On the performance of Bayesian approaches in small samples : A comment on Smid, McNeish, Miocevic, and van de Schoot (2020). / Zitzmann, Steffen; Lüdtke, Oliver; Robitzsch, Alexander et al.
In: Structural Equation Modeling: A Multidisciplinary Journal, Vol. 28, No. 1, 02.01.2021, p. 40-50.Research output: Contribution to journal › Journal article › Research › peer-review
}
TY - JOUR
T1 - On the performance of Bayesian approaches in small samples
T2 - A comment on Smid, McNeish, Miocevic, and van de Schoot (2020)
AU - Zitzmann, Steffen
AU - Lüdtke, Oliver
AU - Robitzsch, Alexander
AU - Hecht, Martin
PY - 2021/1/2
Y1 - 2021/1/2
N2 - This journal recently published a systematic review of simulation studies on the performance of Bayesianapproaches for estimating latent variable models in small samples. The authors of this review high-lighted that Bayesian approaches can perform poorly (i.e., by exhibiting bias) when the prior distribu-tions are not thoughtfully constructed on the basis of previous knowledge. In this comment, wequestion whether the bias is the most important criterion when the sample size is small. We arguethat the variability is more important and should therefore not be ignored. Moreover, because one ofthe most important selling points of Bayesian approaches was not addressed in the article, we arguethat although somewhat biased, Bayesian approaches allow for more accurate estimates (i.e., a smallermean squared error) than Maximum Likelihood (ML) in small samples, and we show one such approachthat is more accurate than ML.
AB - This journal recently published a systematic review of simulation studies on the performance of Bayesianapproaches for estimating latent variable models in small samples. The authors of this review high-lighted that Bayesian approaches can perform poorly (i.e., by exhibiting bias) when the prior distribu-tions are not thoughtfully constructed on the basis of previous knowledge. In this comment, wequestion whether the bias is the most important criterion when the sample size is small. We arguethat the variability is more important and should therefore not be ignored. Moreover, because one ofthe most important selling points of Bayesian approaches was not addressed in the article, we arguethat although somewhat biased, Bayesian approaches allow for more accurate estimates (i.e., a smallermean squared error) than Maximum Likelihood (ML) in small samples, and we show one such approachthat is more accurate than ML.
KW - Bayesian estimation
KW - Markov chain Monte Carlo
KW - structural equation modeling
KW - small sample
U2 - 10.1080/10705511.2020.1752216
DO - 10.1080/10705511.2020.1752216
M3 - Journal article
VL - 28
SP - 40
EP - 50
JO - Structural Equation Modeling: A Multidisciplinary Journal
JF - Structural Equation Modeling: A Multidisciplinary Journal
SN - 1070-5511
IS - 1
ER -
ID: 1388370