# And what did we learn today, kids?

When you estimate a logit over ratios with significantly independent numerator and denominator with significant heterogeneity, your model can get really mangled.

imported from: yetanothergeographer