This post is provide by guest blogger, Dianna Doyle, graduate student at the University of St. Francis in Joliet, Illinois, MS in Training and Development program.
According to Phillips (2018) test questions can have a huge impact on the reputation and efficacy of a trainer and program. Questions that are too easy or give embedded clues are easily passed but when the test taker does not really measure up to expectations of competency the validity of instruction is questioned. Questions that are too difficult where few pass the test may lead to the shut down of programs since it would appear no learning actually took place even though time and resources were invested into the instruction. An instructor may know the content of a subject but may have very little to no training in creating test questions that truly measure if and what training took place.
Phillips (2018) recommends using Criterion-Referenced Test Development written by instructional design professors Sharon Shrock and William C. Coscarelli. Three test item statistics are discussed to evaluate the quality of test items: 1) difficulty index, 2) p-values, and 3) point-biserial correlation. To apply these statistics, you first need to create your test and then administer it to a group of at least 25-30 program participants.
The article defines and discusses each of these test item statistics as well as how to go about developing a quality test that measures up statistically as valid. This article and recommended resources would be an excellent tool for test designers and e-learning professors. This information should be incorporated into all programs training trainers and instructors. A quality measurement of such a course should include creating a test that meets the criterion outlined.
Phillips, K. (2018, August 1). Write test questions that actually measure something.
Association for Talent Development. Retrieved from https://www.td.org/insights/write-