Validity of score interpretations on an online English placement writing test

Abstract

AbstractA much-debated question in the L2 assessment field is if computer familiarity should be considered a potential source of construct-irrelevant variance in computer-based writing (CBW) tests. This study aims to make a partial validity argument for an online source-based writing test (OSWT) designed for English placement testing (EPT), focusing on the explanation inference. Score interpretations on the OWST are proposed and supporting evidence is sought in terms of test-takers’ self-confidence in and preferences for CBW tests (two interrelated aspects of computer familiarity) and L2 writing ability. Ninety-seven ESL students demonstrating two different levels (higher and lower levels) of L2 writing ability studying at a US university completed the OSWT and an online questionnaire asking about their attitudes towards CBW tests. A series of statistical and thematic analyses revealed that most of the test-takers held self-confidence in and preferences for CBW tests for reasons related to previous CBW experience (e.g., familiarity with CBW, useful tools/functions available on computers) regardless of L2 writing ability. The higher-level test-takers obtained significantly higher scores on the OSWT than their lower-level counterparts. Test-takers’ preferences were a significant predictor of the OSWT scores only in the higher-level group. The findings largely support the validity of proposed score interpretations on the OSWT. Implications are discussed in terms of test fairness and the construct of CBW tests.