Racial Justice Act Tool
This site provides summary county-level data representing the raw numbers, rates per population,
and disparity
gaps by race of adults in the California criminal justice
system using criminal records data provided by the California Department of Justice as well as population data
collected by the U.S. Census Bureau. This version of the tool
(v2) covers the period 1990-2023. It replaces the first version of the tool (v1), which covered 2010-2021 and
went offline July 31, 2026. See the “Updates and Changes”
section below for details about what changed between the versions.
(See Proving Actionable
Racial Disparity Under the California Racial Justice Act, 76
UC L. Journal 1 (2023); see also The Paper Prisons Racial Justice Act Data Tool, 29 Berkeley J. Crim. L. 29
(2024).)
This site may experience bugs or interruptions in service. Please send questions, comments, or requests to be on the updates list to rja@paperprisons.org.
How to Use the RJA Tool
Getting started using the Tool:
- Select a county (or California for the entire state)
- Select a year (10-year spans are an option and provide more complete data)
- Select an offense. You can enter a search term (such as "burglary") or a code number (such as 459)
- Select a measurement
About the Tool
This tool provides a way to explore racial disparities in support of the California Racial Justice Act (CRJA) (California Penal Code (CPC) section 745). Enacted in 2020 and amended in 2022, the CRJA provides a mechanism for defendants (and the convicted) in a particular county to challenge a charge, conviction, or sentence if it is sought or obtained in a racially disparate manner. Throughout the CRJA, racial disparities also encompass ethnicity and national origin.
The CRJA specifically addresses four types of conduct. The first two forms focus on the particulars of the case at hand: first, the exhibition of bias or animus towards the defendant by the state, a witness, or juror, and, second, the use of discriminatory language about or exhibition of bias or animus towards the defendant in court (unless quoting another person). The third and fourth forms of conduct, which concern charging and sentencing, require “evidence that the prosecution more frequently sought or obtained” harsher charging, conviction, or sentencing outcomes against people who are of the same race as the defendant. This tool focuses on the latter two forms of conduct.
To make a “pattern of disparity” claim under the CRJA requires two showings. First, in relevant part, CPC section 745(a)(3) states that, in the charging or conviction context, the defendant must be “charged or convicted of a more serious offense than defendants of other races, ethnicities, or national origins who have engaged in similar conduct and are similarly situated.” Second, CPC section 745(a)(3) further provides that “the evidence [must] establish[] that the prosecution more frequently sought or obtained convictions for more serious offenses against people who share the defendant’s race, ethnicity, or national origin in the county where the convictions were sought or obtained.” Similarly, in the context of sentencing, CPC section 745(a)(4) states that a defendant must first show that “a longer or more severe sentence was imposed on the defendant than was imposed on other similarly situated individuals convicted of the same offense,” then show either that “longer or more severe sentences were more frequently imposed for that offense on people that share the defendant’s race, ethnicity, or national origin than on defendants of other races, ethnicities, or national origins in the county where the sentence was imposed” or that “longer or more severe sentences were more frequently imposed for the same offense on defendants in cases with victims of one race, ethnicity, or national origin than in cases with victims of other races, ethnicities, or national origins, in the county where the sentence was imposed.”
CPC section 745(c)(1) provides that “[i]f a motion is filed in the trial court and the defendant makes a prima facie showing of a violation of subdivision (a), the trial court shall hold a hearing.” Furthermore, “evidence may be presented by either party [at the hearing], including, but not limited to, statistical evidence, aggregate data, expert testimony, and the sworn testimony of witnesses.” CPC section 745(d) states that data may be sought for good cause. According to CPC section 745(h)(1), “more frequently sought or obtained” or “more frequently imposed” means that "the totality of the evidence demonstrates a significant difference in seeking or obtaining convictions or in imposing sentences comparing individuals who have engaged in similar conduct and are similarly situated, and the prosecution cannot establish race-neutral reasons for the disparity. . . Statistical significance is a factor the court may consider, but is not necessary to establish a significant difference. ”
Disclaimer: This tool is intended to help you access and analyze criminal justice data and identify potential racial disparities for counties across California and the state as a whole. The tool and the accompanying data are provided as a public service 'as-is', and do not constitute legal advice or 'official proof' of actionable disparity or lack thereof. Please reach out to us with any questions or potential errors you spot by emailing rja@paperprisons.org. Your feedback will help us improve the tool. Thank you.
