師範院校量化取向科學教育碩士論文之內容分析研究

A Content Analysis for Quantitative Masters Theses in Science Education from Normal Colleges and Universities

許良榮(Liang-Rong Hsu) ； 李永進(Yung-Jin Li)

《科學教育》 13期 (2007/07) Pp. 1-22

https://doi.org/10.6767/JSE.200707.0001

摘要

本研究旨在針對國內師範院校量化取向之科學教育碩士論文進行內容分析，研究工具為自行發展之檢核表，包括「量化研究工具發展狀況」、「信效度檢驗方式」、「統計考驗方式」與「量化研究缺失」，共計分析2001年至2003年之碩士論文134篇。研究結果發現：（一）研究工具發展方式以研究者自行發展者為最多，但是有14.2%的論文沒有說明研究工具之信度或效度之檢驗方式；（二）信度檢驗方式以「Cronbach α」為最多數，效度檢驗方式則以「內容關聯的效度證據」最多；（三）考驗方式以「t考驗」被使用次數最多，且統計考驗之次數有過多的現象；（四）論文的缺失分析顯示，除了「論文沒有電子全文於國家圖書館上網公開」有88篇(65.7%)為最多，其他尚有四項的缺失超過50篇次。

關鍵字

科學教育 ； 師範院校 ； 量化研究

並列摘要

The purpose of this study was to analyze the quantitative approach of Masters theses in science education from normal colleges and universities in Taiwan. The instruments of content analysis used in this study was based on four interests, which include: 'the conditions under which the instruments were developed', 'the methods of reliability and validity', 'the methods of statistical testing' and 'the problems associated with the content of the theses'. A total of 134 Masters theses collected from 2001 to 2003 were used in the analysis. The major findings are summarized as follow: (1) the researches usually develop the instruments used in these theses themselves. In addition, 14.2% of these theses did not describe the methods that were used to test the reliability and validity. (2) The sequence of frequency of the methods for reliability testing was: the Cronbach α＞test-retest reliability＞inter-scorer reliability. On the other hand, the sequence of frequency of the methods for validity testing was: content-related validity＞criterion-related validity＞construct validity. (3) With respect to statistical testing methods, the sequence of the frequency was: T-test＞one-way ANOVA＞Pearson correlation. It was also found that these theses tend to overuse statistical testing. (4) 88 theses (65.7%) was not uploaded to the National Library, moreover, 64 theses (47.0%) had not tested the homogeneity of variance before ANOVA and finally, 58 theses (43.3%) had not describing the sampling error. In view of the above findings, we propose several recommendations and suggestions that attempt to address the development of research instruments, the methods of statistical testing and the problem on the 'openness' of theses for quantitative study in science education.

並列關鍵字

normal universities and colleges ； quantitative research ； science education