Author:

Keywords:

assessment, international standards

Abstract:

Background: This study of Flemish/English translation effects was conducted in conjunction with development of the International Foundations of Medicine Examination program. This collaborative effort involving the National Board of Medical Examiners and schools in Belgium, Italy, Portugal, the US, and other countries is designed to facilitate the interchange of students and mobility of graduates internationally. Summary of work: One-hundred basic science items recently retired from USMLE were translated into Flemish and divided into two 50-item blocks. 201 bilingual students from Katholieke Universiteit Leuven were randomly assigned to four groups, each group took a web-based test consisting of one 50-item block in English and one in Flemish, with counterbalancing of block content, order and language. Performance was analyzed to determine effects of language on item difficulty and response times. Summary of results: Mean scores on items presented in Flemish were 1% higher than when the same items were presented in English, three additional seconds were required to respond to items in English. The true (disattenuated) correlation between scores by language was 1.0, correlations between item difficulties and durations by language were 0.93 and 0.97, respectively. Issues in item translation were identified for a few items. Conclusions: Pending replication, it appears feasible to develop comparable forms of basic science examinations in Flemish and English starting with USMLE material. Take-home messages: With care in translation, the effects of language on test performance can be small.