Validate child-level eligibility before labeling population counts as qualifying children

The population generator maps a family's positive child_care_subsidies amount to every person, then counts every child under 13 with positive imputed care hours. It does not check each child's own eligibility or payment.

For a family containing an eligible child and an ineligible sibling, the logic can count both as “Children who would qualify.” The explanatory paragraph more precisely describes children whose family receives a subsidy, but the headline/table labels imply individual qualification.

Determine the intended estimand. If this is individually eligible children, count child-level eligibility/payment under each state's rules and add a mixed-eligibility sibling regression. If it is children living in recipient families, rename the metric consistently. Quantify the national and state impact before publishing revised totals; the magnitude of any overcount has not yet been measured.

Current headline: 16,853,856 children. This issue identifies a counting/labeling concern, not a verified corrected national total.

Source: scripts/microsim.py.

Start at scripts/microsim.py around line 159 and trace how child_care_subsidies and imputed care hours become the population count. Resolve whether the estimand is individually eligible children or children in recipient families, then inspect the relevant state eligibility rules. Done means the metric and labels agree, mixed-eligibility siblings are covered by regression, and national and state impacts are quantified before totals are published.