Objective: In mediation analysis, the use of effect size measures is extremely important to understand the strength and direction of the relationship between variables in depth and to determine the importance of the mediating variable effect. In the literature, it is seen that there are few studies comparing the performance of effect size measures for mediation analysis. The aim of this study is to investigate the relationships between continuous variables and to compare the performances of effect size measures for mediation model. Material and Methods: In line with the objective of the study, the performance of effect size measures for the mediation model was examined through a simulation study, considering different sample sizes and small, medium, and large effect sizes. The comparison of effect size measures for the mediation model was conducted by examining bias values. Results: For the mediation model, it was observed that R2 , as a measure of explained variance, had the least bias across all scenarios considered in the simulation. While mediation ratio measures required a minimum sample size of 500, R2 as a measure of explained variance exhibited good performance even with smaller sample sizes, such as 100. Conclusion: In models involving mediator variables, it is recommended to use alternative effect size measures in research, in addition to a single measure, to comprehensively capture the strength of the relationship between variables.
Keywords: Mediation analysis; mediator variable; effect size; mediation proportion; explained variance
Amaç: Aracı değişken içeren modellerde, değişkenler arasındaki ilişkinin gücünü ve yönünü derinlemesine anlamanın yanı sıra aracı değişken etkisinin önemini ortaya koymak için etki büyüklüğü ölçülerine yer verilmesi son derecede önemlidir. Literatürde aracılık analizi için etki büyüklüğü ölçülerinin performanslarının karşılaştırmalı olarak ele alındığı çalışma sayısının az olduğu görülmektedir. Bu çalışmanın amacı, sürekli türdeki değişkenler arasındaki ilişkilerin araştırılması ve aracılık modeli için etki büyüklüğü ölçülerinin performanslarının karşılaştırılmasıdır. Gereç ve Yöntemler: Çalışmanın amacı doğrultusunda, farklı örnek genişliklerinde ve küçük, orta, geniş etki büyüklüğü durumlarında aracılık modeli için etki büyüklüğü ölçülerinin performansları bir simülasyon çalışması ile incelenmiştir. Aracılık modeli için etki büyüklüğü ölçülerinin performans karşılaştırması yanlılık değerleri göz önünde bulundurularak yapılmıştır. Bulgular: Aracılık modeli için simülasyonda ele alınan tüm senaryolarda R2 açıklanan varyans ölçülerinin en az yanlılığa sahip olduğu görülmüştür. Aracılık oran ölçüleri için en az 300 örnek genişliği gerekirken, R2 açıklanan varyans ölçüsünün 100 gibi daha küçük örnek genişliklerinde de iyi bir performansa sahip olduğu görülmüştür. Sonuç: Aracı değişkenli modellerde değişkenler arasındaki ilişkinin gücünün daha kapsamlı şekilde ele alınabilmesi için tek bir etki büyüklüğü ölçüsü yerine alternatif ölçülerin de araştırmalarda kullanılması önerilmektedir.
Anahtar Kelimeler: Aracılık analizi; aracı değişken; etki büyüklüğü; aracılık oranı; açıklanan varyans
- Zhang Z, Zheng C, Kim C, Van Poucke S, Lin S, Lan P. Causal mediation analysis in the context of clinical research. Ann Transl Med. 2016;4(21):425. [Crossref] [PubMed] [PMC]
- Memona MA, Cheahb JH, Ramayah T, Ting H, Chuahe F. Mediation analysis issues and recommendations. Journal of Applied Structural Equation Modeling. 2018;2(1):i-ix. [Crossref]
- Gunzler D, Chen T, Wu P, Zhang H. Introduction to mediation analysis with structural equation modeling. Shanghai Arch Psychiatry. 2013;25(6):390-4. [PubMed] [PMC]
- Walters GD. Why are mediation effects so small? International Journal of Social Research Methodology. 2019;22(2):219-32. [Crossref]
- Lachowicz MJ, Preacher KJ, Kelley K. A novel measure of effect size for mediation analysis. Psychol Methods. 2018;23(2):244-61. [Crossref] [PubMed]
