upperclassmen (N = 76) were taught 30 general science concepts. Frames of the program were multiple-choice items dealing with commonly misunderstood science concepts. One response to each item was a correct response, another response was a common misunderstanding of the concept, and the other two responses were reasonable and plausible distractors. Equipment used was an IBM 1410 computer and four IBM 1060 teletypewriter terminals equipped with random-access slide projectors. The five modes of feedback compared were (A) no feedback, (B) feedback of "correct" or "wrong," (C) feedback of the correct response choice, (D) feedback appropriate to the student's response, and (E) a combination of B, C, and D. The means of those groups which were guided to the correct response by feedback (C, D, and E) were significantly better (p < .01) than those groups which were required to "discover" the correct response. Posttest results indicated better immediate retention for those Ss receiving a combination of feedback modes.

Fifty-four graduate students were administered a 66 item four-response multiple choice test on self-scoring tests forms. Each test was scored by the traditional right-wrong method of scoring tests and was also scored by the selfscoring method of counting the number of responses necessary to respond to all items correctly. Results indicate that scoring tests by the self-scoring method can result in a higher split half reliability than tests scored by the traditional right-wrong method.Self-scoring tests are forms or devices which allow the student being tested to obtain immediate feedback concerning the correctness of each response by means of symbols which are revealed to him immediately after he has made a response to a question. Students who take self-scoring tests are required to continue responding to each item on the test until a correct response is obtained. Tests administered by the self-scoring method (SSM) are scored by counting the number of responses required to answer correctly all the items on the test.This scoring method differs, of course, from the right-wrong scoring method (RWSM) in which a student responds to each item one time only and receives one point for each correct response and no points for other responses.Self-scoring tests have been used as an aid to instruction since they were invented by Pressey (1926). Several studies (Pressey, 1926; Jensen, 1949)have demonstrated the advantages for student learning that occur when tests are administered by SSM.It is also logical to speculate about the comparison of reliability indices of tests scored by SSM with those of tests scored by traditional methods. Since the SSM scoring technique requires students to continue responding to each item until the correct response has been obtained, each item, once answered, could be considered another item with k-I of the k response choices. The range of scores is therefore larger for SSM than for RWSM, with a corresponding increase in the standard deviation. These increases may be somewhat similar to the effects of increasing the length of a test. The apparent increase in test length and the increase in the standard deviation would produce a higher reliability index.It is also possible that SSM may cause an increase in reliability indices since students are given partial credit for items not answered on the first attempt. Ebel (1965) found that scoring methods designed to give partial credit for incorrect responses when items are scored on the basis of degree of certainty estimates--and which consequently give partial credit for items not answered correctly by the student's first response--can result in greater precision of measurement than would have been obtained through RWSM.This study compared the test reliability of SSM with the reliability of inferred RWSM. A comparison was made between odd-even item correlation coefficients and split-half reliability indices computed from scores obtained through SSM and RWSM. 205

No abstract

No abstract