Podcast About the RJA Tool
About the Data
The source for data on this site is a comprehensive dataset of arrests, court actions, convictions, and sentences in California, the Criminal Offender Record Information (CORI) database, available to researchers through the California Department of Justice Automated Criminal History System (ACHS) under the provisions of the CA DOJ Research Data Request process. The records for this version of the tool were extracted from the ACHS system during the period August 21-28, 2024, and downloaded in early September 2024. The data we present are restricted to records dating to the period January 1, 1990 - December 31, 2023. We excluded records from 2024 both because our sample ends in August of that year, but also because of presumed lags in fully recording records data. We exclude records before 1990 because the completeness of the CORI electronic records data is questionable for earlier years.The first version of this tool (v1), available until July 31, 2026, provided data from a shorter time period, and was based on data provided by the CA DOJ that has since been refreshed, as described below in the “Updates and Changes” section below. V1 remains available in archival form, please see below.
While the CORI dataset provides valuable comprehensive statistics, it has some limitations. Among the known disadvantages of the CORI data are that it does not include systematic information on the conditions of the arrest (such as whether or not a weapon was present) or other aspects of the defendant’s conduct that might influence the evaluation of the “similar conduct” standard under the RJA statute. Two additional limitations of our database are that it does not include juvenile or out-of-state records. Errors in underlying data are due to reporting errors and/or fundamental limitations to the Automated Criminal History System set up and maintained by CA DOJ.
The CORI source data used for the tool are anonymous, with names removed and separate individuals identified only by an internal ID code. Personally identifying information, such as date of birth or social security number, has been removed. The tool reports summaries of the CORI data that have been processed to calculate counts (raw numbers) or rates for the specified user query. These summaries are never reported at the level of a particular individual. To protect confidentiality, our data sharing agreement with the CA DOJ requires that the tool only report metrics when the raw number of incidents reported is greater than 10.
Methodology
The "Paper Prisons Racial Justice Act Tool" allows visitors to customize the data in various ways. This methodology section presents important information about the data provided in this website and how you can use it for your own analysis.
Customization
You can customize the data presented by year, county, event point, race/ethnicity, measurement (metric), and offense. Data you see will depend upon your customization for each category. The event points include arrest, court action (of any kind), conviction, felony conviction, sentence to incarceration (jail or prison), and sentence to prison. The measurements calculate the rate at which an event occurs for selected racial-ethnic groups, relative to their representation in the population of the selected county. We explain the different options and measurements available and how they are calculated, as well as some limitations of the data, below.
Incidents & Lifecycle of a Case
The source (CORI) data set records each of the events associated with a given cycle of an individual’s involvement with the criminal justice system, where a cycle is defined as the series of events in the criminal justice system that flow from a specific initial incident for an individual. A cycle typically begins with an arrest, in which a person may be accused of one or more offenses, and then may proceed through a series of additional steps; for example, the arrest step is frequently followed by a prosecutor filing charges in court on one or more offenses. Offenses charged in court may be the same as or different from those recorded at arrest. The offenses charged in court lead to a disposition on each offense, such as a conviction, dismissal, acquittal, diversion, or other outcome. A conviction results in a sentence, which may include incarceration in jail or prison, a fine, probation, etc. The CORI data does not provide information on actual incarcerations or completed length of term served, only on sentences to incarceration.
The tool counts the number of incidents in which a particular offense has been charged. An incident is defined as a particular cycle for a particular individual. For example, if a person has at least one arrest for violating Penal Code § 242 (battery) in a particular cycle, one incident of arrest on PC 242 is added to the number of arrest incidents. Multiple counts of the same offense charged in the same arrest cycle are only counted once in the tool. For example, suppose that following a particular arrest incident, a person ended up convicted of three counts of Penal Code § 242 (battery) and two counts of Penal Code § 148(A)(1) (obstructing/resisting arrest). For this cycle, we would count one incident of conviction for Penal Code § 242 and one incident of conviction for Penal Code § 148(A)(1). We apply the same approach at the arrest and court levels.
If the same individual is arrested and charged with a certain offense on more than one occasion (in different arrest cycles), each cycle will be counted separately. For that reason, the number of incidents counted in the tool is greater than the number of individuals involved.
