The term discrimination, as used in testing, refers to the extent to which a test distinguishes between strong and weak students. When a test is pre-tested, attention will be paid to each items item discrimination index. If all the members of the cohort get that items right, or if all get it wrong, the item will be rejected - it gives no evidence of the differing knowledge/ability of the members of the cohort and is therefore useless for testing purposes. If, on the other hand, only 20% of the pre-test cohort get an item right, and are the same 20% who score highest on the test overall, then the item clearly discriminates between the "top" group and the rest.
Davies (1999) suggests that, in general, an item would be expected to have a difficulty index of 30-70% to be useful in discriminating between candidates. He points out, however, that this will vary depending on test purpose. In a placement test or diagnostic test, difficult items may be included because of the aim of the test - to find out the candidates exact level, or what they do and dont know. For a test such as an achievement test, which expect a high pass rate, items which the candidates are expected to find easy may be included in order to give representative coverage to all the items taught on the course.
Bibliography : Davies, A. (1999) A Dictionary of Language Testing, CUP
0 komentar:
Posting Komentar