Rankings and scale

Most chapters of this book sort places from least to most deprived on the Area Deprivation Index and then compare an outcome along that ranking. Before the life course begins, this part asks what the ranking measures and what it says about the places inside a ranked unit. The answers decide how every later comparison should be read.

Chapter 1.1, What the deprivation index measures rebuilds the index from census measures of income, education, employment and housing and finds that it orders census tracts almost as median home value does. The index adds its measures in their own units, so the dollar amounts among them, home value, family income, gross rent and the monthly mortgage cost, set most of a tract's score. Versions that put every measure on a common scale first rank expensive places with high poverty as more deprived than the published index does. The chapter's report holds the full reconstruction.

That construction carries into later parts. An outcome that is itself a housing price or a housing cost, such as a home value or a renter's cost burden, is partly sorted on itself when it is compared along this ranking, and the chapters that make such a comparison say so. The book keeps the published index as its main ranking because it is the index that health researchers and agencies use, and because a replacement would have to be chosen before seeing how it changes the results. Each chapter that uses the index also shows its main gradient under a standardized index, median household income and the share of adults with a bachelor's degree. Under a rule fixed before those comparisons were run, the standardized index would have replaced the published index if it had changed enough of those gradients, and it did not (Chapter 9.1, How far one ranking goes).

Chapter 1.2, What a county average leaves out builds one index for tracts and for counties and asks what a county's percentile says about the tracts inside it. A county's rank follows the average of its tracts, yet many residents live in a tract far from their county's percentile, and a correlation between two measures changes when the unit changes from tracts to counties. Its methods, including the split of variation among states, counties and tracts, continue in Chapter 10.1, Gradients, slopes and variance.

The title of the book names the ZIP code, the unit by which most Americans locate a place. Almost none of the comparisons are made at that level. The sources report census tracts, counties, school districts, colleges or survey respondents, and the book keeps each comparison at the unit its source reports. Where a source is keyed to ZIP codes, the chapters use Census ZIP Code Tabulation Areas, which approximate postal ZIP codes and are not the same areas.