• 作者： 黃毅志
• 作者服務機構： 台東師範學院教育研究所
• 中文摘要： Although the questionnaire survey is widely used in social science researches, the quality of the measurement is doubtful. How serious is the error of the measurement in Taiwan questionnaire survey? Did it make most research conclusion incorrect? It deserves to make the systematic quantitative research, but there's few research in this area in Taiwan. Most of the researches explore the measurement quality of the questionnaire survey which were used to take adults as sample to engage in "face to face interview." The elementary school students' self-report questionnaire is widely used in educational research, but the quality of the measurement is doubtful, too, and there's few systematic quantitative research about it. This research took the 5th grade students in elementary school in Taitung county as sample which were sampled through stratified random culster sampling, and got 764 valid sample. Data was collected about the students' studying and living condition, and it adopted the students' self-report questionnaire. Then making three questionnaires from the original questionnaire, one for fathers, another for mothers, and the other for students to assess the test-retest reliability. The questionnaires from the parents reflect parents' character more correct than the elementary school students' self-report. to make a check on the reality of students' self-report (the criterion-related validity). The major findings are: 1. Many students didn't answer their parents' education and occupation. Even though some of them answered, the correlation of father's education, mother's education, and father's occupational prestige which students answered with what parents answered were .72, .74, .78. Many students answered their fathers with college degree, but only 1/3 was real. It showed that many students couldn't know the family SES, especially in parents' education, but not occupation. 2. The range of the test-retest reliability in behavior and attitude variables was from .2 to .6, it's much lower than "the Social Change Survey" that took adults as sample by face to face interview. It could attribute to that many students didn't realize the meaning of the questionnaire, or they didn't concentrate on answering the questionnaire. 3. The correlation of the academic grades by students' self-report and the family SES was unexpectedly not lower than the standardize test and the family SES, it's closed. Then the construct validity of students' self-report was close to the standardize test. It can attribute to that the standardize test which was evaluated by research team workers wasn't judged by background, and it was more objective. The other reason may be the students didn't do it seriously because they didn't regard the test as the formal exam.
• 英文摘要： 在社會科學研究中，雖然問卷調查法被運用得非常廣泛，然而它的測量品質，受到許多質疑。就國內的問卷調查法而言，它的測量品質問題究竟有多嚴重，是否會使多數調查研究的結論遭到嚴重扭曲，值得作有系統的量化研究。然而，國內在這方面的研究，還做得很少，而所探討的大都侷限於「對成年民眾作面訪」的調查方法；至於在教育研究中，常用的學生（童） 自陳問卷方法，所得到的測量品質，可能更令人感到懷疑。然而更不容易看到對此問卷測量品質，作有系統的分析量化研究。本研究以台東縣國小五年級學童為母體，採用分層叢集隨機取樣，到被抽中的班級對學童的學習、生活情況作調查，請學童以自陳方式填答問卷，共得到七六四名有效樣本。再從此問卷中篩選部分題目，編製成三份問卷，在學童作完問卷之後，一份請帶回家給父親填答，一份給母親，另一份隨後對學童進行再測，以考驗這項學童自陳問卷的再測信度。父母所填答的問卷，較能真實反映父母本人的特質，用來比對學童自陳問卷的真實性（即效標效度）。而主要的研究發現與結論是：1. 許多學童沒回答父母教育、職業；而根據回答者所顯示的父親、母親教育年數，父親職業聲望，與父母回答所顯示的數值之相關（r） 分別為.72，.74，.78，並不高；根據父親的回答，學童回答父親為大學程度者，其實僅有三分之一確實為大學，其餘大都為國中、高中職。研究顯示：有許多學童弄不清家庭社經地位，特別是父母教育，而非職業。2. 至於許多行為、態度變項的再測信度（r），則大都界於.2至.6之間，比起中研院社會變遷調查對成年民眾作一對一面訪的再測信度低得多，這可歸因於許多學童對於題意不很了解，或者是在自陳問卷的情境下，不很專心作答。3. 不過學童自陳的月考成績，與理論上很重要的解釋變項「家庭社經地位等背景變項」的關連性，竟然沒有比測量繁複費時之標準化測驗成績與背景變項的關連性為低，而顯得很接近；學童自陳月考成績的建構效度，也就很接近標準化測驗成績。這除了可歸因於本研究工作人員閱卷的標準化測驗比較客觀，不會視背景而給分之外，標準化測驗的情境，不如學校的月考，有些學生認為不是正式考試，而不認真作答，可能也是造成標準化測驗與背景的相關，並沒有比較高的原因。
• 中文關鍵字： educational research; students' self-report questionnaire; reliability; validity
• 英文關鍵字： 教育研究；學童自陳問卷；信度；效度