Event Points
Racial disparities can occur at each of a number of specific event points or steps in the criminal legal system. Criminal records are complex and present an array of event types that may be defined in various ways; in designing the tool we have striven to use simple definitions of key events based on unambiguous interpretation of variables in the CORI data source. The tool provides metrics at the following specific event points or decisions, as derived from the CORI data.
| Event Point | Definition and Source |
| Arrest | Step in the CORI data for which the CORI variable STP_ORI_TYPE_DESCR, which records the type or originating agency, takes the value “Arrest". This identification method includes initial arrests (ARREST/DETAINED/CITED) as well as supplemental arrests and custody records. |
| Court action | Any step for which the CORI variable STP_ORI_TYPE_DESCR takes the value "Court". |
| Conviction (including misdemeanors and felonies) | Court step for which the CORI disposition variable DISP_CODE is in the range of values 2500-2799, which encompass a variety of conviction categories. Misdemeanor charges are identified by the CORI variable OFFENSE_TOC taking the value “M”, and felony charges by OFFENSE_TOC equal to “F”. |
| Felony conviction | Conviction on a felony charge, using the definitions above. |
| Incarceration sentence (a sentence to prison or to county jail) | Conviction for which the CORI variable SENT_LOC_CODE takes the value "0", "A", or “J”, indicating a sentence to prison or jail. For both prison and incarceration sentences, we do not count suspended sentences, sentences to “fine or jail,” or sentences associated with non-conviction events, such as parole violations. |
| Prison sentence | Conviction for which the CORI variable SENT_LOC_CODE takes the value "0" or "A", indicating a sentence to prison. |
Year
For arrest event points, the year of the incident recorded in the tool is the minimum (first) calendar year for the cycle. For court events (all non-arrest events), the year is the maximum (last) year for the cycle. We count any event point that occurs in a cycle up to the last event. We assign years this way to take account of the fact that someone might be arrested in one year and go to court in a subsequent year.
The tool allows the user to select a single year or rolling 10-year spans. For example, selecting Year = “2009-2018” will report counts that add up all incidents during that range of years for the selected county, event point, race, and offense. Adding up incidents across multiple years allows us to report comparisons that would be excluded for single years because of the restriction to selections involving more than 10 cases.
Offenses
The CORI data set includes information on all the categories of criminal offenses that are chargeable as a misdemeanor or felony, most of which are in the penal code (PC), but also appear in a variety of additional California codes, including vehicular (VC) and health and safety (HS) codes. Offenses chargeable only as infractions, for example certain vehicular code violations, are excluded, as are probation violations (e.g. PC 1203.2).
Each code subsection is treated as a distinct offense (for example, PC 148(A) is treated as distinct from PC 148(B)). In situations where subsections might more appropriately be combined, users can select multiple code subsections in the tool, and access aggregate data outputs (as described more fully in the methodology section below, “Combining Data.”) At the same time, users should also be aware that our approach of identifying offenses with code subsections lumps together some charges that come under a single subsection but have different consequences. Examples would include so-called “wobbler” offenses that are felonies by default but may be charged as misdemeanors under some conditions. Burglary, which is PC 459, may be charged as first-degree or second-degree burglary, but the source data do not always distinguish the degree, so these charges are combined into a single offense code. In addition, offenses with different levels of detail are also lumped together, so that "459 PC-BURGLARY" and "459 PC-BURGLARY:FIRST DEGREE" charges are aggregated in our tool. Some offenses may only be associated with a limited set of event points.
For any given incident, the offense charged at court may be different from the offense cited at arrest, given prosecutorial decisions and plea bargaining. To ease searching, we include not just the code section number but the offense description (such as “PC 148(A) PC-OBSTRUCTS/RESISTS PUBLIC OFFICER/ETC”) most commonly attached to that offense code in the CORI data. We have excluded from the tool any records in the CORI data set for which a specific code section is not provided or cannot be identified.
County
The county reported in the tool is the county of the originating agency for the specific incident recorded, whether an arrest by a local law enforcement agency or an action by a county superior court. In a very small number of cases (0.3% of individuals and 0.08% of incidents), the county is recorded as “Unknown.” These cases are included in the California totals reported in the tool but do not appear in individual county data.
