Is the test constructor a facet?

Abstract

Most modern language testers believe that the writing of a successful instrument begins with specifications, and that instruments constructed devoid of specifications are likely to go astray. The purpose of this study was to explore the relative impact of the test-developer on the performance of test-takers using multiple-choice reading comprehension tests that had no specifications. Traditional reading comprehension tests often consist of short prose passages followed by sets of multiple-choicecomprehension questions. The test constructor formulates the stem and the choices on the basis of his or her own renditions of the passage. The test-takers provide evidence of their comprehension in terms of their written responses to the items designed by the test constructor. Since the characteristics of the test method limit the responses the testees can provide, the expected response becomes part of the test method. Accordingly, it seems reasonable to suggest that there may be a facet associated with the test-developer.Six passages each with 3 different sets of multiple-choice items constructed by 3 (groups of) individuals were trialled on 335 Iranian EFL (English as a foreign language) learners. The results revealed differential performance on almost all sets, suggesting a test constructor effect.