利用多層線性模式瞭解題目無反應

Using Multilevel Modeling to Understand Item Nonresponse

楊孟麗(Meng-Li Yang)

《調查研究》 12期 (2002/10) Pp. 59-90

https://doi.org/10.7014/TCYC.200210.0059

摘要

研究己顯示，題目無反應的形成與受訪者及題目的特質有關，若不加以瞭解，可能會使研究結果產生偏誤。另一方面，也發現訪員特質與受訪者在某些特殊議題的回答行為有關。因此討論題目無反應時，必須將訪員、受訪者與題目，三方區的特質同時納入考慮。已有研究者將多層線性模式（HLM）應用於題目無反應，但這些研究只限於探討題目與回答者兩者，或受訪者與訪員兩者之間的互動，或者無反應的總數與受訪者及訪員之間的關係。本研究以HLM中的三層模式來探討題目特質、受訪者與訪員三者對於無反應的影響。所使用的方法是先依據題目的特質（面向）分類後，再以三層模式分析這些面向中的無反應，同時瞭解訪員特質、受訪者特質、與題目面向三者的影響。我以分析台灣地區社會變遷基本調查第三期四次的傳播組中的十個題目做為例子，並討論其結果。

關鍵字

多層線性模式 ； 題項無反應 ； 訪員特色 ； 受訪者特色

並列摘要

Item nonresponse has been shown related to respondent and item characteristics. On the other hand, interviewer characteristics also have been found related to response behavior on some special issues. Therefore, in studying item nonresponse, we need to take into consideration characteristics of all three above, namely, the interviewer, the respondent and the items. To solve the problem, this study groups items based on what they measure or what mechanism they might activate (dimension), and then uses the 3-level logistic model to analyze them, so that the three-way interaction among the items' dimensions, the respondent and the interviewer can be investigated. Nonresponse from 10 questions in the Mass Media Volume (volume 1) of the Taiwan Social Change fourth survey of the third period are analyzed as an example. Results and implications are discussed.

並列關鍵字

HLM ； multilevel logistic model ； item nonresponse ； interviewer characteristics ； respondent characteristics

參考文獻

Anderson, B. A.,Silver, B. D.,Abramson, P. R.(1988).The Effects of Race of the Interviewer on Measures of Electroal Participation by Blacks in SRC National Election Studies.Public Opinion Quarterly.52,53-83.

Berinsky, A. J.(1999).The Two Faces of Public Opinion.American Journal of Political Science.43,1209-1230.

Blair, E.(1979).Improving Interview Method and Questionnaire Design.San Francisco:Jossey-Bass.

Borgers, N.,Hox, J.(2001).Item Nonresponse in Questionnaire Research with Children.Journal of Official Statistics.17,321-335.

Bryk, A. S.,Raudenbush, S. W.(1992).Hierarchical Linear Models.Newbury Park:Sage Publications.