The paper investigates the validity of a multiple-choice reading test using Weir and Khalifa’s (2008) model of reading. Fifty (50) multiple-choice reading items taken from the 2011 English state examination were administered to 496 Grade 12 secondary school students. A concurrent strategies questionnaire was used to elicit information on participants’ use of strategies during test writing. The results indicate that there are more items that target careful reading than those that target expeditious reading. This is to suggest that the ESE tasks appear to be inappropriate and may need some modifications to more closely reflect the actual test context.

Abstract-The study examines the construct validity of a reading test based on narrative texts. It aims to evaluate the extent to which the test tasks target the narrative macrostructure and microstructure elements in both the text passage and the test questions as well as the extent to which the order of test questions is sensitive to the narrative story line. The study uses 40 test items randomly selected from the DR Congo English state examination (ESE). The results indicate that the construct validity of the ESE based on narrative texts is threatened by the inclusion of a limited number of episodes in the narratives, the absence of some elements of episodic structure, the use of truncated narratives that fail to include coherence elements, the failure to examine all the critical aspects of the narrative construct, the inclusion of some test items that require examinees' abilities that are irrelevant to narrative story grammar, and the ordering of information requested by the test items that does not relate to the way information is ordered in the narratives.

This paper investigates the use of elimination of distractors strategy by exploring examinees’ response decision processes while answering test items. It aims to evaluate the extent to which examinees’ use of elimination of distractors strategy is construct-irrelevant. As instrument, the paper uses verbal reports in order to elicit response decision processes the selected six respondents used when they eliminated item options in some selected test materials. The results indicate five main patterns of response decision processes respondents used when eliminating distractors. However, of these five patterns, four patterns include response decision processes that can be qualified construct-irrelevant to reading construct; while one pattern included response decision processes that appear to be construct-relevant. This suggests that the use of elimination of distractors is a strategy that can sometimes be construct- relevant.

The paper examines the construct validity of a high stake multiple-choice reading test from a context-based perspective that proposes to link the different reading processes examinees engage in answering the test items to the actual context of reading activity. Using two questionnaires administered to the 126 student-participants and 29 teacher-participants in order to investigate the context of the DR Congo English state examination (ESE) and results from a mixed-methods study that has comprehensively investigated the construct validity of the same test, the results indicate that the ESE appears to have low validity as the test tasks do not appear to grip with the ESE context. The paper then suggests that the ESE tasks be adjusted so as to reflect the actual context of reading activity.

This paper examines the construct validity of a multiple-choice reading test from the epistemological stance that supports a close relationship between test construct and test context. It aims to evaluate the extent to which the test tasks not only target the reading construct, but also come to grip with the specific test context (Katalayi & Sivasubramaniam, 2013). Fifty (50) multiple-choice items taken from the 2017 English state examination (ESE) were administered to 496 Grade 12 secondary school students and a concurrent strategies questionnaire was used to elicit information on participants’ use of strategies during test writing (Katalayi & Sivasubramaniam, 2014).  The findings indicate that, although the fifty item questions cover the whole range of reading construct by targeting all the processing levels; there is, however, a predominance of items that require text processing at higher level than those items that require text processing at lower level. This is to suggest that since the ESE includes higher level test items it does not reflect the context of the ESE; and therefore, it is untenable in the educational practices of language teaching and testing.