An argument-based validation study of the English Placement Test (EPT): Focusing on the inferences of extrapolation and ramification.

摘要

English placement tests have been widely used in higher education as post-admission assessment instruments to measure admitted English as a second language (ESL) students' English proficiency or readiness in academic English, usually upon their arrival at universities in English-speaking countries. Unlike commercial standardized English proficiency tests, many English placement tests are locally developed with comparatively limited resources and are relatively under-investigated in the field of language testing. Even less attention has been directed to the score interpretation and the impact of placement decisions on ESL students' English learning and academic achievement. Undoubtedly, this scarcity of research on English placement tests is inappropriate in view of their status as one of the most frequently used language testing instruments, which may exert immediate and strong impact on ESL students' learning in general.;By employing a mixed-methods approach, this dissertation project investigates the validity of test score interpretation and use of the English Placement Test (EPT) used at Iowa State University (ISU) under an argument-based validity framework. More specifically, this study started with an interpretation and use argument for the EPT, which states the score meaning and intended impact of the EPT explicitly, and focused on the last two inferences in the interpretation and use argument, namely extrapolation and ramification. The extrapolation inference links expected scores of the EPT (scores that exhibit adequate test reliability) to target scores or actual performance in the target domain. In this study, the extrapolation inference requires investigation of the relationship between ESL students' English placement test performance and two external criteria of English performance, including the TOEFL iBT and a self-assessment. The ramification inference links the use of the EPT results to its actual impact and in this study the ramification inference requires investigation of the impact of the placement decisions in a specific educational context.;For the extrapolation inference, quantitative data such as test performance data on the EPT, the TOEFL iBT, and the self-assessment were collected and analyzed using multitrait-multimethod (MTMM) analysis techniques. The findings indicated that the EPT was found to have moderate relationships with the TOEFL iBT and weak to moderate relationships with the self-assessment. The EPT showed some of the expected convergent correlations as well as discriminant correlations based on the MTMM correlation coefficient matrix as well as the factor loading parameters in a correlated trait-correlated uniqueness (CTCU) model. For the ramification inference, three types of analyses were conducted to seek support with regard to 1) test stakeholders' perceptions of the EPT placement decisions, 2) the impact of the EPT placement on ESL students' English learning, and 3) the relationship between ESL students' EPT performance and their first-semester academic achievement. The interviews with test stakeholders were coded and analyzed to identify statements indicating their perceptions of the impact of the placement decisions. The qualitative findings are also utilized to help interpret the quantitative findings. Multiple paired-samples t-tests are used to investigate ESL students' progress in the ESL courses that they were placed into. In addition, a structural equation modeling (SEM) approach was used to model the relationship among students' performance on the EPT, ESL courses, and their first-semester GPA, mediated by individual difference constructs, such as learning motivation, academic self-efficacy, and self-regulated learning strategies.;The qualitative analyses of the interviews with four groups of test stakeholders show that the interviewed ESL students in general experienced initial frustration regarding the placement decisions, in retrospect, they understood why they were placed into ESL courses and appreciated the benefits of taking the required courses, especially ESL writing courses. The ESL course instructors were satisfied with the placement accuracy, even though occasionally they identified a few cases of potentially misplaced students in the ESL courses. The interviewed undergraduate academic advisors showed positive perceptions of the EPT and the placement decisions. They also reported observing that the majority of the ESL advisees were receptive to the EPT placement decisions.;The analyses of ESL course performance data collected at the beginning and the end of the course indicate that ESL students in Engl99L, an ESL listening course focusing on listening strategies, made statistically significant progress in terms of score gain on the same listening test administered at two time points. However, only nine out of 38 ESL students made satisfactory progress with reference to the course standard. Students in Engl101B (a lower-level ESL academic English writing course) and Engl101C (a higher-level ESL academic English writing course) did not show much progress in terms of lexical complexity, syntactic complexity, and grammatical accuracy. However, the Engl101C students on average wrote longer essays at the end of the course. Based on the ratings of the essays written in the final exams using the EPT scoring rubric, 14 out 18 Engl101B students (77.8%) and eight out of 16 Engl101C students (50%) showed satisfactory progress in these classes and were deemed ready for the next level of English study. The SEM analysis results indicate that ESL students' EPT performance had significant and direct impact on their academic achievement. What's more, students' EPT performance predicted their academic self-efficacy and affected extrinsic goal orientation. However, these motivational factors did not have direct impact on academic achievement.;The findings in this study contribute to building the validity argument for the EPT with two of the assumptions underlying the warrant for the extrapolation inference and ramification inference found supported and the other three partially supported. This findings in this study contributed to a better understanding of the score interpretation and use of the EPT at Iowa State University through constructing a validity argument. These findings shed light on the future development of the EPT and other similar English placement tests. The findings in this study as well as the research methodology can be informative for other institutions where English placement tests are used.