Journal of the Korean Data and Information Science Society

The Korean Data and Information Science Society (한국데이터정보과학회)
권호 표지
DOI QR코드

DOI QR Code

A study on the performance of three methods of estimation in SEM under conditions of misspecification and small sample sizes

모형명세화 오류와 소표본에서 구조방정식모형 모수추정 방법들 비교: 모수추정 정확도와 이론모형 검정력을 중심으로

  • Received : 2017.06.09
  • Accepted : 2017.09.21
  • Published : 2017.09.30
Download PDF
⟨ Previous Next ⟩

Abstract

Structural equation modeling (SEM) is a basic tool for testing theories in a variety of disciplines. A maximum likelihood (ML) method for parameter estimation is by far the most widely used in SEM. Alternatively, two-stage least squares (2SLS) estimator has been proposed as a more robust procedure to address model misspecification. A regularized extension of 2SLS, two-stage ridge least squares (2SRLS) has recently been introduced as an alternative to ML to effectively handle the small-sample-size issue. However, it is unclear whether and when misspecification and small sample sizes may pose problems in theory testing with 2SLS, 2SRLS, and ML. The purpose of this article is to evaluate the three estimation methods in terms of inferences errors as well as parameter recovery under two experimental conditions. We find that: 1) when the model is misspecified, 2SRLS tends to recover parameters better than the other two estimation methods; 2) Regardless of specification errors, 2SRLS produces small or relatively acceptable Type II error rates for the small sample sizes.

구조방정식모형은 사회과학 및 행동과학 연구 분야에서 이론검정을 위해 주로 사용되는 통계방법이다. 최근 이 통계기법에 대한 방법론적 이슈로서 모형명세화 오류와 소표본 문제가 부각되고 있다. 그런데 이 문제들이 구조방정식모형의 대표 추정 방법인 최대우도법에 위한 이론검정에 어떤 영향을 주는지에 대해 여전히 명확하지 않다. 따라서 본 연구에서 최대우도법 그러고 이에 대한 대안으로 개발된 2단계최소자승법과 2단계능형최소자승법을 정확도와 검정력 관점에서 시뮬레이션을 통해 체계적으로 비교해 본다. 이 실험 결과에 따르면, 모형이 정확하게 설정된 경우, 정확도 기준에서 추정방법들 간의 차이는 미미했다. 하지만 모형오류가 발생한 경우, 2단계능형최소자승법은 다른 방법들보다 표본 크기가 작을 때 훨씬 더 정확한 모수추정치를 산출해 내었다. 그러고 이 방법은 명세화 오류에 관계없이 표본 크기가 작을 때에도 제 2종 오류 (Type II error) 수준이 상대적으로 작거나 만족할만한 수준의 검정력을 보여주었다. 이에 반해 다른 두 방법들은 표본이 작은 경우 또는 명세화 오류가 있는 경우 상당히 높은 수준의 제 2종 오류를 나타내었다.

