City+ ACS expansion batch (equity and children indicators)

Seven indicators computed first-hand from published ACS 5-year detail tables, 2010-2024 (2012+ where a table starts later), at nation, state, county, place, and census-tract levels. Built because the aggregators that publish these for cities (KIDS COUNT, CityHealth Dashboard) carry restrictive licenses, while every underlying table is public. | variable | construction | |---|---| | `dw/GiniIndex_HouseholdIncome` | B19083 direct value (0-1) | | `dw/Percent_Household_NoVehicle` | B08201: no-vehicle / all households | | `dw/Percent_Household_CostBurdened` | B25070 30%+ renter cells + B25091 30%+ owner cells, over households whose burden could be computed (the "not computed" cells leave the denominator) | | `dw/Percent_Child_SingleParentFamily` | B09002: own children in "other families" (unmarried householder) / all own children | | `dw/Percent_Youth16To19_NotInSchoolNotWorking` | B14005: not enrolled x (unemployed + not in labor force), both sexes, both education branches, / all 16-19 | | `dw/Percent_Child_NoHealthInsurance` | B27001: under-6 + 6-18 uninsured / those age universes (under 19; table starts with the 2012 vintage) | | `dw/Percent_ChildUnder6_AllParentsInLaborForce` | B23008: both-parents-in-LF + single-parent-in-LF / all under 6 living with parents | Every share ships with a `dw/MarginOfError_` twin in percentage points, derived with the Census handbook formulas (MOE of a sum = root sum of squares; MOE of a proportion via the subtraction form, falling back to the ratio form when the radicand goes negative). Negative jam values are skipped; a controlled estimate (MOE sentinel -555555555) counts as zero error rather than blocking the derived MOE. The batch also carries eleven MOE-ONLY twins (`dw/MarginOfError_CityPlus_*`) for the City+ metrics whose VALUES come from base Data Commons (population, median household income, poverty, child poverty, homeownership, broadband, uninsured, SNAP, preschool enrollment, bachelor's-or-higher, children share): base DC does not republish ACS margins, so the twins are computed from the same source-table cells and power the City+ significance tiers (change vs the combined published margin at 90% confidence). All margin variables are statType `marginOfError` support metadata: the AI chat's resolvers exclude them from answers unless the question is about the uncertainty itself.