The use of generalized linear models and generalized estimating equations in bioarchaeological studies.

摘要: The American School of Classical Studies at Athens, 54 Souidias Street, 106 76 AthensKEY WORDS statistical analysis; generalized linear models; entheses; archaeologicalcollectionsABSTRACT current article explores whether theapplication models (GLM) and gen-eralized estimating equations (GEE) can be used inplace conventional analyses in the study ofordinal data that code an underlying continuous vari-able, like entheseal changes. analysis artificialdata ordinal expressing changes inarchaeological North African populations gave fol-lowing results. Parametric nonparametric tests giveconvergent results particularly for P values <0.1, irre-spective variable is normallydistributed or not under condition samplesinvolved exhibit approximately equal sizes.If this prerequisite valid provided sam-ples are variances, covariance maybe adopted. GLM subject to constraints giveresults converge those obtained from all nonpara-metric tests. Therefore, they instead tra-ditional as give same amount ofinformation them, but with advantage allowingthe simultaneous impact multiple predic-tors their interactions modeling theexperimental data. However, should replaced byGEE bilateral asymmetry generalwhen paired samples tested, because GEE areappropriate correlated Am J Phys Anthropol153:473–483, 2014.