Research questions about how and why health trends differ between populations require decisions about data analytic procedure. The objective was to document and compare the information returned from stratified, fixed effect and random effect approaches to data modelling for two prototypical descriptive research questions about comparative trends in toothbrushing.
Data included five cycles of the Health Behaviour in School-aged Children 2006 to 2022, which provided a sample of 980192 11- to 15- year olds from 35 countries. Using logistic regression models and generalized linear mixed models, toothbrushing daily was regressed on time, following the three approaches to analysis of trends.
The stratified approach suggested a positive but non-linear trend in toothbrushing from 2006 to 2022 in most countries but provided no statistical inference on the variation. The fixed effect and the random effect approach converged on a positive but flattening overall trend, with a statistically significant country variation in trends.
Only the fixed effect approach and the random effects approach provided clear answers to the research question. Additional methodological considerations for making an informed choice of analytical approach are discussed.