I am looking to see the association between metabolites and disease. It doesn't necessarily matter whether I have the metabolites as a predictor or as the response, I am just looking to see whether there is a significant association and the direction ( negative or positive effect size).
I have both fixed effects ( age, BMI, medication usage) and random effects (family ID).
Please could anybody tell me whether there are any benefits over utilizing the lmer over glmer? My disease status is a binary variable ( control as 0, disease as 1). I have run them both and the magnitude of the effect size is different, although for both it is a negative value and the association for both is significant.
lmer(Metabolites ~ Disease status + Fixed effects ( 7 in total) + random effects
glmer(Disease status ~ Metabolites + Fixed effects (7 in total) + random effects + family="binomial"