Evaluation of a Streaming Instrument
摘要
Throughout the last decade, the Rasch model has frequently been used as a tool to evaluate assessments. Rasch analysis provides estimates of item-difficulty and person-ability that are not dependent on raw scores. It is therefore particularly suitable for estimating ability from a test for the purposes of streaming students into levelled classes. The analysis can also be used to evaluate the assessment by highlighting items that are causing unexpected response patterns. The current results demonstrate that a multiple-choice test of English designed to stream students acted as a fairly effective tool for distinguishing the abilities of the population of test takers. However, several items appear to require modifications prior to future administrations of the test, with the hope that the streaming process will be rendered more precise. The suitability of using Rasch-based methods to evaluate streaming tools is discussed. Introduction Recently, Rasch analysis has replaced classical test theory analyses of pedagogical assessment because of its differing theoretical stance and practical applications. Compared to deterministic methods, Rasch analysis is based on a prescriptive model, meaning that conditions for the data to meet are prescribed, rather than being pre-determined (Salzberger, 2010). In other words, instead of rather than building