain measured by the exam. The size of the discrimination index indicates the relation of the item to the total domain of knowledge – as represented by the total score. Zero correlation indicates that the item does not relate at all to the total domain of knowledge – as represented by the total score. Meaning that examinees can have the item right (or wrong) regardless of their overall performance in the entire exam. Column 3 Column 4 Omits A (True) Column 5 Column 6 Column 7 Column 8 B (False) C D E Page 2 of 2 While a correlation of 1 indicates a perfect relation between the item and the total domain of knowledge measured by the test, a negative correlation indicates the reversal – which may be a result of an item being keyed incorrectly. Number of students who left this item blank. Corresponds with option A on the answer sheet. The headers listed above for column two also apply here: admis or number of students choosing this option, p or percent of students choosing this item, rpb or point-biserial of this item. Reminder that the correct answer statistics are highlighted in bold. So if option A is correct, all stats under this column are in bold. Corresponds with option B on the answer sheet. Corresponds with option C on the answer sheet. Corresponds with option D on the answer sheet. Corresponds with option E on the answer sheet. Pattern continues if using answer sheet with options A-J. Revised 12/2004