Skip to main content icon/video/no-internet

An adaptive test is a test, or other psychological or educational measuring instrument, in which the items or questions administered to each examinee are dynamically selected for that examinee based on the responses during the test. Adaptive tests contrast with conventional fixed-form tests that administer a constant set of items to each examinee. Well-designed adaptive tests can allow the number of items, as well as item content, to vary across examinees while permitting constant measurement precision for all examinees; with conventional tests, item content and number of items are constant but measurement precision is permitted to vary across examinees.

Adaptive tests can be designed to measure continuous variables, such as ability, achievement, personality, and preferences, and other psychological and educational variables; they can also be used to make classifications, such as pass-fail, accept-reject, as well as multiple classifications (e.g., definite pass, probable pass, probable fail, definite fail). Adaptive tests can also be designed to measure individual change, including growth and decline. Adaptive tests achieve these differing objectives by varying the characteristics involved in adaptive test design and implementation. These design characteristics are as follows: (a) an item bank of items with known psychometric characteristics, (b) a rule for beginning the test, (c) an item selection rule, (d) a scoring procedure designed to place all examinees on the same scale even though they have answered different items, and (e) a procedure for terminating the test. Because implementation of some of these characteristics requires complex decisions and computations, most current adaptive tests are administered by computer and are called computerized adaptive tests (CATs). This entry focuses on the various types of CATs and issues associated with the implementation of CATs, including their use for educational testing.

Types

Alfred Binet’s IQ (intelligence quotient) test, first developed in the early 1900s, had all of the characteristics of an adaptive test: (a) the item bank for Binet’s IQ test (later known as the Stanford-Binet IQ test) was normed on children and young adults by placing items into levels based on the age at which the items were answered correctly by about 50% of a given age grouping; (b) Binet’s test began for each examinee based on the information that the psychologist had about the examinee obtained from a teacher or a parent; (c) items were administered and scored immediately by the examiner; (d) at the end of each age level, the examiner chose the next level to be administered based on the proportion correct; and (e) the test ended when the examiner had identified a ceiling level at which all items were answered correctly and a basal level at which all items were answered incorrectly. Between these two levels the examiner will have located the range of effective measurement for each examinee. The procedure is efficient as well as effective because it administers only items in that part of the larger item bank that are necessary to measure each individual.

Computerized adaptive tests extend the basic ideas of adaptive testing and generally combine them for maximum efficiency and effectiveness. Partially adaptive CATs are similar in some respects to the Binet adaptive tests and include CATs that are based on item banks that are stratified by levels of item difficulty or CATs that use structured item banks and administer items in stages. The latter are referred to as multistage tests (MSTs). Multistage tests administer a small set of items (sometimes called a testlet), score that set of items, and then select a new set of items based on the score from the first set. Multistage tests typically are based on a predesigned structure of testlets and have several stages of testlets arranged in a branching structure. Movement from one stage to the next is based on an examinee’s score on the previous stages. The number of stages is predetermined, as is the number of items in each testlet.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading