A Comparative Analysis Of Reading Tests In CET, TOEFL & IELTS

This author has conducted a research on the comparative analysis of various characteristics of reading tests among CET4, CET6, TOEFL and IELTS (A) in order to detect possible errors concerning test validity and reliability (particularly content validity, instrument-related reliability, and parallel or alternate forms reliability) and to provide implications for test designers, teachers and learners, especially CET designers on the improvement of reading test quality.To enable a sufficient sampling effort, the author has collected 20 latest consecutive versions of reading-related subtests in each of the above four tests, thus making up altogether 80 versions of reading subtests, within which about 360 reading texts (passages or essays) are involved.The comparative analysis of reading test characteristics in CET4 and CET6, TOEFL and IELTS (A) is conducted from two aspects——the analysis of critical characteristics of reading texts and the analysis of critical features of reading questions, and on two dimensions——horizontally and vertically, corresponding mainly to validity analysis and reliability analysis respectively. Based on a proposed framework of critical facets of reading tests within the four influential English tests, quantitative analyses have been conducted and statistical data obtained. Major findings concerning test validity are: 1) reading tests of IELTS (A) provide texts with more authentic forms and greater length than those of the other three tests, 2) reading texts in the four tests are close in sentence length, percentage of long words and readability (according to Gunning Fox Index), 3) a majority of reading texts in CET4 and CET6 are concerned with social sciences topics while reading texts in TOEFL and IELTS (A) are more evenly distributed among a variety of topics, 4) reading tests in TOEFL have a more balanced distribution between the macro-level and micro-level testing points, while those in the others contain more testing points at the macro level. Major findings related to test reliability are: 1) text length and sentence length in TOEFL reading tests are more stable than those in CET4 and CET6, 2) percentage of long words and readability of reading texts are more stable in CET6 than in TOEFL, in which the two indexes are more stable than in CET4, 3) IELTS (A) does better in providing reading questions in a variety of formats and is thus more instrumentally reliable than the other three, and CET4 and CET6 do a better job in this aspect than TOEFL, 4) reading tests in CET produce errors of unreliability due to the inappropriate use of alternative tasks such as Cloze, SAQ, Error Correction and Translation, which in effect vary greatly in a variety of reading test facets mentioned above.