The whole map

This part steps back from single outcomes and asks what the whole map looks like at once. It asks whether neighborhoods fall into a small number of types, whether the gradient has changed over time, whether neighborhoods with more immigrants follow it, and which places do better or worse than a model built on a few census measures predicts. The chapters work with tracts and counties, and a type or a residual describes a place and not the people who live there.

Chapter 9.1, Eight types of neighborhood sorts census tracts into types on measures that include life expectancy, adult income rank, diabetes and smoking, then maps the types by county and compares them on outcomes. Because outcomes are among the inputs, part of the spread in outcomes between types is built in. The chapter checks how much of it remains when the outcomes are left out of the sort. The types are regions of a continuum and are not sharply separated, and the chapter measures how well tracts fit their own type against simulated data that contain no clusters. The names are this book's labels for groups of tracts that look alike on the inputs, and they do not describe the people who live there. A second sort, built from deprivation measures with outcomes left out, is in Chapter 10.4, Families of places.

Chapter 9.2, Change in the gradient over time (planned) will ask whether the gap in outcomes between more and less deprived neighborhoods widened or narrowed between 2010 and 2023. Chapter 9.3, Immigrants and the gradient (planned) will ask whether neighborhoods with more immigrants follow the same gradient between deprivation and health, and where they depart from it.

Chapter 9.4, Places that beat the model fits a model of upward mobility from a handful of census measures, namely poverty, education, single-parent households, median income and the Black share of residents, and averages its residuals to counties. Counties below the prediction cluster in the South, and the largest counties above it lie around New York. The chapter describes what marks the counties above the prediction, such as older housing and fewer smokers. It then follows tax filers between counties for one pair of filing years and asks whether net migration runs toward the places where children from low-income families rise furthest. Counties where children from low-income families rose further gained fewer filers on net. A residual is the part of mobility that the measures leave unexplained. It is not a measure of what a place does for a child, and the chapter lists what it mixes together, among them noise in the estimates, omitted measures and the gap between the decades of the outcome and of the predictors.