JOE is a product of the CJARS (Criminal Justice Administrative Records System) team, based at the University of Michigan, in collaboration with the U.S. Census Bureau. Any opinions and conclusions expressed are those of the authors and do not necessarily reflect the views of the U.S. Census Bureau.
Criminal justice information displayed in JOE is a product of the CJARS (Criminal Justice Administrative Records System) Database. For more information on becoming a participating data partner with CJARS, please visit https://cjars.org/get-involved/ or contact cjars-data-collection@umich.edu.
JOE provides a unique view of the criminal justice system, combining the unprecedented geographic and temporal scope of CJARS with detailed information on health and economic outcomes for justice-involved individuals from the U.S. Census Bureau.
In the JOE application, hover over the information bubble adjacent to each statistic name for definitions.
All Economic and Health outcomes are linked to CJARS data through data available at the U.S. Census Bureau, including tax filings and participation in federal assistance programs.
Each JOE statistic is defined for a particular “cohort,” or the set of people with a particular form of criminal justice system contact in a given calendar year. The follow-up view tracks outcomes for a single cohort over a five-year period. The cohort view shows outcomes for different cohorts over time, measured at a fixed interval.
Per capita rates are calculated as the number of criminal justice events per 100,000 per selected geographic area. When you filter the data by any of the demographic categories available, both the numerator and denominator change to reflect the newly selected population.
The context metrics provide static state- and county-level information on features like health, unemployment, poverty, and crime rates. They are based on external, publicly available data and are intended to provide context for the statistics available in JOE.
Each context metric is represented as a single value for each geographical region, calculated as an average over the years covered in JOE.
The line shows the best-fit trend between statistics on the y-axis and the context metric on the x-axis.
For per capita rate statistics, the size of the bubbles scale with the population of the given geography. For all other statistics, the size of the bubbles scale with the caseload of the given geography.
You can go to our Methodology page for more information on the methods used to produce the statistics in JOE.
We use a custom aggregation strategy to make sure that we are able to produce the most comprehensive set of statistics as possible without compromising the privacy and confidentiality of justice-involved populations. Statistics for counties with small populations often show erratic trends when viewed individually, so we combine counties in order to give a better representation of the actual trends in the criminal justice system in a given area, while statistics for larger counties are reported separately. See the Methodology page for more information.
The five race and ethnicity categories were chosen for a combination of compatibility with existing categorization schemes used at the U.S. Census Bureau and criminal justice agencies and for simplicity in implementing a national-level database. Race and ethnicity data for population statistics and recidivism outcomes are harmonized from administrative data provided by criminal justice agencies along with an imputation strategy based on geography as well as first and last names (see the CJARS data documentation). Race and ethnicity data for economic and health outcomes are derived from U.S. Census Bureau sources.
Incarceration statistics only include individuals serving time in prison facilities. Jail terms are not included.
For prison inmates, geographical location refers to the county or state of criminal charge.
Recidivism statistics reflect only the population and follow-up event specified. For example, with the “Prison inmates” population selected, the “New felony charge” statistic only measures members of the inmate population with a new felony charge, and does not include other potential forms of recidivism (re-arrest, new misdemeanor charge, re-incarceration, etc.).
Taking history of justice involvement into account helps to provide a more accurate view of the justice system by separating individuals with frequent justice contact - who often appear in the same statistics year after year - from those encountering the system for the first time, who are generally less likely to experience recidivism. Prior justice involvement is defined as any criminal justice event of the same type prior to the selected year. In cases where at least five years of historical data in the relevant domain are not available, statistics based on history of justice involvement are not available.
Raw offense descriptions, as provided by criminal justice agencies, are categorized using the CJARS Text-based Offense Classification algorithm, described here.
For health and economic outcomes, statistics for some demographic groups are measured in two-year intervals in order to protect the privacy of individuals. See the Methodology page for more information.
Yes, in the bottom right corner there is a drop-down menu that allows you to select which data inform the color gradient. The default option scales the colors based on all of the data for the selected statistic and year, which is best for comparing across different demographic subgroups or other filter settings. The “Filtered data” option scales the colors based only on the selected subgroup, which can help bring out more subtle differences in the map but still allows for comparisons across all geographies. The “Visible on map” option scales the colors based on the maximum and minimum values on the screen at the current zoom level, which makes it even easier to see nuances in variation within a smaller geographical area.
All of the statistics available on the JOE dashboard are also available for download as CSV tables on the Get the Data page.
CJARS is always expanding, and collecting decentralized data can take time and requires the cooperation of state and local agencies. CJARS ultimately plans to cover every part of the country.
CJARS collects data from a variety of public and private sources, and temporal coverage varies greatly across sources. Health and economic outcomes are derived from U.S. Census Bureau data, which also varies in temporal availability.
The last year available to view is 2020. It may take 1-2 years before data on health and economic outcomes from the U.S. Census Bureau becomes available for use, which creates a limit to how recent data can be presented. However, available data can vary across jurisdictions and data sources depending upon data provided to or collected by CJARS.
Some CJARS data is sourced from county-level agencies, like county courts. In some cases, state-level data sources may also have gaps at the county level due to inconsistencies in reporting from some counties.
Some of the agencies that CJARS receives data from do not provide information on the county of criminal charge associated with criminal justice events, in which case we cannot produce statistics at the county level.
Both state- and county-level statistics (where available) are derived from the same data sources, so a county that is missing a statistic for a given year will also not contribute to the respective state-level statistic.
Currently, statistics are only available for state and local jurisdictions.
Use the HOW TO button in the bottom left of the application to access guidance for common tasks.
This website supports the most recent 2 major versions of Grade A browsers (these include Firefox, Edge, Chrome, and Safari) across desktop, tablet, and mobile devices.
We would love for you to share your discoveries and insights based on JOE. Instead of taking a screenshot, use the EXPORT button in the bottom left of the application to save a formatted image of the current JOE view, with conveniently formatted legend and attribution.
For help with this site or questions about how to interpret the statistics, contact us at cjars-joe@umich.edu.