Washback Effect Of A High-stakes Test:Preparation For The Writing Tasks Of The Graduate School Entrance English Examination

The present study is an attempt at investigating washback effect of the writing tasks of the Graduate School Entrance English Examination (GSEEE), with a particular focus on how test preparation operates under the washback mechanism. The reconstructed "Participants-Process-Product" washback model (Hughes1993; Bailey1996; Cheng2005; Green2007). encompassing the participants’perceptions, the preparation practices and the outcomes, served as the theoretical framework. A mixed method design was employed, using questionnaire and semi-structured interview to collect data from623test-takers,84raters from a GSEEE marking center and5instructors from two coaching schools.The present study placed the test-driven preparation practices towards the GSEEE writing tasks at the center, and offered valuable insights into the interaction among perceptions, test-driven preparation practices and products. With regard to constructs that may influence preparation practices, motivational and cognitive perceptions of the test-takers including test use, test stakes, task demand, task difficulty, and self-efficacy were specified and analyzed successively. In terms of preparation categories, five factors emerged from factor analysis:memorizing practice, test familiarization, comprehensive learning, skills development, and drilling practice. The present study then applied structural equation modeling to investigate the potential pathway from perceptions to test-driven preparation practices. With regard to the effects of preparation practices, regression analysis with the GSEEE writing scores revealed that the length of test preparation made significantly positive differences in the predictive model, while the comparison of test-takers and raters on the usefulness of individual preparation activities in the in-depth interview revealed the underlying cause of the negative effect of all-purpose sentences and templates in the GSEEE scoring procedure. To sum up, the appraisal of the washback effect of the GSEEE writing tasks should never underestimate the relationship between tests, teaching and learning due to the intensive preparation practices and the examination-oriented learning tradition in China. The "Participants-Process-Product" washback model reconstructed and advocated in the present study provides insights into the complex process of washback on learners and offers research evidence to enhance the consequential validity of high-stakes language tests.