查詢結果分析
來源資料
相關文獻
- 從國小六年級學籍資料預測國中輟學行為的邏輯迴歸分析與區辨分析
- 統計方法在總統選舉預測的應用
- 一個中輟傾向量表的編製及其信度與效度
- 國小導師預測學生國中輟學可能性之甄別研究
- 把孩子找回來--國中中輟生問題之研究
- Analysis of Referral Source, Severity, and Return Among the Elderly in Rural Primary Care
- An Analysis of Referral Initiative, Severity, and the Return of Older Population in a Tertiary Teaching Hospital
- 編製我國小學六年級女生健康體適能效標參照的組合測驗
- 編製國小六年級男生健康體適能效標參照的組合測驗
- 中途輟學青少年福利機構之規劃與經營
頁籤選單縮合
題 名 | 從國小六年級學籍資料預測國中輟學行為的邏輯迴歸分析與區辨分析=Employing Students' Grade-Six Archived Information to Predict Dropout Behavior in the Junior High School |
---|---|
作 者 | 章勝傑; 陳金燕; | 書刊名 | 臺東大學教育學報 |
卷 期 | 14(下) 民92.12 |
頁 次 | 頁125-151 |
分類號 | 527.47 |
關鍵詞 | 中輟; 高危險羣; 早期甄別; 邏輯迴歸; 區辨分析; Dropouts; At-risk; Early-identification; Logistic regression; Discriminant analysis; |
語 文 | 中文(Chinese) |
中文摘要 | 為了能及早甄別臺東的中輟高危險群學生,研究者以臺東縣的201名曾被通報的中輟學生,以及加成隨機抽出的299名非中輟生為研究對象,取得他們國小六年級的檔案資料進行預測研究。研究者以邏輯迴歸與區辨分析兩種方式進行統計分析。研究結果發現,以估計樣本求得之最適邏輯唔歸函數包括了六年級數學科平均成績、父母婚姻狀況、家庭經濟狀況、以及六年級輔懙紀錄等預測變項,此最適邏輯迴歸函數能正確分類驗證樣本中71%的中輟生與70%的非中輟生,整體預測正確率在七成左右,並有40%的偽陽性誤判率。同樣的,以估本所得的最適區辨函數則除了上述四個預測變項外還包括負面的生活適應、六年級社會科平均成績、及曠課日數等三個變項。以此區辨函數對驗證樣本進行再分類的中輟生擊中率為66%,非中輟生擊中率為76%,整體預測正確率與偽陽性誤判率則分別為72%與36%。區辨分析與邏輯迴歸的結果各有所長,但差異不大。整體而言,雖然在預測的準確性上仍有進步的空間,以學童國小學籍卡與輔導紀錄表上記錄的六年級資料來預測他們到國中的輟學行為已具可行性與實用性。 |
英文摘要 | In order to identify dropout-at-risk students early in their elementary years, the researchers retrieved the archived grade-six information of 201 dropouts and 299 non-dropouts to establish predictive models for dropout behavior in their junior high years. Logistic regression and discriminant analysis statistical methods were used in this study. It was found that math grade point, parent marital status, family economic status, and negative counseling records were the significant variables that stayed in the logistic regression equation. When this equation was used to reclassify the validation sample, it can classify 71% of the dropouts and 70% of the non-dropouts correctly. The hit rate for the whole sample is 70% and the false positive rate is 40%. The results pattern for discriminant analysis is similar but with some difference. Beside those significant predictive variables mentioned above, three additional significant variables, negative life adjustment, social studies grade points, and absented days, also stayed in the discriminant function. When the discriminant function was applied to the validation sample, it correctly reclassified 66% of the dropouts and 76% of non-dropouts. The average hit rate and the false positive rate were 72% and 36%, respectively. It is concluded that using students’ archived information in the elementary school to identify those dropout-at-risk students is a feasible and realistic approach, although the accuracy of prediction still left rooms for further improvement. |
本系統中英文摘要資訊取自各篇刊載內容。