Racial classifications
We extracted race data as recorded in the CORI dataset as the basis for calculating racial differences in patterns of arrest, charging, conviction, and sentencing. The CORI data source indicates the racial/ethnic identity of each individual with a single mutually exclusive “race” identifier. Multiple racial identities are not recorded. While white, Hispanic, Black, and American Indian populations are specifically identified in the CORI database, we aggregated several races into the “AAPI” category (Asian Indian, Cambodian, Chinese, Filipino, Guamanian, Hawaiian, Japanese, Korean, Laotian, Other Asian, Pacific Islander, Samoan, and Vietnamese.), masking considerable heterogeneity in this population. Outside of the five racial groups that the tool collapses the various CORI categories into, persons of “other” and “unknown” race, representing about 3% of the total incidents, are not included in the tool. How, specifically, racial/ethnic groups are assigned in the CORI data (whether self-identified or assigned by authorities) is not indicated in the source. Although the CORI database also includes information on national origin (country of birth), this information is not reported in the current version of the tool.
To calculate rates per population, we use National Cancer Institute bridged race estimates of county population by year and race, prepared as part of the Surveillance, Epidemiology, and End Results (SEER) Program (data file ca.1990_2023.20ages.gz, downloaded 4/4/2025 from https://seer.cancer.gov/popdata/download.html). These estimates, produced under a collaborative arrangement between the U. S. Census Bureau and the National Center for Health Statistics, use models to calculate estimated population for consistent single-race categories, separately by Hispanic identity. In the tool, the Hispanic population includes people of any race, the White population is for non-Hispanic Whites only, and all other racial groups include Hispanic and non-Hispanic counts. These population categories are not mutually exclusive; for example, the Black population estimate includes some individuals who are also Hispanic and would thus be counted in the Hispanic population as well.
The following table summarizes the racial categories in the CORI and the corresponding category from the SEER population estimates:
| CORI race | SEER race and Hispanic identity |
| AAPI (combination of Asian and PI groups) | Asian American or Pacific Islander race (Hispanic or non-Hispanic) |
| Black | Black race (Hispanic or non-Hispanic) |
| Hispanic | Hispanic, any race |
| Native American | Native American race (Hispanic or non-Hispanic) |
| White | White race, not Hispanic |
| Other | NA |
NOTE: In downloadable tables, the population reported for 10-year spans is the average of 10 annual estimates.
Measurements
Three different metrics can be viewed in the tool:
-
Raw number is a count of the actual number of incidents in the selected category defined by county, offense, year, race, and event point. For example, a query of the tool for convictions in Alameda County in 2016 for Black persons on the offense PC 148(A)(1) returns a raw number of 55. This means there were 55 incidents involving convictions of Black persons on the charge of PC 148(A)(1) recorded for Alameda County in 2016. The tool counts incidents, not individuals, so 55 conviction incidents might represent fewer than 55 individuals, because a given individual might have been charged with the same offense on more than one occasion. For 10-year spans, the tool reports the total number of incidents over the 10 year period.
-
Annualized rate per population measures the rate at which a given event or decision occurs for a selected racial or ethnic group, relative to that group’s total population in the county. Specifically, it is the raw number of offense incidents of the requested type for the requested ethnic group during the requested year, per indicated number of individuals of that group in the county population, according to the following formula:
rate per population = raw number / county population (race-specific)
Annual estimates of population by race and county come from the National Cancer Institute’s SEER program, as discussed above.
Example: If there were 350 incidents in which Hispanic individuals had been arrested in county X on a charge of PC 459 (burglary) during the year in question, and the Hispanic population of county X was 100,000, the rate per 100 population would be 350/(100,000/100) = 0.35 per 100 population.
In cases where data from more than one year are combined, the rate per population is annualized. That is, the rate is adjusted to represent the average number of incidents per year by dividing by the number of years in the query.
For example: Suppose we wanted to run the above example combining data for 2012, 2013, and 2014, and suppose there were 350 incidents in which Hispanic individuals had been arrested in county X on a charge of PC 459 (Burglary) during 2012, 250 such incidents during 2013, and 300 such incidents in 2014. The annualized rate per 100 population for that time frame would then be [(350+250+300)/3]/(100,000/100) = 0.30 per 100 population per year.
-
Population disparity v. White compares the rate per population of a given racial/ethnic group with that of non-Hispanic white individuals according to the formula below:
population disparity v. white = rate per population (selected race) / rate per population (white)
The "population disparity v. white" can be considered the relative likelihood that a person of the given race/ethnicity experiences the outcome or decision (event point), compared to the likelihood of a non-Hispanic white person, given their respective underlying populations.
