ABSTRACT
Computer-administered and online assessments have the potential to be both desirable and dangerous. They have key advantages; at the same time, they have challenges regarding validity and ethics. Best practices for computer-administered and online assessments are provided. Adaptive testing involves having the respondent complete only those items that are needed to answer an assessment question, which can save immense time without sacrificing validity (if done well). There are many approaches to adaptive testing, including manual administration—such as skip rules, basal and ceiling criteria, and the countdown approach—and computerized adaptive testing (CAT) using item response theory (IRT). A CAT is designed to locate a person's level on the construct with as few items as possible. A CAT administers the items that will provide the most information based on participants' previous responses. CAT typically results in the greatest item savings—around 50% item savings or more.
Well-designed CATs show equivalent reliability and validity to their full-scale counterparts. By contrast, many short forms are not as accurate as their full-scale counterparts.
