Correlation between outcomes and identity groups is particularly not bias if the two are conditionally independent given other variables (e.g., income level).
-
Understanding based on identity is easy since data is easily available Solutions should look at other variables which have a stronger correlation & also offer direct paths for corrective action LowIncomeFam->NoAfterSchoolCare->LessTimeOnHW->PoorAcademicPrf is easier 2 fix \1
If we're committed to doing interventions no matter what it's true that we should focus on those areas where interventions have some efficacy even if that efficacy is low, that's true. I am not sure if Pedro would agree that we should have such an unconditional commitment.