A population disparity v. white value greater than 1.0 indicates that a specific racial/ethnic group experiences a higher rate of a particular outcome or decision compared to non-Hispanic whites, considering their respective population sizes. Conversely, a value less than 1.0 suggests that the specified group is less likely to experience the outcome relative to non-Hispanic whites, given the underlying populations. A value of 1.00 means that the two groups experience the outcome at the same rate relative to population.
Example: Suppose that in a particular county in a particular year, there were 3 incidents in which Black adults experienced felony convictions for burglary for every 100 Black individuals in the population, and 2 such incidents per 100 non-Hispanic white individuals. Then the population disparity for the Black relative to white population is 3/2 = 1.5.
Combining Data
The tool permits data to be combined across years, counties, or offenses. When a user selects multiple values, the values displayed reflect aggregate counts in the case of raw count metrics (e.g. adding counts from 2018 and counts from 2019). In the case of rate per population or disparity gap per population metrics, the values displayed reflect weighted averages, taking into account combined event and combined population counts. Important caution: When an underlying data point is unavailable, due to a small number of cases (N<11), the tool will simply not include that data point in the calculation and will also display a message warning the user that not all selected values are reflected in the displayed values. As such, we encourage users to select the “View Data” to see which values are actually included in the aggregation, and to consider the metrics individually whenever also considering them in combination.
Example where rate data is aggregated: Suppose there were 350 incidents in which Hispanic individuals had been arrested in county X on a charge of PC 459 (burglary) during the year in question, and the Hispanic population of county X was 100,000, The rate per 100 population in county X would be 350/(100,000/100) = 0.35 per 100 population. Suppose that for another country, county Y, there were also 350 incidents in which Hispanic individuals had been arrested on a charge of PC 459 (burglary) during the year in question, but the Hispanic population of county Y was 350,000. The rate per 100 population in county Y would be 350/(350,000/100) = 0.10 per 100 population. To get the aggregate rate across counties X and Y, the numerators and denominators would be added, for a combined number of arrests of Hispanic individuals of 700 and a combined Hispanic population of 350,000 + 100,000 = 450,000. The combined rate per 100 population in county X and Y would be 700/(450,000/100) = 0.157 per 100 population.
Example where data is limited: Suppose that in a particular county in a particular year, say 2019, there were 12 incidents in which Native American adults experienced felony convictions for burglary, and in 2018, the number of equivalent incidents was N/A, due to the total being 10 or fewer. The raw count metrics for 2018 and 2019 in combination would still reflect 12 incidents, and a warning message would appear. In the case where the metric is the population disparity v. white gap, the chance that a person of the given race/ethnicity experiences a certain outcome or decision, relative to the chances of a non-Hispanic white adult, given underlying populations, and the data for either white or non-white populations is insufficient, neither will be included.
Updates and Changes
The original version of the RJA Tool (v1), which is provided as a courtesy for archival access here was restricted to data for incidents recorded in the source data during the years 2010-2021, and was based on CORI data downloaded between 9/23/2021 and 9/29/2021. The current version (v2) has been updated using a refreshed download of CORI records data for the calendar years 1990-2023. The following changes between the versions should be noted:
- Records coverage has been extended from 2010-2021 to 1990-2023.
- Actual records for some incidents and individuals have changed, even for the period of overlap between the versions. These changes may reflect corrections or updates in the ACHS database, including the removal of access to records that were expunged or sealed, for example under the terms of AB1076/SB731.Because of these changes, users should expect some potential discrepancies between the results of the same query across versions v1 and v2.
- The county population estimates used to calculate rate per population have been changed from static estimates using the American Community Survey 5-year estimates for 2016-2020 to the dynamic (year-specific) SEER population estimates discussed above.
- Measurements for 10-year spans of years are provided as an alternative to single-year estimates.
Archival access to v1 of the tool is available here .
