There are two statistics that are looked at: Point Biserial and P Value. The P Value stands for the percent correct or in other words how many students answered the question correctly. You may want to look at the questions where less than 50% of your students answered correctly. It is also important to note whether or not you have one correct answer and three plausible answers. If students are not picking a distractor it may be an indication that it is not a plausible answer. The Point Biserial is used to discriminate between the top 15 % of the class and the bottom 15 % of the class according to the Bell Curve. A Point Biserial of 0.2 is a good question as it discriminates between high and low performers. If you have a negative Point Biserial it might mean that the question doesn’t discriminate as everyone answered the question correctly. Look at the question to see if it is testing necessary knowledge. Questions that have a Point Biserial that is less that 0.2 need to be reviewed: Check the P Value, question wording and distractors before considering removing from the exam.
The other the statistic you may look at is the Kuder-Richardson which is used to measure the reliability of the exam. The reliability number ranges between 0.00 to 1.00 and that you would want a reliability of at least 0.7 .
You can listen to the podcast of this session.Nancy Anweiler podcast