Comparative Reliabilities and Difficulties of the Multiple-Choice and True-False Formats

An Analysis of Textbook Advice About True-False Tests

“…Research findings support this assertion as well (Frisbie, 1973;Oosterhof & Glasnapp, 1974). But because true-false items provide more efficient measurement, this limitation is not so severe.…”

“…Rule 16 indicates that about the same number of true and false statements should be used so that, its proponents argue, an imbalance will not cue the testwise students in a guessing situation. However, those who support Rule 20 maintain that more false items than true should be used because false items have been shown to be more discriminating of achievement levels than true items (Barker & Ebel, 1981;Oosterhof & Glasnapp, 1974). The higher level of discrimination leads to more reliable scores, a major goal of builders of norm-referenced tests.…”

“…Relative to most other item formats, true-false items can be read quickly and responses can be recorded just as efficiently. Considerable research evidence shows that at least 50Vo more true-false items than multiple-choice items can be used in a given amount of testing time (Frisbie, 1973(Frisbie, , 1974Oosterhof & Glasnapp, 1974). Thus, more elements of the content domain can be sampled (breadth) or each element can be sampled more thoroughly (depth) with true-false than with contentparallel multiple-choice items.…”

“…That reliable scores can be obtained with true-false tests is documented by our own classroom experience and by substantial research evidence. Relative to multiple-choice tests, item-for-item discrimination is somewhat less for true-false tests, but the potential for greater length overcomes the discrimination deficits to yield scores of essentially the same level of reliability (Frisbie, 1973(Frisbie, , 1974Green, 1979;Oosterhof & Glasnapp, 1974).…”

True‐False, Alternate‐Choice, and Multiple‐Choice Items

“…This study showed a 2.4 to 1 ratio of TF to multiple-choice items per minute. Oosterhof and Glasnapp (1974) also compared TF and multiple-choice items with respect to reliability, difficulty, and discrimination. Their results showed TF items to be of lower overall reliability.…”