Incompleteness of records for Los Angeles County
On June 1, 2026, the Paper Prisons team was notified by the California Department of Justice that a significant number of records from Los Angeles were missing from the Automated Criminal History System (ACHS), including from the download on which the current version of this tool is based. Specifically, their communication stated that “Los Angeles Superior Court has notified the DOJ that approximately 464,000 criminal record updates from their legacy court case management system were not previously transmitted to the DOJ for inclusion in ACHS. These records primarily span from 1965 to 2023 and include both felony and misdemeanor charge dispositions. The omission of these records may have an impact on state-wide or large regional analyses that relied on data contributed by the Los Angeles Superior Court, including counts of dispositions over time. The omission of these records may have a larger impact on any analyses conducted focusing only on Los Angeles County or Los Angeles Superior Court.”
The DOJ also informed us that they are working to add the missing data to the ACHS system. The RJA tool will be updated to reflect the updated data when it becomes fully available. In the meantime, users should be aware that the reported data for Los Angeles County are incomplete. How these missing records would affect racial counts and comparisons in the tool is unknown.
Racial Justice Act Tool FAQ
-
What is the purpose of the Racial Justice Act (RJA) Tool?
The RJA Tool was developed to help individuals, legal professionals, and researchers explore and analyze potential racial disparities (defined as imbalances between treatment of racial groups) within the California criminal justice system. It does this by providing access to comprehensive data on arrests, court actions, convictions, and sentences, allowing users to examine disparities at different stages of the legal process. This information can be used to support claims under the California Racial Justice Act (CRJA), which prohibits racial bias in convictions and sentencing.
-
What data does the RJA Tool use and where does it come from?
The tool uses anonymized data from the California Department of Justice's Criminal Offender Record Information (CORI) database, obtained through the Automated Criminal History System (ACHS). The current edition of the data covers arrests, court actions, convictions, and sentences in California from 1990 through 2023. The tool also uses annual population estimates by race from National Cancer Institute bridged race estimates of county population by year and race, prepared as part of the Surveillance, Epidemiology, and End Results (SEER) Program, produced under a collaborative arrangement between the U. S. Census Bureau and the National Center for Health Statistics (https://seer.cancer.gov/popdata/download.html).
-
What are the limitations of the data used in the RJA Tool?
- It lacks details about arrest circumstances (like weapon presence) or the defendant's conduct, which might be relevant to evaluating "similar conduct" under the RJA.
- It doesn't include juvenile or out-of-state records.
- It might contain reporting errors from the Automated Criminal History System.
- Data on specific offenses and in smaller counties might be limited due to privacy concerns, as data is not reported if the underlying count is 10 or fewer.
-
How can I use the RJA Tool to explore racial disparities?
The tool allows you to customize data by year, county, event point (arrest, conviction, etc.), race/ethnicity, measurement (raw number, rate per population, disparity v. White), and offense. You can analyze data for specific racial/ethnic groups and compare their rates of arrest, conviction, sentencing, etc., to those of White individuals. This comparison reveals potential disparities and can be used to support claims under the CRJA.
-
What are the different measurements available in the RJA Tool and what do they mean?
The RJA Tool offers three measurements:
- Raw Number: This is the count of incidents for the selected category (e.g., arrests for Black individuals for a specific offense in a particular county and year).
- Annualized Rate per Population: This measures how often a specific event happens for a racial/ethnic group compared to their population in the county. It's calculated by dividing the raw number of incidents by the group's population size, and annualized to represent the average number of incidents per year.
- Population Disparity v. White: This compares the rate per population of a racial/ethnic group to that of White individuals. It indicates the relative likelihood of a person from the selected group experiencing the event compared to a White person, taking population sizes into account.
-
What does a "population disparity v. White" value greater than 1 mean?
A value greater than 1 means the selected racial/ethnic group experiences the specific event (arrest, conviction, etc.) at a higher rate compared to White individuals, even when considering their respective population sizes. For example, a value of 2 would mean the selected group is twice as likely to experience that event compared to White individuals.
-
What does a "population disparity v. White" value less than 1 mean?
This means that the selected racial/ethnic group experiences the specific event (arrest, conviction, etc.) at a lower rate compared to White individuals.
-
When I click on a county, nothing shows up. Why?
Some counties are too small and contain 10 or less data points for some queries. In these instances, the RJA tool will state it is unable to display as there is insufficient data. As we process more data in the future, this may change.
-
How does the RJA Tool handle data when multiple years, counties, or offenses are selected?
