Completion Items

Educational and psychological tests are composed of test items (questions or statements requiring a response) in many formats. One common format is the completion item. In this context, completion items come in several forms with several names.

Most commonly, completion items include some form of constructed response and are frequently called constructed-response items. The item does not contain options from which a person could select a response but requires the individual to construct a response. This may be accomplished by asking a complete question or providing a statement that must be completed, as the following examples demonstrate.

• Item 1: What type of reliability can be estimated from one form of a test administered on a single occasion?
• Answer: Coefficient alpha, split-half reliability.
• Item 2: Describe one advantage of true-false items compared with multiple-choice items. Answer: Less testing time per item, easier to construct.
• Item 3: To be most useful, norms should be representative, relevant, and ____________. Answer: recent.

Some people distinguish between constructed-response items and completion items. Many people consider completion items to be primarily of the short-answer type. Constructed-response items, on the other hand, may also include extended response and essay items (requiring extensive responses, potentially several paragraphs long), configural response items (such as items requiring manipulation of schematic diagrams), or computation problems (commonly found in mathematics, where the individual must compute the answer).

The use of completion items, including constructed-response items, has advantages and disadvantages. Among the advantages, completion items are appropriate when the objective being measured requires a written response; they are relatively easy to construct and, when responses are short or composed of a single word, easy to score; short-answer items can assess higher-order thinking skills; and completion items allow for novel responses or solutions. Disadvantages include scoring difficulty because of the many possible correct responses, which may reduce reliability of scores; need for longer testing time, compared with multiple-choice testing, to achieve adequate reliability; low likelihood of assessing higher-order thinking skills with single-word answer formats; and because constructed-response items take more time to complete, limitation of the content that can be covered in a single test period.

Although it is possible to construct multiple-choice items to measure higher-order thinking skills, it appears that the range of cognitive skills addressed by completion items is larger than the range addressed by multiple-choice testing. Empirical evidence suggests that when written to tap the same content and cognitive skill, completion items and multiple-choice items measure the same construct; when written to tap different cognitive skills, the two formats appear to measure substantially different constructs. So it is not the format that determines what is being measured, but the nature and quality of the problem presented by the item, whatever its format.

Applying Ideas on Statistics and Measurement

The following abstract is adapted from Bornstein, M. H., Hahn, C.-S., & Haynes, O. M. (2004). Specific and general language performance across early childhood: Stability and gender considerations. First Language, 24(3), 267–304.

Children participated in four longitudinal studies of specific and general language performance cumulatively from age 1 year and 1 month to

...