青少年事故傷害自我效能量表的二階驗證性因素分析與多組不變量統計考驗
作者:王國川(國立臺灣師範大學衛生教育研究所)
卷期:43卷第1期
日期:1998年4月
頁碼:33-48
DOI:10.3966/2073753X1998044301003
摘要:
為了更能掌握青少年事故傷害自我效能量表的理論基礎與其特質,評估並探討此量表之高階潛在因素結構是建立此量表過程中不可或缺的重要步驟。本研究以臺灣地區八十四學年度公、私立國中、高中高職日間部學生為研究母群,以「青少年事故傷害自我效能量表」為評估、探討目標,以EQS統計軟體為效度評估與探討的主要分析工具。結果有三項主要的發現:第一、經整體暨個別的殘差分析、整體模式之適配與其取代指標,以及估計參數之統計考驗等結果,均指出假設此量表為二階因素分析測量模式是相當恰當的;第二、經交互確認效度分析結果,顯示此量表之二階潛在因素結構是非常穩定;第三、經多組別不變量統計考驗結果,指出此量表之二階潛在因素結構,對於學校別、性別、年級、用手習慣,以及過去傷害經驗這些社會人口變項,大致上具有因素個數、因素負荷量、因素變異數,以及測量誤差變異數等不變量的特性。
關鍵詞:青少年、事故傷害、自我效能量表、高階因素效度評估
《詳全文》
Journal directory listing - Volume 43 Number 1 (1998/April) - Education【43(1)】
Adolescent Self-Efficacy Scale In Unintentional Injury: Testing For Second-Order Factorial Validity And Invariance
Author: Wang, Kuo-chang(Graduate Institute of Health Education, National Taiwan Normal University)
Vol.&No.:Vol. 43, No. 1
Date:April 1998
Pages:33-48
DOI:10.3966/2073753X1998044301003
Abstract:
It is essential to evaluate the higher-order latent factor structure of the adolescent self-efficacy scale in unintentional injury in order to comprehend its theoretical foundation and characteristics. The 9,770 students, randomly sampled from junior high, senior high, and vocational schools of the 1995-1996 academic year in Taiwan, were administered this scale, and the statistical program, EQS 5.3 for Windows, was used for the major tool of data analysis. There were three major findings. (1) On the basis of the assessment of overall model fit and parameter estimates, the hypothesized second-order confirmatory factor analysis model represented an adequate fit to the sample data. (2) According to the cross-validation analysis, the second-order factorial validity of this scale was very stable. (3) This hypothesized model was invariance across these five important demographical variables, including school, gender, grade, right/left-handedness, and past experience in unintentional injury, via the testing of invariance in the multiple groups.
Keywords:adolescent, unintentional injury, self-efficacy scale, higher-order factorial validity