When combining data across multiple years, counties, or offenses, the tool aggregates/summarizes the data. For raw counts, it adds the counts from each selection. For rate per population and disparity v. White, it calculates weighted averages considering the combined event counts and population sizes. If data is unavailable for any selection due to limitations, the tool excludes it and displays a warning message.
-
If an offense is not listed in the offense list within the RJA tool, can we assume it was excluded from the CORI data used for this analysis?
There are several reasons why an offense may not appear in the RJA tool's offense list:
- Not included in the data: The offense may not have been part of the original CORI dataset used to develop the tool. This could occur if it was not reported or not included in the data provided for analysis.
- Not a felony or misdemeanor: The tool includes only offenses classified as felonies or misdemeanors. Offenses categorized differently (e.g., infractions) would not be included.
- Low Frequency: Offenses with 10 or fewer incidents statewide over the analysis period are excluded from the tool in order to protect privacy.
-
Is there data available on racial disparities in charging decisions for wobbler offenses? (for example to make data visualizations like those that appear in your article,Proving Actionable Racial Disparity, at Figure 2, page 55)
For wobbler offenses, the RJA Tool can be used to explore data on charging decisions, including the proportion of cases charged as felonies versus misdemeanors for different racial or ethnic groups. However, the availability and robustness of such data may vary by offense and location. In many cases, small sample sizes make it challenging to draw meaningful conclusions about disparities in charging practices. If you are analyzing wobbler offenses, you can use the tool to examine specific offenses on a case-by-case basis.
For example, we used the tool to query for SF county data on PC 496(a) over the period 2014-2023. In these data, the ratio of felony to total convictions is 50/234 = 21% for whites, 108/371 = 29% for Blacks, 23/108 = 21% for Hispanics. We do not know if there will be many offenses-counties where you will have sufficient sample size to do this sort of analysis, but it is a piece of information you can look at on a case-by-case basis.
-
Sometimes the tool reports a value of N/A for one or more racial/ethnic group. Why does this happen, and does it invalidate the reported results for other groups?
Under our data agreement with California DOJ, we agree not to report results if the query yields 10 or fewer cases (incidents). So users of the tool should remember that “N/A” does not necessarily imply “zero cases,” but rather that the number of cases is 10 or fewer. This precaution ends up excluding many results, especially for less populous groups and less common offenses. Importantly, the deliberate suppression of data for racial/ethnic groups with 10 or fewer cases does not affect the statistical reliability or validity of the analysis for other groups that have sufficient case numbers. The tool's analyses, such as disparity ratios, are calculated independently for each group relative to the White population, using only the available, non-suppressed data for those specific groups. Therefore, conclusions drawn about racial disparities for groups with reported data remain robust and unaffected by the suppression of data for very small populations. The 'N/A' simply indicates insufficient data for that specific group to avoid re-identification; it does not impact the ratios developed for other groups. Finally, our tool is only as good as the data from which it is derived. So we are unable to improve upon any infirmities with the underlying data from the CalDOJ.
-
I ran a query for Year = “2010-2019” and obtained 55 arrest incidents, and then I ran the same query for each individual year, Year = “2010” through Year = “2019”, and the reported arrests only added up to 42. Why?
If the number of arrests for any individual year is less than 11, it will not be reported individually, but it is still added into the total for the 10-year span.
-
What should I do if I have further questions about the RJA Tool?
Our website has a lot of resources, including a video explainer that walks through a hypothetical, a podcast, and other goodies to come. But you can also contact the Paper Prisons RJA Tool team with questions, comments, or suggestions at rja@paperprisons.org. We will be happy to assist you and address any concerns.
Acknowledgements
The styling of this website was inspired by the California State of Disparities website, a data project of the Burns Institute, whom we thank.
This tool represents the collaboration of many dedicated volunteers and Paper Prisons team members including Bill Sundstrom, Yabo Du, Bennett Cyphers, Rayna Saron, Akhil Raj, Arthi Kundadka, Yangxier Sui, Lukas Pinkston, Navid Shaghaghi and Colleen Chien.
For more information about the RJA, please see:
- Proving Actionable Racial Disparity Under the California Racial Justice Act (exploring the disparities
standard of the RJA and how to apply it)
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4392014 - Paper Prisons’ Testimony at California Penal Committee Hearing on CA Racial Justice Act
https://paperprisons.org/news/2023/04/20/paper-prisons-testimony-at-california-penal-committee-hearing-on-ca-racial-justice-act/
