2015
DOI: 10.2903/sp.efsa.2015.en-869
|View full text |Cite
|
Sign up to set email alerts
|

Separation Issues and Possible Solutions: Part I – Systematic Literature Review on Logistic Models ‐ Part II – Comparison of different methods for separation in logistic regression

Abstract: The separation issue is a complication frequently occurring when sparse data are analysed by logistic regression models. Such analyses of sparse datasets tend to be biased, resulting in misleading conclusions, or may even not be feasible, due to computational problems such as non-convergence. In this report, a systematic literature review (SLR) was carried out to describe the phenomenon as well as the methods that deal with the separation issue in binary response models, especially in logistic regression type … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1

Citation Types

0
3
0

Year Published

2016
2016
2023
2023

Publication Types

Select...
2
1

Relationship

0
3

Authors

Journals

citations
Cited by 3 publications
(3 citation statements)
references
References 24 publications
0
3
0
Order By: Relevance
“…Since both bivariate and multi-variate analysis returned high odds ratios and confidence intervals for the pre-covid violence variable, which can be common in multilevel modelling (Ensoy, Rakhmawati, Faes, & Aerts, 2015;Ju, Lin, Chu, Cheng, & Xu, 2020), Firth logistic regression was explored which returned smaller odds ratios (between 4 and 6) for experiencing violence during COVID-19 if a participant had experienced violence before COVID-19 restrictions. The mixed-effects models were then re-run using penalized quasilikelihood (PQL) and Bayes estimation (Supplemental Tables 7 and 8) (Benedetti, Platt, & Atherton, 2014;Bolker et al, 2009;Breslow & Clayton, 1993;Ju et al, 2020).…”
Section: Discussionmentioning
confidence: 99%
“…Since both bivariate and multi-variate analysis returned high odds ratios and confidence intervals for the pre-covid violence variable, which can be common in multilevel modelling (Ensoy, Rakhmawati, Faes, & Aerts, 2015;Ju, Lin, Chu, Cheng, & Xu, 2020), Firth logistic regression was explored which returned smaller odds ratios (between 4 and 6) for experiencing violence during COVID-19 if a participant had experienced violence before COVID-19 restrictions. The mixed-effects models were then re-run using penalized quasilikelihood (PQL) and Bayes estimation (Supplemental Tables 7 and 8) (Benedetti, Platt, & Atherton, 2014;Bolker et al, 2009;Breslow & Clayton, 1993;Ju et al, 2020).…”
Section: Discussionmentioning
confidence: 99%
“…In this study, we consider two types of penalization for explanatory methods, Firth and Log-F penalization, although many more exist. See Ensoy et al (2015) for an overview of methods often used in cases of separation, where an input feature perfectly predicts the outcome, or rare events.…”
Section: Penalized Regressionmentioning
confidence: 99%
“…As the primary objective of the survey is not to assess specific risk factors effects, sparseness/separation could be an issue when carrying out the analysis. Ensoy et. al (2015) proposed a structure and harmonised guidance on how to deal with such issues, depending on the type of analysis performed, this will be used when analysing the Nov baseline survey data.…”
Section: Plan Of Analysismentioning
confidence: 99%