Keywords

References

  1. Ansari, A., Kamel J. and Sharan J. (2000). A hierarchical Bayesian methodology for treating heterogeneity in structural equation models. Marketing Science, 19, 328-347. https://doi.org/10.1287/mksc.19.4.328.11789
  2. Bagozzi, R. P. and Yi, Y. (2012). Specification, evaluation, and interpretation of structural equation models. Journal of the Academy of Marketing Science, 40, 8-34. https://doi.org/10.1007/s11747-011-0278-x
  3. Baron, R. M. and 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. https://doi.org/10.1037/0022-3514.51.6.1173
  4. Bollen, K. A. (1989). Structural equations with latent variables, John Wiley & Sons, New York.
  5. Bollen, K. A. (1996). An alternative two stage least squares (2SLS) estimator for latent variable equations. Psychometrika, 61, 109-121. https://doi.org/10.1007/BF02296961
  6. Bollen, K. A. and Bauer, D. J. (2004). Automating the selection of model-implied instrumental variables. Sociological Methods and Research, 32, 425-452. https://doi.org/10.1177/0049124103260341
  7. Bollen, K. A., Kirby, J. B., Curran, P. J., Paxton, P. and Chen, F. (2007). Latent variable models under misspecification: Two-stage least squares (2SLS) and maximum likelihood (ML) estimators. Sociological Methods and Research, 36, 48-86. https://doi.org/10.1177/0049124107301947
  8. Boomsma, A. and Hoogland, J. J. (2001). The robustness of LISREL modeling revisited. In Structural Equation Modeling: Present and Future (Cudeck, R., Du Toit, S. & Sorbom, D., eds), SSI Scientific Software, Chicago, 139-168.
  9. Budaev, S. V. (2010). Using principal components and factor analysis in animal behaviour research: Caveats and guidelines. Ethology, 116, 472-480. https://doi.org/10.1111/j.1439-0310.2010.01758.x
  10. Chen, F., Bollen, K., Paxton, P., Curran, P. J. and Kirby, J. (2001). Improper solutions in structural equation models: Causes, consequences, and strategies. Sociological Methods and Research, 29, 468-508. https://doi.org/10.1177/0049124101029004003
  11. Choi, H. S., Kwon, Y. J. and Ha, J. C. (2013). Study of university students' perceptions on participation in elections via structural equation model-Focusing on K university students. Journal of the Korean Data & Information Science Society, 24, 379-390. https://doi.org/10.7465/jkdi.2013.24.2.379
  12. Cohen, J. (1988). Statistical power for the behavioral sciences, 2nd ed., Lawrence Erlbaum, New Jersey.
  13. Efron, B. (1982). The jackknife, the bootstrap and other resampling Plans, SIAM, Philadelphia.
  14. Fan, X. and Sivo, S. A. (2005). Sensitivity of fit indices to misspecified structural or measurement model components: Rationale of two-index strategy revisited. Structural Equation Modeling, 12, 343-367. https://doi.org/10.1207/s15328007sem1203_1
  15. Fox, J. (2006). Structural equation modeling with the sem package in R . Structural Equation Modeling, 13, 465-486. https://doi.org/10.1207/s15328007sem1303_7
  16. Grace, J. B., Anderson, T. M., Olff, H. and Scheiner, S. M. (2010). On the specification of structural equation models for ecological systems. Ecological Monographs, 80, 67-87. https://doi.org/10.1890/09-0464.1
  17. Grewal, R., Cote, J. A. and Baumgartner, H. (2004). Multicollinearity and measurement error in structural equation models: Implications for theory testing. Marketing Science, 23, 519-529. https://doi.org/10.1287/mksc.1040.0070
  18. Hair, J. F., Ringle, C. M. and Sarstedt, M. (2011). PLS-SEM: Indeed a silver bullet. Journal of Marketing Theory and Practice, 19, 139-151. https://doi.org/10.2753/MTP1069-6679190202
  19. Hastie, T., Tibshirani, R. and Friedman, J. (2001). The elements of statistical learning; Data mining, inference, and prediction, Sringer-Verlag, New York.
  20. Hayduk, L., Cummings, G. G., Boadu, K., Pazderka-Robinson, H. and Boulianne, S. (2007). Testing! testing! one, two three-Testing the theory in structural equation models. Personality and Individual Differences, 42, 841-850. https://doi.org/10.1016/j.paid.2006.10.001
  21. Henseler, J. (2012). Why generalized structured component analysis is not universally preferable to structural equation modeling. Journal of the Academy of Marketing Science, 40, 402-413. https://doi.org/10.1007/s11747-011-0298-6
  22. Hoerl, A. E. and Kennard, R. W. (1970). Ridge regression: Biased estimation for nonorthogonal problems. Technometrics, 12, 55-67. https://doi.org/10.1080/00401706.1970.10488634
  23. Hong, Y., Jang, G. and Choi, C. (2016). Life satisfaction and self-esteem of children from low-income class: Testing mediation model of depression. Journal of the Korean Data & Information Science Society, 27, 179-189. https://doi.org/10.7465/jkdi.2016.27.1.179
  24. James, L. R., Mulaik, S. A. and Brett, J. M. (2006). A tale of two methods. Organizational Research Methods, 9, 233-244. https://doi.org/10.1177/1094428105285144
  25. Jarvis, C. B., MacKenzie, S. B. and Podsakoff, P. M. (2003). A critical review of construct indicators and measurement model misspecification in marketing and consumer research. Journal of Consumer Research, 30, 199-218. https://doi.org/10.1086/376806
  26. Jung, S. (2013). Structural equation modeling with small sample sizes using two-stage least-squares estimation. Behavior Research Methods, 45, 75-81. https://doi.org/10.3758/s13428-012-0206-0
  27. Kaplan, D. (2009). Structural equation modeling: Foundations and extensions, 2nd ed., Sage, California.
  28. Lee, S. (2007). Problems caused by model equivalence in developing and testing models. Journal of Educational Evaluation, 20, 125-146.
  29. McArdle, J. J. and McDonald, R. P. (1984). Some algebraic properties of the reticular action model. British Journal of Mathematical and Statistical Psychology, 37, 234-251. https://doi.org/10.1111/j.2044-8317.1984.tb00802.x
  30. MacKenzie, S., Podsakoff, P. and Jarvis, C. (2005). The problem of measurement model misspecification in behavioural and organizational research and some recommended solutions. Journal of Applied Psychology, 90, 710-730. https://doi.org/10.1037/0021-9010.90.4.710
  31. Min, D. K. and Choi, M. K. (2016). How depression affects girls who experienced violence in home or at school: Using mixed model. Journal of the Korean Data & Information Science Society, 27, 101-110. https://doi.org/10.7465/jkdi.2016.27.1.101
  32. Olsson, U. H., Foss, T., Troye, S. and Howell, R. (2000). The Performance of ML, GLS, and WLS estimation in structural equation modeling under conditions of misspecification and non-normality. Structural Equation Modeling, 7, 557-595. https://doi.org/10.1207/S15328007SEM0704_3
  33. Sideridis, G., Simos, P., Papanicolaou, A. and Fletcher, J. (2014). Using structural equation modeling to assess functional connectivity in the brain: Power and sample size considerations. Educational & Psychological Measurement, 74, 733-758. https://doi.org/10.1177/0013164414525397
  34. Steenkamp, J. B. E. M. and Baumgartner, H. (2000). On the use of structural equation models for marketing modeling. International Journal of Research in Marketing, 17, 195-202. https://doi.org/10.1016/S0167-8116(00)00016-1
  35. Tenenhaus, M., Pages, J., Ambroisine, L. and Guinot, C. (2005). PLS methodology to study relationships between hedonic judgments and product characteristics. Food Quality and Preference, 16, 315-325. https://doi.org/10.1016/j.foodqual.2004.05.013
  36. Tenenhaus, M. (2008). Structural equation modeling for small samples, HEC school of management (GRECHEC).

Cited by

  1. A Comparison of Regularized Maximum-Likelihood, Regularized 2-Stage Least Squares, and Maximum-Likelihood Estimation with Misspecified Models, Small Samples, and Weak Factor Structure vol.56, pp.4, 2021, https://doi.org/10.1080/00273171.2020.1753005