The Construct Validity of a Test: A Triangulation of Approaches
- PDF / 211,251 Bytes
- 18 Pages / 595.22 x 842 pts (A4) Page_size
- 59 Downloads / 208 Views
Volume two, Issue two
May 2012
The Construct Validity of a Test: A Triangulation of Approaches MOHAMMAD SALEHI Sharif University of Technology Bio Data: Mohammad Salehi is currently a faculty member of the Languages & Linguistics center at Sharif University of Technology and holds a PhD degree in TEFL from the University of Tehran. He got his MA degree from the University of Allameh Tabatabai in TEFL. He earned a BA degree in English literature from Shiraz University. He has taught in University of Kashan, University of Teacher Training, University of Amirkabir, Azad University of Karaj, University of Tehran, and University of applied sciences. He has presented articles in Turkey, Cyprus, Dubai, Armenia, Australia, Italy and Iran. His research interests include language testing and second language acquisition research. He has also written books on language testing and vocabulary. Abstract Three approaches of validation enquiry were applied on the data obtained from a proficiency test carried out with 3,398 PhD candidates as a partial requirement for entering PhD program in Iran. The data obtained from the reading section were subjected to an exploratory factor analysis (EFA). The EFA yielded nine factors in the reading section. Furthermore, a design of multitrait-multimethod was also investigated, the traits being grammar and vocabulary and the methods, multiple choice and cloze test. The test demonstrated both convergent and discriminant validity. A third approach was intersubtest correlations as elaborated on by Alderson et al. (1995). For this, different sub-parts of the instrument were correlated with one another and with the total test. The test also showed construct validity as investigated via this approach. It can be concluded that the test had construct validity as revealed by these three approaches. Keywords: Exploratory factor analysis, Multi -trait multi-method, Varimax rotation, Factor structure, Construct validity Introduction Validation is an important enterprise especially when the test is a high stakes one. Admission tests for universities or other professional programs, certification exams, or citizenship tests are all high-stakes assessment situations (Roever, 2001). According to Messick (1988), if the validity of a test is not known it might have 102 | P a g e
Language Testing in Asia
Volume two, Issue two
May 2012
undesirable consequences for the society at large. University of Tehran English Proficiency Test (the UTEPT) qualifies as a high stakes test as almost ten thousands test takers take it on a yearly basis. It also has life changing implications for the test takers involved. Although studies in language testing have used single lines of enquiry in test validation (LLosa, 2007, Bachman and Palmer, 1983), this study uses triangulation in test validation and enquiry. The value of using a triangulation approach is clear from Anderson, Bachman, Perkins, and Cohen's (1991) statements in favor of it. Perhaps the greatest insight gained from this investigation is that more than one s
Data Loading...
