MFJ begins by collecting data from U.S. statewide court datasets. Often, these populate the majority of our Measures. If feasible, we also approach local agencies (prosecutors’ offices, sheriffs’ offices, public defenders’ offices, etc.) to acquire supplemental data in an effort to ensure we capture the entire criminal justice system. We do this in writing, but also, if possible, in person, via pairs of MFJ researchers who travel our states county by county.
MFJ understands that every county is different. Our process is iterative and responsive to stakeholder feedback. We continually seek input, local context, and information at multiple stages of our work to ensure our data are reliable and take into account county and state differences.
MFJ also recognizes that criminal justice data are not collected, recorded, or systematized uniformly across local jurisdictions. Thus, addressing data quality and uniformity has been a critical MFJ goal from the outset. We have developed a process to clean and code our data in-house and then run them through three audits—two internal and one external—to make sure they are correct. This process involves:
Per these reviews, MFJ does not use any data elements identified as problematic.
Our complete methodology is available for download.
All our case-level data are aggregated to the county level so that the Measures can be calculated and centralized in an online repository—a public Data Portal—that is free and available to anyone. The Portal has been designed so that it’s easy to use but also comprehensive. We provide multiple views to satisfy users of all skill and interest levels (and no personal information is published for any parties involved in a case).
On the Portal, the Measures can be visualized as maps, bar graphs, and tables presenting county-level data. These views are available for download. Users can also compare counties within and across states.
Every Measure will be accompanied by:
When viewing the measures users will be asked to keep the following in mind:
Our Measures are meant to be a starting point for a conversation about the criminal justice system that addresses what’s working well and what needs further attention. The aim is to create transparency.
Our system measures only the performance of counties on the processing of adult criminal cases. Therefore, we do not measure how juvenile, family, civil, and other cases may fare. Nonetheless, our Measures can be filtered by the age group of the defendant, including those under 18 (juvenile defendants who were waived to adult court).
Our Measures can be filtered by defendant characteristics (race/ethnicity, indigent status, sex, and age) and by case characteristics (offense type, offense severity, court type, attorney type, and drug type–only for drug-specific Measures). We encourage users to explore the Measures using these filters. Some filters calculate disparities between two groups. However, we don't test the statistical significance of such disparities.
Measures for Justice (MFJ) works with data extracted from administrative case management systems. These data were originally collected by the sources for the purpose of tracking the processing of individual cases and not necessarily for the purpose of measurement. Nevertheless, they are suitable for measurement provided they are handled correctly. Often, these data are reliable. Just as often, they can be entered incorrectly or not at all, may be subject to errors at any stage of the recording and collection process, and may not be standardized across counties. MFJ has taken steps to account and adjust for these problems but cannot correct entirely for errors in data entry. For these reasons, and because jurisdictions use a variety of calculation methods, we encourage examining overall patterns instead of exact percentages when comparing to reports produced by local agencies.
Criminal justice agencies use different methods to record cases. Some jurisdictions file all charges against a defendant under the same docket number and sometimes they do so even when the charges stemmed from different incidents. Others file each charge under separate docket numbers even when the charges are for the same incident. To standardize the definition of case across jurisdictions, we count all charges associated with the same defendant that were filed (or referred for prosecution, in the case of declinations) on the same date as a single case. We assume that when a prosecutor files multiple charges together, even when they originated from different incidents, they intend to resolve these charges simultaneously. Since the focus of our Measures is case processing, not case clearance, we believe this approach is currently the best way to standardize case definition across jurisdictions.
Because cases often involve multiple charges of differing severities, we define cases based on the most serious charge, according to the state's offense severity classification, that was present at each stage of the case processing, respectively referral, filing, and conviction.
MFJ’s research is descriptive and does not, by definition, tell us why things happen. As such, we do not test hypotheses about the reasons for the patterns the data reveal. When our Measures show differences between states, counties, or groups (e.g., in medians, percentages, or rates), we make no claim about the reasons for these differences.
MFJ uses a Relative Rate Index (RRI) to assess disparities on case processing outcomes between white defendants and defendants of color, males and females, and indigent and non-indigent defendants. The RRI compares how two groups fare on the same outcome by dividing the results of one group by those of the other. An RRI equal to 1 indicates that there is no disparity in outcomes between the two groups. Disparities are not calculated when there are fewer than four cases in the denominator of the rate for either group. We also test the statistical and substantive significance of disparities. Disparities that are neither statistically nor substantively significant are suppressed from publication.
MFJ estimates confidence intervals to test whether the disparity in outcomes for the two groups is beyond what could be expected by random chance. In this sense, statistical significance provides information about the precision and certainty of the measurement. When a disparity is statistically significant, we can be 95% confident that the rates for the two groups are unequal. Statistically significant disparities are noted with an asterisk (*).
Because statistical significance is affected by sample size, MFJ also evaluates whether the size of the disparity merits attention irrespective of statistical significance. When a disparity is substantively significant, this means it is large enough to warrant attention. Disparities equal to or greater than 1.05 are considered substantively significant, and attempts should be made to understand and address them.
Each Measure sheds light on a corner of a local criminal justice system, but to evaluate the health of that system in a more comprehensive way, all available Measures should be assessed together and interpreted with county context in mind.
We measure criminal justice performance at the county level because it is usually at this level that charging, disposition, and sentencing decisions are made.
The maximum permissible percentage of cases with missing values for any given measure is 10 percent. Performance measures for counties with more than 10 percent of cases missing values in the numerator or in the pool to calculate the median are suppressed from publication. In addition, performance measures for counties with more than 5 percent and up to 10 percent of cases with missing values display a “high missing rate” warning.
MFJ uses statistical simulations to estimate the amount of bias that may result from missing data. The bias depends both on the percentage of missing data and the actual value of the measure being estimated. For example, in a county where the pretrial diversion rate is low (e.g., 3%) and there is a considerable proportion of cases missing data (e.g., 7%), the estimate of the pretrial diversion rate could be inaccurate. Bias is estimated as a function of the sample mean and the percentage of missing data. Whenever the sample mean and the percentage of missing data suggest a level of bias greater than 5 percent, MFJ suppresses the data from publication.
MFJ continues to seek out more data—especially law enforcement data—as part of our effort to measure all corners of the criminal justice system.
If you’ve given us data and don’t see them represented in the Portal yet, it’s because we are still working on them to ensure accuracy. Thank you for your participation and patience.
We provide a complete history of portal updates that allows you to track when data changes or new data have been released to the portal or when new versions of the portal are made available.
Our complete methodology for creating the Measures in PDF format.