電腦輔助分析軟體運用於質性研究訪談稿內容分析之探討
No Thumbnail Available
Date
2008
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
隨著社會思潮的湧現，質性研究在社會科學的領域裡逐漸嶄露頭角。其中，質性訪談是運用最廣泛的資料蒐集方法。研究者在質性訪談後，先建立訪談稿，再依照研究者所採用的理論架構，將問題的反應做分群，再著手進行分析。然而傳統人工分群往往耗費大量時間與精力。因此，本研究以電腦輔助分析軟體進行訪談稿分群分析，探討電腦輔助分析軟體與傳統人工分群的一致性，並說明電腦輔助分析軟體如何有助於訪談稿的分析。
本研究使用電腦輔助分析軟體ClustanGraphics8.02及主題萃取系統，針對社會科學領域畢業研究生（研究對象）之訪談稿進行實驗。研究中，先以電腦輔助分析軟體針對訪談稿進行分群分析。接著，根據電腦分析結果，對研究對象進行訪談，並利用rand index交相驗證電腦與人工分析結果之相似度，以探究電腦輔助分析軟體對質性研究訪談稿分析之幫助。
實驗與訪談結果顯示，雖然電腦輔助分析軟體與傳統人工分析結果的一致性不算高，但電腦分析之結果，在人工分析訪談稿前，可以給予初步之分群的構想；同時，電腦輔助分析軟體若運用於結構式或半結構式之訪談稿，可能得到較好的助益。此外，透過rand index計算所得之電腦與人工及綜合結果之相似度(一致性)數值，並無明顯差距；另於訪談亦發現綜合分析結果並未獲得青睞，皆顯示出使用單一或多種輔助軟體實驗，可能並不會有太大之差異。
With the rising of social thoughts, qualitative research has been gradually emphasized in the domain of social science. Qualitative interview is the most extensively used method of data collection. After qualitative interview, researchers need to create an interview manuscript and cluster the interviewees’ responses based on the theoretical framework of the research, before analyzing the collected data. Conventional manual clustering requires a large amount of time and efforts. Therefore, this study aimed to use computer assisted qualitative data analysis software (CAQDAS) to analyze interview manuscripts and investigate if consistency exists between the results of CADQAS and manual clustering. Finally, how CAQDAS could benefit the analysis of interview manuscripts would also be explained. In this study, a CAQDAS, ClustanGraphics 8.02, and a topic extraction system were utilized in the experiment on the interview manuscripts of graduates (interviewees) in the domain of social science. CAQDAS was first used to analyze the interview manuscripts and obtain computer assisted analysis results. Later, according to the analysis results, the interviewees were interviewed again, and rand index was applied to cross verify the similarity between the results of CAQDAS and manual analysis to find out the benefits of CAQDAS for the analysis of qualitative research interview manuscripts. The experimental and interview results revealed that CAQDAS and manual clustering results were not consistent but the results produced by CAQDAS could provide a primary concept of clustering for manual analysis of interview manuscripts. Moreover, if CAQDAS was applied to structured or semi-structured interview manuscripts, the results would be better. From the similarity calculated by rand index between CAQDAS and manual clustering results and comprehensive results, no significant difference was found. From the interview results, it was also found that the general analysis results were not reliable, indicating that the use of a single or multiple computer assisted software systems would not produce any significant difference.
With the rising of social thoughts, qualitative research has been gradually emphasized in the domain of social science. Qualitative interview is the most extensively used method of data collection. After qualitative interview, researchers need to create an interview manuscript and cluster the interviewees’ responses based on the theoretical framework of the research, before analyzing the collected data. Conventional manual clustering requires a large amount of time and efforts. Therefore, this study aimed to use computer assisted qualitative data analysis software (CAQDAS) to analyze interview manuscripts and investigate if consistency exists between the results of CADQAS and manual clustering. Finally, how CAQDAS could benefit the analysis of interview manuscripts would also be explained. In this study, a CAQDAS, ClustanGraphics 8.02, and a topic extraction system were utilized in the experiment on the interview manuscripts of graduates (interviewees) in the domain of social science. CAQDAS was first used to analyze the interview manuscripts and obtain computer assisted analysis results. Later, according to the analysis results, the interviewees were interviewed again, and rand index was applied to cross verify the similarity between the results of CAQDAS and manual analysis to find out the benefits of CAQDAS for the analysis of qualitative research interview manuscripts. The experimental and interview results revealed that CAQDAS and manual clustering results were not consistent but the results produced by CAQDAS could provide a primary concept of clustering for manual analysis of interview manuscripts. Moreover, if CAQDAS was applied to structured or semi-structured interview manuscripts, the results would be better. From the similarity calculated by rand index between CAQDAS and manual clustering results and comprehensive results, no significant difference was found. From the interview results, it was also found that the general analysis results were not reliable, indicating that the use of a single or multiple computer assisted software systems would not produce any significant difference.
Description
Keywords
電腦輔助質性資料分析軟體, 質性研究, 訪談稿, 文件分群, 質性資料分析, 分群分析, Computer assisted qualitative data analysis software(CAQDAS), qualitative research, interview manuscript, document clustering, qualitative data analysis, cluster analysis