| Summary: | 碩士 === 國立臺中教育大學 === 教育測驗統計研究所 === 102 === The purpose of this research is to compare binary structure and multiple scoring with unlimited multiple scoring. Traditional binary structure is a answer mode of second election, application convenient and clear. In this research, 4-points multiple scoring、5-points multiple scoring and 6-points multiple scoring which more commonly used in the multivariate, and the newly developed unlimited multiple scoring was planted with the "brace" feature scoring. Use the Monte Carlo simulation to simulate the data sets of three different related levels, each relevance type of data set is subdivided into three different sizes of data samples, various of scoring estimation subjects ability true ability, and calculated Mean Square Error values.
The results are as follows：
1. When the number of different samples, various methods obtain mean square error values are different from each other the difference.
2. When the number of different questions, can't change the comparison result between scoring.
3. When the number of different K-value in cross-validation method, and will not affect the result of the comparison between the scoring.
4. In data of different of correlation levels, compared to binary structure and multiple scoring, unlimited multiple scoring ability to get closer to the true value of subjects.
|
|---|
