不同測量方式下潛在調節效果估計的比較

The Comparison of Estimated Moderating Effect with Latent Variables

10.6147/JHRM.2015.1503.03

林秀娟(Hsiu-Chuan Lin)；陳順宇(Shun-Yu Chen)

調節因子 ； 調節效果 ； 潛在變數 ； 調節迴歸分析 ； 結構方程模式 ； moderator ； moderation effect ； latent variable ； moderated regression analysis ； structural equation modeling

人力資源管理學報

15卷3期（2015 / 09 / 01）

53 - 74

繁體中文

當潛在自變數及／或調節因子有多個問項時，則其交互作用項也是潛在變數，但有些研究者會將每個量表的多個問項分數分別平均，各得到單一的分數，而交互作用項以自變數與調節因子的個別平均數相乘項當測量變數，然後進行傳統的迴歸分析，此種方式稱為單一測量法（或稱調節迴歸分析；moderated regression analysis, MRA）。另一種測量方式是將自變數、依變數、調節因子等三個量表每個量表的多個問項分成兩組，分別各以所得兩組平均分數當測量變數，而交互作用項是以不重疊的部分配對做為測量變數，此種方式稱為兩次測量法。本研究以模擬資料利用Amos 的結構方程模式（structural equation modeling, SEM），比較兩次測量法與單一測量法在調節效果估計、調節效果估計標準誤、調節模式解釋能力（R^2）及調節效果顯著性等四項的表現。研究結果發現兩次測量法在調節效果接近不偏的估計，單一測量法調節效果是偏的估計，且有低估現象，尤其當測量信度低及自變數與調節因子的相關係數小時低估更嚴重。兩次測量法模式解釋能力及調節效果估計標準誤都比單一測量法大，至於調節效果顯著性檢定兩種方法差別不大。

When independent variable and/or moderator have multiple items, the interaction between independent variable and moderator is also a latent variable. Some researchers using the means of multiple items as measurement of independent variable and/or moderator, then using these means product as measurement of interaction. Such an approach is called single-measurement approach (or moderated regression analysis, MRA). Another measurement method is called two-measurement approach. In this method, for each independent variable, dependent variable, and moderator, we group their multiple measurement items into two groups and use their average as the measurement variables. While their non-overlapping interaction term is paired as measurement variables. Structural Equation Modeling (SEM) has been suggested to study the moderating effect with latent variables. This paper compares two-measurement approach and single-measurement approach, including estimated moderating effect, estimated standard deviation, coefficient of determination (R^2) and significance level. The results show that the estimated moderating effect under the unconstrained approach is nearly unbiased but it is biased and possibly underestimated under the traditional approach. While the two approaches perform similarly in the significance testing for moderating effect, unconstrained approach results in higher R2 and estimated standard deviation.

1. Baron, R. M.,Kenny, D. A.(1986).The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations.Journal of Personality and Social Psychology,51,1173-1182.
2. Batista-Foguet, J. M.,Coenders, G.,Saris, W. E.(2004).A parsimonious approach to interaction effects in structural equation models: An application to consumer behavior.Working Paper d'ESADE,183,1-28.
3. Coenders, G. M.,Batista-Foguet, J. M.,Saris, W.(2008).Simple, efficient and distribution-free approach to interaction effects in complex structural equation models.Quality & Quantity,42,369-396.
4. Cohen, J.,Cohen, P.(1983).Applied multiple regression/correlation analysis for the behavior sciences.Hillsdale, NJ:Erlbaum.
5. Conner, M.,Sparks, P.,Povey, R.,James, R.,Shepherd, R.,Armitage, C. J.(2002).Moderator effects of attitudinal ambivalence on attitude-behaviour relationships.European Journal of Social Psychology,32,705-718.
6. Frazier, P. A.,Tix, A. P.,Barron, K. E.(2004).Testing moderator and mediator effects in counseling psychology research.Journal of Counseling Psychology,51,115-134.
7. Irwin, J. R.,McClelland, G. H.(2001).Misleading heuristics and moderate regression models.Journal of Marketing Research,38,100-109.
8. Jaccard, J.,Wan, C. K.(1995).Measurement error in the analysis of interaction effects between continuous predictors using multiple regression.Psychological Bulletin,117,348-357.
9. Kenny, D. A.,Judd, C. M.(1984).Estimating the nonlinear and interactive effects of latent variables.Psychological Bulletin,96,201-210.
10. Lin, G. C.,Wen, Z.,Marsh, H. W.,Lin, H. S.(2010).Structural equation models of latent interactions: Clarification of orthogonalizing and double-mean-centering strategies.Structural Equation Modeling,17,374-391.
11. Marcoulides, G. A.(Ed.),Schumacker, R. E.(Ed.)(1996).Advanced structural equation modeling: Issues and techniques.Mahwah, NJ:Erlbaum.
12. Marsh, H. W.,Wen, Z.,Hau, K. T.(2004).Structural equation models of latent interactions: Evaluation of alternative estimation strategies and indicator construction.Psychological Methods,9,275-300.
13. Moulder, B. C.,Algina, J.(2002).Comparison of methods for estimating and testing latent variable interactions.Structural Equation Modeling,9,1-19.
14. Newsom, J. T.,Prigerson, H. G.,Schulz, R.,Reynolds, C. F.(2003).Investigating moderator hypotheses in aging research: Statistical, methodological, and conceptual difficulties with comparing separate regressions.International Journal of Aging and Human Development,57,119-150.
15. Ping, R. A.(1996).Latent variable interaction and quadratic effect estimation: A two-step technique using structural equation analysis.Psychological Bulletin,119,166-175.
16. Schumacker, R. E.(Ed.),Marcoulides, G. A.(Ed.)(1998).Interaction and nonlinear effects in structural equation modeling.Mahwah, NJ:Erlbaum.

1. (2024)。粉絲對YouTube旅遊網紅之擬社會互動和行為反應：整合來源吸引性與內容滿足之觀點。電子商務學報，26(1)，41-74。