以試題反應理論分析CES-D量表不同調查方式的差異效果

Survey Mode Effect Analyzed by Item Response Theory on the Center for Epidemiological Studies Depression (CES-D) Scale

本研究主要是探討網路問卷與紙本問卷這兩種調查方式是否有所差異，問卷內容是採用國際間廣泛使用的CES-D憂鬱量表。2400位樣本分層隨機取自台灣中小學教師，其中共有1141位教師在20題CES-D量表上作答，包括535位網路組及606位紙本組教師。本研究藉由單參數試題反應理論軟體ConQuest開發的多向度潛在回歸模組來比較網路問卷組與紙本問卷組在問卷內容上的差異情形。研究結果顯示，儘管網路問卷回收率稍低於紙本問卷，但並未導致兩組受試者在背景變項上有所差異。在問卷內容的比較上，雖然網路問卷組在問卷內容的潛在平均數顯著地低於紙本問卷組，但是並無實質上的重要差異，亦即問卷組別變項對CES-D量表各因素的解釋力趨近於0。本研究認為仍有必要持續進行調查方式的比較研究，以確認網路問卷是否可取代紙本問卷，以減少研究及社會成本。

This study investigated whether differences exist between Internet and paper questionnaires in survey mode, where the internationally- and widely-used ”Center for Epidemiological Studies Depression” (CES-D) scale was adopted as content of the questionnaire. Of 2400 primary and secondary school teachers stratified and sampled randomly from Taiwan, 1141 teachers responded to the 20_item CES-D Scale, including 535 teachers in the Internet group, and 606 teachers in the paper group. This research used multidimensional latent regression module developed by ConQuest software. Using one-parameter item response theory, we compared the differences between the Internet and paper questionnaire groups. The results of this study showed no significant difference between the two groups in background variables; nonetheless, response rate in the Internet group was lower than that in the paper group. As for comparisons between content of questionnaires, although the latent means of the Internet group were significantly lower than those of the paper group, these differences had no substantial meanings. The group variable of survey mode explained close to zero variance in factors of the CES-D Scale. To examine whether Internet questionnaire survey can be used to substitute for paper survey in order to reduce research and administration cost, we encourage more comparisons between the two survey modes.