The effect of changed responses in true-false tests.

Abstract

Many teachers have been advising their students to give preference to their second judgments on true-false tests if these judgments differ from the initial impression. This advice has been based in part on studies tending to show that second judgments are the more often correct. Lamson 1 found that the ratio of correct to incorrect changes when second judgments were given preference was about two to one. In her investigation, however, the students were asked to identify any changes in answers and were thus placed in the position of being very conscious of all such changes.Mathews' 2 results were much the same; but his students marked the true-false items with plus and minus signs, making it difficult to identify all changes. Lowe and Crawford,' from an analysis of answers made under normal testing conditions, report that two-thirds of the changed answers were changed correctly.The tests analyzed in the present study were given under ordinary testing conditions in three different Education courses: Character Education, Methods of Teaching in High Schools, and Extra-curricular Activities. Nothing was said to the students about changing their responses. They had plenty of time and were told to check over their papers when finished. In no case did any student have insufficient time for this purpose, although not all of them so made use of the time. The students had written instructions not to check an item unless reasonably sure about it, the R-W marking method being used. All students were required to use pencil and to check the items plus or zero. There were very few cases in which a changed answer was difficult to identify. All papers were analyzed by the same person, the writer.Eleven different true-false tests taken by five hundred two thirdand fourth-year college students were analyzed. The tests ranged in

Staying with Initial Answers on Objective Tests: Is it a Myth?

“…In one study (Archer & Pippert, 1962) this approach also included checks by another independent judge. The high interobserver agreement provides support for a number of authors who have commented on the ease and unambiguousness of detecting answer changes (Hill, 1937;Jarrett, 1948;Archer & Pippert, 1962;Foote & Belinky, 1972). Vol.…”

Implications of Changing Answers on Objective Test Items

“…The percentage of total responses made by all test-takers for which answers were changed is universally reported to be very small, with a median, from seven studies reporting this data, of 3.2 percent (Hill, 1937;Jarrett, 1948;Lamson, 1935;Lowe & Crawford, 1929;Mathews, 1929;Mueller & Shwedel, 1975;Reile & Briggs, 1952). In only one study (Reile & Briggs, 1952) was the percent of answers changed reported to be greater than 4 percent.…”

“…For true-false and multiple-choice tests the percentages of students changing one or more answers varies from 70 percent to 96 percent, with a median, from seven studies reporting this information, of 89 percent (Archer & Pippert, 1962;Bath, 1967;Foote & Belinky, 1972;Hill, 1937;Jarrett, 1948;Lehman, 1928;Mueller & Shwedel, 1975). Only one of these studies (Lehman, 1928) employed all true-false items.…”

“…Virtually all studies reviewed report that when students change answers their changes result in a net gain in points. Using changes (as opposed to students) as the unit of comparison, several studies reported gains/loss ratios for true-false items ranging from 1.4/1 to 2.5/I, with a median of 2/1 (Hill, 1937;Jarrett, 1948;Lamson, 1935;Lowe & Crawford, 1929). Using students as the unit of comparison, two studies reported the ratio of students gaining to students losing points as 1.4/I (Lehman, 1928) and 3/1 (Lowe & Crawford, 1929).…”

Oral Versus Visual Presentation of True-False Achievement Tests in the First Course in Psychology

“…Hill (5) found that only 2.5 per cent of more than ~,000 responses were changed, and with little effect, as the changes resulted in raising scores in only i per cent of the cases.…”