Copy the page URI to the clipboard
Dawadi, Saraswati
(2016).
DOI: https://doi.org/10.21954/ou.ro.0000ef83
Abstract
There are a number of studies on construct validity of a language test in general, and a reading test in particular; most of them are quantitative in nature. As quantitative studies may not capture the cognitive processes the test takers follow to respond to a test item, the importance of collecting verbal reports of test takers has been recommended in order to establish the construct validity of a test (Alderson, 1990b). Additionally, no research to date, has investigated the construct validity of the reading test of the school leaving examination in Nepal. Therefore, the present study was designed to fill this gap i.e. to investigate whether the reading strategies, the test claims to test, are measured by the test. In order to investigate the issue, eight grade 10 students were asked to take the reading test and think-aloud when responding to the test items. Immediately after they completed the test, they were asked to take part in a retrospective debriefing. Additionally, in order to triangulate the data, the views of seven language testing experts on the reading strategies tested by the test were collected. The experts were first asked to make judgements individually with regard to the skills tested by the test. Then, they were asked to take part in a focus group discussion. The findings provide grounded insights into the response behaviours prompted by the reading tasks on the test and indicate some threats to the construct validity of the test as the test does not seem to be measuring all the reading strategies it claims to measure. The study also indicates a very low level of agreement among the subject experts regarding the strategies tested by the test. Consequently, there was a big gap between the strategy use and expert judgements. Therefore, the usefulness of expert judgements to predict which reading strategies the test items are testing is questioned and the need for further research concerning the methodologies used in the study has been suggested. More importantly, the study highlights the need to make the test more valid.