2017
DOI: 10.1002/ajpa.23206
|View full text |Cite
|
Sign up to set email alerts
|

Comparisons of statistical techniques to assess age‐related skeletal markers in bioarchaeology

Abstract: ANCOVAs or Factorial ANOVAs that incorporate age as a covariate should be considered more often in studies that test different prevalences of age-related osteological markers among past populations.

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

1
8
0

Year Published

2017
2017
2024
2024

Publication Types

Select...
5
1

Relationship

0
6

Authors

Journals

citations
Cited by 11 publications
(9 citation statements)
references
References 47 publications
(57 reference statements)
1
8
0
Order By: Relevance
“…Variation in effect sizes was expected, as ORs from the Fisher's exact test are a direct expression of the raw ratio of DJD prevalence between the two categories within each predictor, while ORs from GLMs and GLMMs are conditional on all other terms within each model. This influences their interpretation but, in our analyses, most of the results of each method were within the 95% CI of other methods (Table 5), supporting previous studies which discussed the consistency of contingency table and regression approaches in bioarchaeology (Cheverko & Hubbe, 2017;Nikita et al, 2013). However, we also observed that GLMMs resulted in larger numbers of statistically significant associations, even after multiple testing correction.…”
Section: Statistical Assessment Of Djd Prevalencesupporting
confidence: 89%
See 2 more Smart Citations
“…Variation in effect sizes was expected, as ORs from the Fisher's exact test are a direct expression of the raw ratio of DJD prevalence between the two categories within each predictor, while ORs from GLMs and GLMMs are conditional on all other terms within each model. This influences their interpretation but, in our analyses, most of the results of each method were within the 95% CI of other methods (Table 5), supporting previous studies which discussed the consistency of contingency table and regression approaches in bioarchaeology (Cheverko & Hubbe, 2017;Nikita et al, 2013). However, we also observed that GLMMs resulted in larger numbers of statistically significant associations, even after multiple testing correction.…”
Section: Statistical Assessment Of Djd Prevalencesupporting
confidence: 89%
“…Statistical methodologies are thus a core part of the assessment of DJD in past populations. Traditionally, a plethora of tests have been applied to study the distribution of this and other pathologies in the bioarchaeological literature (Cheverko & Hubbe, 2017), with the most common being contingency table tests (Chi‐square and Fisher's exact) and analyses of variance (ANOVA and ANCOVA). These tests are often used to answer broad questions, such as the existence of differences in DJD prevalence between sexes and age categories (Shimoda et al, 2012; Suzuki et al, 2016; Woo & Pak, 2013), and for this purpose their performance is equivalent (Cheverko & Hubbe, 2017).…”
Section: Introductionmentioning
confidence: 99%
See 1 more Smart Citation
“…ANCOVA adjusts the mean of the dependent variable once the variation of the covariate has been accounted for (both mean and adjusted mean are reported in Tables 5 and 6). Others have found ANCOVA to be more effective than alternative nonparametric tests (Cheverko et al 2016;Hubbe et al 2012).…”
Section: Methodsmentioning
confidence: 99%
“…However, these procedures can incur the loss of a substantial part of the individuals available for study, and thus the use of statistical approaches based on the linear regression model (including classic ANOVA designs) has been preferred. The use of many of these methods for bioarchaeological research has been reviewed and discussed elsewhere (Alonso‐Llamazares et al, 2021; Cheverko & Hubbe, 2017; Nikita et al, 2013). Nevertheless, a prevailing drawback of the usual framework for these analyses, based on “frequentist” null hypothesis significance testing, is that each new study carried out disregards that a large body of evidence might already exist on the questions being explored (van de Schoot et al, 2014).…”
Section: Introductionmentioning
confidence: 99%