本研究旨在探討網路與紙本兩種問卷調查方式及對問卷所調查議題的熟悉度兩個變項，對問卷回收率及問卷調查內容是否有顯著差異的影響存在。本研究以分層隨機取樣方式抽取臺灣地區1,200位中小學教師進行研究，實際回收425份問卷，整體回收率35.4%，其中網路組195人（佔45.9%），紙本組230人（佔54.1%）。本研究以卡方檢定來進行回收率的比較，並且使用Rasch模式，將李克特式評定量尺資料轉換成具連續量尺資料後，再進行問卷內容的分析。結果發現：（1）在問卷回收率上，調查方式與議題熟悉度之間並無交互作用效果存在，也無主要效果存在；（2）在問卷內容上，調查方式與議題熟悉度之間有交互作用效果存在，其中，針對差異試題功能（DIF）的單純主要效果檢定發現，在高議題熟悉組或低議題熟悉組中，網路組與紙本組之間幾乎沒有出現DIF的試題；但是就高低議題熟悉組之間的比較而言，在網路組出現DIF的試題與在紙本組出現DIF的試題有較大的不同；（3）調查方式的DIF主要效果顯示，網路組與紙本組在各試題上並無DIF現象；而議題熟悉度的DIF主要效果顯示，高低議題熟悉組之間出現許多DIF試題，且高議題熟悉組的反應較低議題熟悉組極端；（4）未來的問卷調查方式，似乎可以考慮使用網路問卷來取代紙本問卷，或者是兩者並行，因為，其效果是相同的。最後，本研究亦提出許多建議，供未來研究時參考。本研究旨在探討網路與紙本兩種問卷調查方式及對問卷所調查議題的熟悉度兩個變項，對問卷回收率及問卷調查內容是否有顯著差異的影響存在。本研究以分層隨機取樣方式抽取臺灣地區1,200位中小學教師進行研究，實際回收425份問卷，整體回收率35.4%，其中網路組195人（佔45.9%），紙本組230人（佔54.1%）。本研究以卡方檢定來進行回收率的比較，並且使用Rasch模式，將李克特式評定量尺資料轉換成具連續量尺資料後，再進行問卷內容的分析。結果發現：（1）在問卷回收率上，調查方式與議題熟悉度之間並無交互作用效果存在，也無主要效果存在；（2）在問卷內容上，調查方式與議題熟悉度之間有交互作用效果存在，其中，針對差異試題功能（DIF）的單純主要效果檢定發現，在高議題熟悉組或低議題熟悉組中，網路組與紙本組之間幾乎沒有出現DIF的試題；但是就高低議題熟悉組之間的比較而言，在網路組出現DIF的試題與在紙本組出現DIF的試題有較大的不同；（3）調查方式的DIF主要效果顯示，網路組與紙本組在各試題上並無DIF現象；而議題熟悉度的DIF主要效果顯示，高低議題熟悉組之間出現許多DIF試題，且高議題熟悉組的反應較低議題熟悉組極端；（4）未來的問卷調查方式，似乎可以考慮使用網路問卷來取代紙本問卷，或者是兩者並行，因為，其效果是相同的。最後，本研究亦提出許多建議，供未來研究時參考。

The main purpose of this study was to investigate the effects of survey modes and issue-familiarity of questionnaires on the response rates and the surveyed contents. The questionnaires were administered to a stratified sample of 1,200 teachers who were randomly selected from elementary and secondary schools in Taiwan. The overall returned samples were 425 teachers and the response rate was 35.4%. Of the returned samples, the Web-based questionnaire group had 195 teachers (45.9%), and the paper-based questionnaire group had 230 teachers (54.1%). This study not only used the chi-square test on the comparison of response rates, but also used the Rasch model converting Likert-type scale into continuous scale data and followed by analyses on surveyed content. This study found that: (1) there were neither interaction effects nor main effects on the response rates between the survey modes and issue-familiarity of questionnaires; (2) there were interactions on surveyed content between survey modes and issue-familiarity of questionnaire; yet, by simple main effect tests of differential item functioning (DIF), no matter in high or low issue-familiarity group, there was almost no DIF item appearing between Web-based group and paper-based group, but items that appeared differences between high issue-familiarity group and low issue-familiarity group in Web-based group were much more different from those in paper-based group; (3) main effect test of DIF for survey modes showed that there was no DIF between Web-based group and paper-based group. However, there was serious DIF phenomenon between high issue-familiarity group and low issue-familiarity group, and responses in high issue-familiarity group were more extreme than in low issue-familiarity group; (4) in sum, Web-based questionnaires could be suggested for substitution for paper-based questionnaires or for using them both together on the future researches, because they almost have the same effects. Finally, some suggestions for future studies were also proposed.The main purpose of this study was to investigate the effects of survey modes and issue-familiarity of questionnaires on the response rates and the surveyed contents. The questionnaires were administered to a stratified sample of 1,200 teachers who were randomly selected from elementary and secondary schools in Taiwan. The overall returned samples were 425 teachers and the response rate was 35.4%. Of the returned samples, the Web-based questionnaire group had 195 teachers (45.9%), and the paper-based questionnaire group had 230 teachers (54.1%). This study not only used the chi-square test on the comparison of response rates, but also used the Rasch model converting Likert-type scale into continuous scale data and followed by analyses on surveyed content. This study found that: (1) there were neither interaction effects nor main effects on the response rates between the survey modes and issue-familiarity of questionnaires; (2) there were interactions on surveyed content between survey modes and issue-familiarity of questionnaire; yet, by simple main effect tests of differential item functioning (DIF), no matter in high or low issue-familiarity group, there was almost no DIF item appearing between Web-based group and paper-based group, but items that appeared differences between high issue-familiarity group and low issue-familiarity group in Web-based group were much more different from those in paper-based group; (3) main effect test of DIF for survey modes showed that there was no DIF between Web-based group and paper-based group. However, there was serious DIF phenomenon between high issue-familiarity group and low issue-familiarity group, and responses in high issue-familiarity group were more extreme than in low issue-familiarity group; (4) in sum, Web-based questionnaires could be suggested for substitution for paper-based questionnaires or for using them both together on the future researches, because they almost have the same effects. Finally, some suggestions for future studies were also proposed.