- Ditlevsen S, Christensen U, Lynch J, Damsgaard MT, Keiding N. The mediation proportion: a structural equation approach for estimating the proportion of exposure effect on outcome explained by an intermediate variable. Epidemiology. 2005;16(1):114-20. [Crossref] [PubMed]
- Fairchild AJ, Mackinnon DP, Taborga MP, Taylor AB. R2 effect-size measures for mediation analysis. Behav Res Methods. 2009;41(2):486-98. [Crossref] [PubMed] [PMC]
- Hayes AF, Preacher KJ. Quantifying and testing indirect effects in simple mediation models when the constituent paths are nonlinear. Multivariate Behav Res. 2010;45(4):627-60. [Crossref] [PubMed]
- MacKinnon DP, Fairchild AJ, Fritz MS. Mediation analysis. Annu Rev Psychol. 2007;58:593-614. [Crossref] [PubMed] [PMC]
- Hayes AF, Rockwood NJ. Regression-based statistical mediation and moderation analysis in clinical research: Observations, recommendations, and implementation. Behav Res Ther. 2017;98:39-57. [Crossref] [PubMed]
- Gu J, Strauss C, Bond R, Cavanagh K. How do mindfulness-based cognitive therapy and mindfulness-based stress reduction improve mental health and wellbeing? A systematic review and meta-analysis of mediation studies. Clin Psychol Rev. 2015;37:1-12. Erratum in: Clin Psychol Rev. 2016;49:119. [Crossref] [PubMed]
- MacKinnon D. Introduction to Statistical Mediation Analysis. 1st ed. New York: Routledge; 2008. [Crossref]
- Preacher KJ, Kelley K. Effect size measures for mediation models: quantitative strategies for communicating indirect effects. Psychol Methods. 2011;16(2):93-115. [Crossref] [PubMed]
- Miočević M, O'Rourke HP, MacKinnon DP, Brown HC. Statistical properties of four effect-size measures for mediation models. Behav Res Methods. 2018;50(1):285-301. [Crossref] [PubMed] [PMC]
- Fairchild AJ, MacKinnon DP. A general model for testing mediation and moderation effects. Prev Sci. 2009;10(2):87-99. [Crossref] [PubMed] [PMC]
- Fairchild AJ, McDaniel HL. Best (but oft-forgotten) practices: mediation analysis. Am J Clin Nutr. 2017;105(6):1259-71. [Crossref] [PubMed] [PMC]
- Lachowicz M. A general measure of effect size for mediation analysis [Doctoral thesis]. Nashville, Tennessee: Vanderbilt University; 2018. [Link]
- Field-Fote EE. Mediators and moderators, confounders and covariates: exploring the variables that illuminate or obscure the "active ingredients" in neurorehabilitation. J Neurol Phys Ther. 2019;43(2):83-4. [Crossref] [PubMed]
- de Heus P. R squared effect-size measures and overlap between direct and indirect effect in mediation analysis. Behav Res Methods. 2012;44(1):213-21. [Crossref] [PubMed] [PMC]
- Hair Jr JF, Hult GTM, Ringle CM, Sarstedt M, Danks NP, Ray S. Mediation analysis. Partial Least Squares Structural Equation Modeling (PLS-SEM) Using R: A Workbook. 1st ed. Switzerland: Springer; 2021. p.139-53. [Crossref]
- Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York: Routledge; 1988.
- Sayın A, Gelbal S. Yapısal eşitlik modellemesinde parametrelerin klasik test kuramı ve madde tepki kuramına göre sınırlandırılmasının uyum indekslerine etkisi [Structural equation modeling effect of restricted with parameters from the classical test theory and item response theory]. Uluslararası Eğitim Bilim ve Teknoloji Dergisi. 2016;2(2):57-71. [Link]
- Tingley D, Yamamoto T, Hirose K, Keele L, Imai K. Mediation: R package for causal mediation analysis. Journal of Statistical Software. 2014;59(5):1-36. [Crossref]
- Tomcza, M, Tomczak E. The need to report effect size estimates revisited. An overview of some recommended measures of effect size. Trends in Sport Sciences. 2014;21(1):19-25. [Link]
- Sobel ME. Asymptotic intervals for indirect effects in structural equations models. In: Leinhart S, ed. Sociological Methodology. 1st ed. San Francisco: Jossey-Bass; 1982. p.290-312. [Crossref]
- MacKinnon DP, Dwyer JH. Estimating mediated effects in prevention studies. Evaluation Review. 1993;17(2):144-58. [Crossref]
- Preacher KJ. Quantifying parsimony in structural equation modeling. Multivariate Behav Res. 2006;41(3):227-59. [Crossref] [PubMed]
- Mackinnon DP, Warsi G, Dwyer JH. A simulation study of mediated effect measures. Multivariate Behav Res. 1995;30(1):41. [Crossref] [PubMed] [PMC]
.: Process List