Log transform or log link? And confounding variables.

I check the robustness of last week’s analysis of height -> weight by trying a different method of specifying and fitting the model, and checking to see if socioeconomic status is acting as a confounder (because better-off people are both taller and …