查詢結果分析
來源資料
頁籤選單縮合
題 名 | 貝氏隱藏式類別分析統計法在數學科教育評量的應用=Bayesian and Classical Latent Class Analysis: An Application of LSAY Mathematics Assessment |
---|---|
作 者 | 楊志堅; 劉湘川; 楊志強; | 書刊名 | 臺中師院學報 |
卷 期 | 13 1999.06[民88.06] |
頁 次 | 頁1-13 |
分類號 | 521.36 |
關鍵詞 | 隱藏式類別分析; 吉氏取樣; E-M算法; Latent class analysis; Gibbs sampling; E-M algorithms; |
語 文 | 中文(Chinese) |
中文摘要 | 聯考式的一試定終身的測驗方法在現今進步的社會下,逐漸被人們所遺棄。取而 代之的是以多次的在校學科評量成績為主要的學習成就參考依據,但是如何以這些多次的測 量(repaat measurements)結果來看出學生真正的學習成就與能力呢?最簡單的方法便是取平 均數再以之為分類標準,但這似乎不是最好的方法,因為每項試題難易不同,取平均數並無 法兼顧到這個事實,也無法鑑別出受測者的真正能力。本研究便以隱藏式類別分析(Latent Classes Analysis)模式來分析出受測者的真正能力,並以兩種主要的統計方法:E-M算法(E-M algorithms)及吉氏取樣(Gibbs Sampling)做統計計算學的比較研究。 |
英文摘要 | Latent class analysis has been shown as an important and practical analytic tool for substantive researchers (see e.g. Wang, et al., 1996, Biometrics; Qu, et al., 1996, Biometrics). Applications of latent class can be seen in a lot of areas, e.g., biomedical, educational, and psychological studies. Though several related models(e.g. Follmann and Lambert, 1989, JASA) were proposed years ago, they have not been used much in practice until recently. One of the reasons is due to the numerical difficulties of xture modeling, especially in the framework of Bayesian statistics. This paper investigates latent class analysis under both classical and Bayesian statistics frameworks and suggests more flexible models with covariates and including possible random effects. Estimation results using EM algorithms and Gibbs sampling are compared. An educational example using LSAY data is analyzed and illustrated in this paper. |
本系統中英文摘要資訊取自各篇刊載內容。