BEST Plus 3.0: Assessing Speaking Using a Multi-Stage Adaptive Test

8:00 - 8:30am

Decatur A

Presented at: LTRC 2019

We will demonstrate BEST Plus 3.0, a computer-based speaking assessment used to assess the oral proficiency of adult English learners in the United States. The BEST Plus 3.0 is an individually administered, face-to-face adaptive oral interview designed to assesses the ability to understand and use unrehearsed, conversational, everyday language within topic areas generally covered in adult English language courses. During the assessment, test administrators read scripted interview questions and enter task scores into the computer. A computer application selects the next test question based on the examinee’s performance on previous tasks.

BEST Plus 3.0 is built on an innovative multi-stage adaptive test (MST) design. In a multi-stage adaptive test, examinees proceed through the item pool in “stages,” in which each stage has a fixed set of questions. The MST approach allows for greater measurement precision with fewer items than one fixed form that is administered to all examinees. It is particularly suited to cases where several items need to be administered together (in contrast to “item-level” computer adaptive testing).

In this demonstration, we will present a sample test. We will also discuss the underlying test design and show a custom-built computer application that was developed in order to support the MST design. The computer application includes a web portal that allows test users to manage test administration and score information, and also allows test developers to access test data and easily embed field test items for ongoing test refreshment. The demonstration will thus include features of the test user interface while also highlighting the behind-the-scenes components of the program that are of interest to language testers.