Correcting the bias of the Root Mean Squared Error of Approximation under missing data
Author(s) / Creator(s)
Fitzgerald, Cailey E.
Estabrook, Ryne
Martin, Daniel P.
Brandmaier, Andreas M.
von Oertzen, Timo
Abstract / Description
Missing data are ubiquitous in psychological research. They may come about as an unwanted result of coding or computer error, participants' non-response or absence, or missing values may be intentional, as in planned missing designs. We discuss the effects of missing data on χ²-based goodness-of-fit indices in Structural Equation Modeling (SEM), specifically on the Root Mean Squared Error of Approximation (RMSEA). We use simulations to show that naive implementations of the RMSEA have a downward bias in the presence of missing data and, thus, overestimate model goodness-of-fit. Unfortunately, many state-of-the-art software packages report the biased form of RMSEA. As a consequence, the scientific community may have been accepting a much larger fraction of models with non-acceptable model fit. We propose a bias-correction for the RMSEA based on information-theoretic considerations that take into account the expected misfit of a person with fully observed data. The corrected RMSEA is asymptotically independent of the proportion of missing data for misspecified models. Importantly, results of the corrected RMSEA computation are identical to naive RMSEA if there are no missing data.
Keyword(s)
missing data structural equation modeling fit indices Kullback Leibler divergence relative entropyPersistent Identifier
Date of first publication
2021-09-30
Journal title
Methodology
Volume
17
Issue
3
Page numbers
189–204
Publisher
PsychOpen GOLD
Publication status
publishedVersion
Review status
peerReviewed
Is version of
Citation
Fitzgerald, C. E., Estabrook, R., Martin, D. P., Brandmaier, A. M., & von Oertzen, T. (2021). Correcting the bias of the Root Mean Squared Error of Approximation under missing data. Methodology, 17(3), 189-204. https://doi.org/10.5964/meth.2333
-
meth.v17i3.2333.pdfAdobe PDF - 1.16MBMD5: 73705bed64ce5c8e455dd6ef78edd77d
-
There are no other versions of this object.
-
Author(s) / Creator(s)Fitzgerald, Cailey E.
-
Author(s) / Creator(s)Estabrook, Ryne
-
Author(s) / Creator(s)Martin, Daniel P.
-
Author(s) / Creator(s)Brandmaier, Andreas M.
-
Author(s) / Creator(s)von Oertzen, Timo
-
PsychArchives acquisition timestamp2022-04-14T11:24:54Z
-
Made available on2022-04-14T11:24:54Z
-
Date of first publication2021-09-30
-
Abstract / DescriptionMissing data are ubiquitous in psychological research. They may come about as an unwanted result of coding or computer error, participants' non-response or absence, or missing values may be intentional, as in planned missing designs. We discuss the effects of missing data on χ²-based goodness-of-fit indices in Structural Equation Modeling (SEM), specifically on the Root Mean Squared Error of Approximation (RMSEA). We use simulations to show that naive implementations of the RMSEA have a downward bias in the presence of missing data and, thus, overestimate model goodness-of-fit. Unfortunately, many state-of-the-art software packages report the biased form of RMSEA. As a consequence, the scientific community may have been accepting a much larger fraction of models with non-acceptable model fit. We propose a bias-correction for the RMSEA based on information-theoretic considerations that take into account the expected misfit of a person with fully observed data. The corrected RMSEA is asymptotically independent of the proportion of missing data for misspecified models. Importantly, results of the corrected RMSEA computation are identical to naive RMSEA if there are no missing data.en_US
-
Publication statuspublishedVersion
-
Review statuspeerReviewed
-
CitationFitzgerald, C. E., Estabrook, R., Martin, D. P., Brandmaier, A. M., & von Oertzen, T. (2021). Correcting the bias of the Root Mean Squared Error of Approximation under missing data. Methodology, 17(3), 189-204. https://doi.org/10.5964/meth.2333en_US
-
ISSN1614-2241
-
Persistent Identifierhttps://hdl.handle.net/20.500.12034/5707
-
Persistent Identifierhttps://doi.org/10.23668/psycharchives.6311
-
Language of contenteng
-
PublisherPsychOpen GOLD
-
Is version ofhttps://doi.org/10.5964/meth.2333
-
Is related tohttps://doi.org/10.23668/psycharchives.5135
-
Keyword(s)missing dataen_US
-
Keyword(s)structural equation modelingen_US
-
Keyword(s)fit indicesen_US
-
Keyword(s)Kullback Leibler divergenceen_US
-
Keyword(s)relative entropyen_US
-
Dewey Decimal Classification number(s)150
-
TitleCorrecting the bias of the Root Mean Squared Error of Approximation under missing dataen_US
-
DRO typearticle
-
Issue3
-
Journal titleMethodology
-
Page numbers189–204
-
Volume17
-
Visible tag(s)Version of Recorden_US