新北市國中語文類學術性向資優鑑定工具之效度研究

本研究旨在考驗新北市國中語文類學術性向資優鑑定測驗的效度，以102、103及104學年度參加新北市三所設有語文類資優資源班國中資優鑑定的學生為樣本，三個年度的人數依次為57人、60人及73人。研究分為三個部分：初複選各測驗之間的相關與內部結構、初複選測驗的預測效度、各種切截方式之篩選結果的比較。本研究運用的分析方法包括積差相關、因素分析、區別分析、卡方分析、獨立樣本t考驗與單因子變異數分析。主要的研究發現臚列如下：一、除了103學年度的初選「英語能力」測驗外，各年度的初選測驗之間都有中度以上的相關，複選測驗間則僅有「填空重組」和「看圖作文」兩者之間有顯著相關。二、除103學年度「英語能力」測驗、104學年度「影片理解」測驗獨立為一個因素之外，各年度的初選評量皆只有一個因素，複選評量則可分為「國文」及「英文」兩個因素。三、複選「填空重組」測驗是預測通過鑑定與否的最佳指標。四、除了103學年度的「英語能力」測驗之外，初選評量和「在校語文科成績」、「會考語文科成績」及「資優行為表現」皆有顯著相關。五、複選評量和「在校語文科成績」及「資優行為表現」大多沒有顯著相關，但是和「會考語文科成績」有顯著相關。六、除了103學年度之外，通過語文資優鑑定複選之學生，其「在校語文科成績」、「會考語文科成績」和「資優行為表現」皆優於未過初選的學生。七、比較「原始分數加權」、「標準分數加權」、「多元截斷標準」等三種複選計分方式和切截標準，發現不同方式與切截標準之篩選結果並無明顯差異。

關鍵字

語文資優；資優鑑定；測驗效度

並列摘要

This research aimed to investigate the validity of the junior high school “Language Talented Identification Exams (LTIE)” in New Taipei City, Taiwan. Participants were students from three junior high school with “language talented resource classroom” in New Taipei City, who took the LTIE in 2013 (N=57), 2014 (N=60) and 2015 (N=73), respectively. The research was divided into three parts – correlation between tests and internal structure, predictive validity of students’ future performance and comparison of different scoring methods and cut-off points. The analytical methods adopted included Pearson correlation, factor analysis, discriminate analysis, chi-squared test, independent sample t-test and one-way analysis of variance (ANOVA). The major findings are as follows: 1. While all of the tests in the first round of LTIE were at least moderately correlated (except the “English ability” test in 2014), in the second round significant correlation was found only between the “fill-in-the-blanks/ rearrangement” and “picture writing” tests. 2. The first round of LTIE was composed of a single factor, whereas the second round consisted of two factors (Chinese and English). However, “English ability” and “video comprehension” constituted an additional factor in 2014 and 2015, respectively. 3. The “fill-in-the-blanks/ rearrangement” test was the most powerful predictor for whether the participant would pass LTIE. 4. All the tests in the first round were significantly correlated with GPA, Comprehensive Assessment Program for Junior High School Students (CAP), and Observation Scale for Gifted Behavior (OSGB), except the “English ability” test in 2014. 5. Most of the tests in the second round did not correlate with GPA and OSGB yet were significantly correlated with CAP. 6. Except for 2014, langugage talented students performed better than “non-passers” in GPA, CAP and OSGB. 7. When three methods for scoring and setting cut-off points ( “sum of weighted raw scores”, “sum of weighted standard scores” and “multiple cutoff points”) were compared, no significant difference was found.