Control for race, and the two maps split apart

Turnout-weighted correlation of each tract measure with the 2020 Democratic margin, before (hollow) and after (filled) holding the tract's Black population share constant. Deprivation runs high = deprived; the mobility measures run high = more upward mobility.

Takeaway: net of Black share, deprivation's bar lengthens — the deprivation–Republican link gets stronger. The childhood-mobility bar crosses zero and flips: once race is held constant, higher-mobility tracts lean Democratic, not Republican.

Partial correlations are weighted-residual correlations net of pct_black_alone. Childhood mobility is the mean adult household-income rank of children raised at the 25th parental percentile (Opportunity Atlas, kfr); the individual-income variant is kir. Source: see notes.

Up the deprivation ladder vs. down the mobility ladder

Tracts in ten equal groups by deprivation rank (ochre) and, separately, by pooled childhood mobility (ink). Each point is the turnout-weighted mean 2020 two-party Democratic margin of that decile.

Takeaway: the deprivation line slides steadily Republican as places get poorer — until the most-deprived decile, the heavily Black urban core, snaps back to Democratic. The raw mobility line is dominated by that same low-mobility, Democratic-leaning core: a race artifact, not an opportunity gradient.

Deciles formed within tracts reporting each measure (deprivation n = 81,281; pooled childhood mobility n = 81,526). Margins above zero are Democratic, below zero Republican. Source: see notes.