Is “I Don’t Know” a Viable Answer Choice on the Vocabulary Size Test? TIM STOECKEL The University of Niigata Prefecture Niigata, Japan PHIL BENNETT Miyazaki International College Miyazaki, Japan STUART McLEAN Osaka Jogakuin University Kansai University Osaka, Japan doi: 10.1002/tesq.325 T he Vocabulary Size Test (VST) (Nation & Beglar, 2007) has attracted considerable attention in studies of second language acquisition. Designed to estimate the overall written receptive vocabu- lary size of English language learners, the test was created for a variety of pedagogical purposes—to guide syllabus design, inform decisions regarding extensive reading and vocabulary instruction, and monitor lexical growth over time (Beglar, 2010; Nation, 2012). Because VST scores can be inflated by blindly guessing unknown words under its multiple-choice format (Stewart, 2014), the inclusion of an I don’t know (IDK) answer choice has been explored as a means of achieving more accurate estimates of vocabulary size. Using this convention, researchers have reported both reduced guessing and improved estimates of reliability (Lucovich, 2014; Zhang, 2013). It has also been observed, however, that when faced with unknown words some learners are more likely than others to use IDK (Bennett & Stoeckel, 2012; Zhang, 2013). This means that examinees with the same vocabulary knowledge could achieve significantly different test scores, rendering the instrument ineffectual for many of the purposes listed above. Although use of IDK improves reliability, it is unclear whether the resultant scores correlate to actual vocabulary knowledge as well as scores without IDK, as this has gone unreported in previous studies. To address this gap, the goals of the present study are to BRIEF REPORTS AND SUMMARIES 965