The above data files supply the variables that populate the Upward Mobility Data Dashboard. The data files each contain a number prefix to offer easy ways to differentiate the many files. Each file is organized to group similar types of information together for as many mobility metrics as it relates to. 

Overall values for counties across multiple years have their own files, and we have a separate file for cities, which we label as “places” to stand for census places. For metrics that we disaggregate by certain categories, we have a separate file to capture those different categories, including race/ethnicity, income, disability, age, gender, housing tenure, industry, mother’s education, and the racial characteristics of neighborhoods. All of these, if applicable, are in two separate files for those in counties vs those in cities, where available. Each of these files will have geographic indicators and year indicators, and the geographic indicators will have both a numeric code and character variables that spell out the name of the state and county or city so that users can more easily filter the data and see all the communities available.  

The table below provides a brief summary of each file as well as a list of the subgroups that can be found in them. For more detailed information around the variables within these files visit the data dictionary.

Each metric will have at least one main variable that is the metric value.  Some metrics require more than one main variable, for example the mobility metric “household income at the 20th, 50th, and 80th percentiles” has three main variables: pctl_income_20, pctl_income_50, and pctl_income_80. Each mani variable will have a group of accompanying variables that will include its data quality index (“_quality”) and, if confidence intervals are calculated, a pair of variables that indicate the lower-bound (“_lb”) and upper-bound (“_ub”) values all of which will have a prefix that matches the main variable name.  

In addition to the mobility metric files, this catalog includes the file county_medians_year_suppressed.csv which contains approximate annual national medians for metrics. To approximate the median, we calculate the median county-level metric weighted by county population. We suppress medians when the proportion of the population in counties with data quality equal to 3 or missing mobility metrics exceeds 15% of the national population. This median represents the county-level metric for the median person in the country. The names of the median metrics in county_medians_year_suppressed.csv match the names of metrics in the other data files. Missing values and suppressed values are both represented by `NA`.

The “Revisions” tab (next to the default “View” tab) is where we store an archive of past years of data files. We maintain these for posterity, but the current data files will contain all of the years of data and variables tied to each metric. 

To see the programming code that generate each of these variables, visit our public GitHub site: https://github.com/UI-Research/mobility-from-poverty. See the data dashboard Appendix to learn how each metric performed against our selection criteria, alternative metrics considered, and a summary of the data available in the Dashboard for each metric (i.e., years, categories of disaggregation, and data sources).