| Summary: | 碩士 === 國立臺南大學 === 教育學系測驗統計碩士班 === 100 === Task difficulty is a significant issue that the test developers are concerned. Purposes of many large-scale assessments are to select examinees for certain purposes. To control or balance item difficulty of a whole assessment helps to improve the validity and discrimination. Test developers construct tests from the cognitive perspective; make the test constructing process more scientific and efficient, and the scores obtained represent the proficient level more precisely.
An analysis framework of cognitive components related to English listening comprehension (LC) test is proposed in the research. The data used in this study is derived from the norm of the After School Alternative Program English test (ASAP-ENGLC) item pool. 150 items are analyzed; item difficulty is concurrently calibrated under Three-Parameter-Logistic Item Response Theory from the response of 1459 participants from the fifth to seventh grade. Components may affect item difficulty are proposed and discussed from the cognitive perspectives. All items are coded with cognitive components, e.g. phonetic/ non- phonetic discrimination item type, number of plausible distractors, and content familiarity etc.. Multiple regression analysis is conducted to calibrate the correlation between the components and item difficulty. The result of this research indicates that phonetic/ non- phonetic discrimination item type, number of plausible distractors, necessity of inference, heterogeneity of sentence patterns in options and lexical overlap in key are the best predictors in the predicting model. The total explained variance of this cognitive predicting model is 53.5%. The research result is in expectation to make the task constructing procedure more organized, to make the task difficulty under controlled within the desired extent, so that the intended ability can be measure effectively. Besides, the purpose of ASAP English test is to assist in remedial instruction for low achievement students; the cognitive components and research results proposed in the present research will provide reference for developing teaching materials of English listening comprehension remedial instruction and constructing listening comprehension test items.
|
|---|
