Unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices

Nested data structures create statistical dependence that influences the effective sample size and statistical power of a study. Several methods are available for dealing with nested data, including the summary-statistics approach and multilevel modelling (MLM). Recent publications have heralded MLM...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Carolyn Beth McNabb, Kou Murayama
Formato: article
Lenguaje:EN
Publicado: Elsevier 2021
Materias:
Acceso en línea:https://doaj.org/article/0bf4faf0e04744bbaca58bde1a357f07
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:0bf4faf0e04744bbaca58bde1a357f07
record_format dspace
spelling oai:doaj.org-article:0bf4faf0e04744bbaca58bde1a357f072021-11-24T04:34:24ZUnnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices2665-945X10.1016/j.crneur.2021.100024https://doaj.org/article/0bf4faf0e04744bbaca58bde1a357f072021-01-01T00:00:00Zhttp://www.sciencedirect.com/science/article/pii/S2665945X21000206https://doaj.org/toc/2665-945XNested data structures create statistical dependence that influences the effective sample size and statistical power of a study. Several methods are available for dealing with nested data, including the summary-statistics approach and multilevel modelling (MLM). Recent publications have heralded MLM as the best method for analysing nested data, claiming benefits in power over summary-statistics approaches (e.g., the t-test). However, when cluster size is equal, these approaches are mathematically equivalent. We conducted statistical simulations demonstrating equivalence of MLM and summary-statistics approaches for analysing nested data and provide supportive cases for the utility of the conventional summary-statistics approach in nested experiments. Using statistical simulations, we demonstrate that losses in power in the summary-statistics approach discussed in the previous literature are unsubstantiated. We also show that MLM sometimes suffers from frequent singular fit errors, especially when intraclass correlation is low. There are indeed many situations in which MLM is more appropriate and desirable, but researchers should be aware of the possibility that simpler analysis (i.e., summary-statistics approach) does an equally good or even better job in some situations.Carolyn Beth McNabbKou MurayamaElsevierarticleHierarchical linear modelMixed modelt-testClusteringNeurosciences. Biological psychiatry. NeuropsychiatryRC321-571ENCurrent Research in Neurobiology, Vol 2, Iss , Pp 100024- (2021)
institution DOAJ
collection DOAJ
language EN
topic Hierarchical linear model
Mixed model
t-test
Clustering
Neurosciences. Biological psychiatry. Neuropsychiatry
RC321-571
spellingShingle Hierarchical linear model
Mixed model
t-test
Clustering
Neurosciences. Biological psychiatry. Neuropsychiatry
RC321-571
Carolyn Beth McNabb
Kou Murayama
Unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices
description Nested data structures create statistical dependence that influences the effective sample size and statistical power of a study. Several methods are available for dealing with nested data, including the summary-statistics approach and multilevel modelling (MLM). Recent publications have heralded MLM as the best method for analysing nested data, claiming benefits in power over summary-statistics approaches (e.g., the t-test). However, when cluster size is equal, these approaches are mathematically equivalent. We conducted statistical simulations demonstrating equivalence of MLM and summary-statistics approaches for analysing nested data and provide supportive cases for the utility of the conventional summary-statistics approach in nested experiments. Using statistical simulations, we demonstrate that losses in power in the summary-statistics approach discussed in the previous literature are unsubstantiated. We also show that MLM sometimes suffers from frequent singular fit errors, especially when intraclass correlation is low. There are indeed many situations in which MLM is more appropriate and desirable, but researchers should be aware of the possibility that simpler analysis (i.e., summary-statistics approach) does an equally good or even better job in some situations.
format article
author Carolyn Beth McNabb
Kou Murayama
author_facet Carolyn Beth McNabb
Kou Murayama
author_sort Carolyn Beth McNabb
title Unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices
title_short Unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices
title_full Unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices
title_fullStr Unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices
title_full_unstemmed Unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: When a traditional summary-statistics approach suffices
title_sort unnecessary reliance on multilevel modelling to analyse nested data in neuroscience: when a traditional summary-statistics approach suffices
publisher Elsevier
publishDate 2021
url https://doaj.org/article/0bf4faf0e04744bbaca58bde1a357f07
work_keys_str_mv AT carolynbethmcnabb unnecessaryrelianceonmultilevelmodellingtoanalysenesteddatainneurosciencewhenatraditionalsummarystatisticsapproachsuffices
AT koumurayama unnecessaryrelianceonmultilevelmodellingtoanalysenesteddatainneurosciencewhenatraditionalsummarystatisticsapproachsuffices
_version_ 1718415898311655424