Generalised Estimating Equations
The study selected an exchangeable working correlation matrix over AR(1) because it had lower QICC. GEE modelling was considered appropriate for within-subject correlation in the repeated-measures data. The binomial logit model allowed exp…
1 sources - 4 claims
The study selected an exchangeable working correlation matrix over AR(1) because it had lower QICC. GEE modelling was considered appropriate for within-subject correlation in the repeated-measures data. The binomial logit model allowed exponentiated coefficients to be interpreted as odds ratios. Generalised Estimating Equations were used to model longitudinal associations while accounting for repeated measures within participants.