Let's assume we would like to examine regional disparities in income. We are NOT interested in country-wide effects. A DAG tells us to adjust for age and education. DAGs do not tell anything about additive or hierarchical model specification. Should we specify region additively or hierarchically?
income ~ age + education + region
income ~ age + education + (1 | region)
Isn't it so that regional-level effects would be very similar with both models? What would be your considerations in this case?