Learn to build a world where everyone belongs. Take free classes at OBI University.   Start Now

The Othering and Belonging Institute is pleased to launch the Racial Disparities Dashboard. A racial disparity is defined as a persistent difference in outcomes between racial groups.1 This project is designed to provide data on these outcomes for American society, and to highlight areas of progress and regress in closing disparities in critical life outcomes. We currently only have data for Black and White Americans, but we hope to expand this project and include data for more races in future iterations. This dashboard allows users to track changes within the United States from 1970 to 2020 across 15 different important variables:


We find that some outcomes have improved greatly, both in absolute and relative terms, between racial groups, while other disparities remained stubbornly wide, or have even gotten worse. We can see, for instance, significant reductions in disparities in high school graduation rates, rates of health insurance coverage, and voting rates, even as disparities in homeownership rates, maternal mortality, and the racial wealth gap have gotten worse. A complete and nuanced portrait of race in American life requires an understanding and awareness of the range of these outcomes. As you click through each variable, you should be able to quickly see whether the disparity is improving or widening, as well as observe absolute progress (or regress) for each racial group.

Our goals are to provide clear and easily understood information on the differences in outcomes between racial groups, to show and juxtapose areas of progress and regress, and suggest a more nuanced and complex story about race in America.2 We hope this project is useful to researchers, journalists, students and concerned citizens. 

We caution that our dashboard does not provide an explanation for why some disparities have been reduced, grown larger, or persist. Research shows that, when presented with disparities, people tend to either explain it in terms of discrimination or cultural or intellectual differences between groups. We caution against drawing immediate conclusions or inferences about why these disparities exist. We believe that structural forces play a large role behind many of these disparities, but we deliberately avoid offering an explanation for why they exist or have changed. Our goals are to display and illustrate how they have changed, to provide easily accessible data on these disparities, and to deepen our understanding about what it means to say that a disparity has changed. Explanations for why these disparities exist or may have changed is beyond our immediate purpose with this project.

Progress Grading Report Card:


The Progress Grading Report is based upon four inputs: the change in the absolute and relative magnitude of the disparity, and the absolute and relative progress toward closing the disparity. The direction and strength of these inputs determines the final grade (as explained and shown in the legend and summer table further below). 

We decided to use a grading scale (A, B, C, ...) as the least confusing and most intuitive way to convey information about whether a key life outcome or disparity has improved or gotten worse over our selected time period. A decent grade does not denote that the situation is fine or acceptable. Rather, it is a grade of how America has performed from 1970 to 2020 on that particular variable, whether it has improved or gotten worse, not an evaluation of current performance. Thus, the C- grade for infant mortality should not be interpreted to mean that the current rates of Black infant mortality is a passing grade or even acceptable. Rather, it means that there has been a significant reduction in the overall absolute rate of Black infant deaths and the Black-white disparity since 1970. Alternatives to the letter grading scale were far less intuitive to reviewers prior to launching this project.

An important nuance which became more evident and dramatic in the development of this project is that determining whether a disparity widened or partially closed was not straightforward. As you may have noticed by clicking through some of the grades, disparity data is more complicated than it may superficially appear. In a few cases, such as median home values, the absolute value of the disparity grew or widened even though the disparity shrank in relative terms. But there is more to the story of progress than whether a disparity widened or closed. A disparity can grow wider even if the condition of the marginalized group improved. Conversely, a disparity can close or shrink because the condition of the non-marginalized group declined or got worse. Neither of those outcomes can be considered "progress" in the broadest sense. Therefore, we wanted our Progress Grading Report to account for both whether there was absolute progress for the racially marginalized group as well as the relative rates of progress. Progress has to account for the absolute performance or outcomes for groups, not just the disparities between them, and the relative rates of progress between them. Ultimately, we need to know whether a disparity is on pace to widen or close, not just whether it is getting larger or not. Our Progress Grading Report card takes account of all four inputs.

Disparities are measures of progress towards a more racially just society, a yardstick by which we can and should gauge progress. However, closing out the disparity should not necessarily be the primary policy goal (we do, however, link each grade card to our Structural Racism Remedies Repository, which offers policies that may be able to do this).3 Consistent with our Targeted Universalism framework, we recommend instead setting a universal goal for each issue area (where appropriate), and striving, through policy, to get all groups to that threshold. A byproduct should be to reduce disparities, but that is not the policy goal. This, again, is why our Progress Report Card grades are based on more factors than simply whether the disparity widened or not.

Below is a legend and a summary of the four factors we used to calculate the grades. All four factors are equally weighted. Click any item on the summary table to highlight that row.

  1.  Absolute Disparity: Percentage change in the absolute disparity between Black and White Americans from 1970 to 2020
  2.  Relative Disparity: Percentage change in the relative disparity between Black and White Americans from 1970 to 2020
  3.  Absolute Progress: The percentage change in the outcome or performance of that condition or indicator for Black Americans from 1970 to 2020
  4.  Relative Progress: The percentage difference in the rate of growth for the variable for Black and White Americans from 1970 to 2020

Summary Chart:




Closing Notes:

Thank you for your interest in this project. This is the first iteration of this project. Future versions may include additional timeframes (starting with 1990), other racial groups, gender/sex and possibly even age. There were other factors we would like to have included (such as maternal mortality), but for which we cannot get data for in 1970.  If you have a request or suggestion for items that should be included or considered for inclusion, please send us a note. Your feedback will be important to us as we consider further iterations of this project.

If you have any questions or suggestions, email us at belonging@berkeley.edu.


This project was a joint effort with contributions from the Equity Metrics team, Communications, and leadership under Stephen Menendian. This project was conceived and led by Stephen Menendian, and the methodology was designed and refined by Stephen and Samir Gambhir. The data was collected and organized by Samir Gambhir and Eve Liao, who provided invaluable suggestions from a data science perspective. The dashboard with visualizations was designed by Jake Tompkins with input from the rest of the project team.

  • 1 For purposes of this project, we use racial group definitions as supplied by the US Census. We recognize that racial categories are social constructions and under continuous evolution rather than immutable categorical distinctions. For purposes of this project, however, we follow conventions used by the Census.
  • 2 One of the complexities which we lack space to illuminate more broadly is the reality that disparities can shrink or grow wider even if the conditions or outcomes of either group decline or get worse. At the outset of this project, we theorized four possible outcomes in this regard: 1) The disparity improved (that is, it shrank), and this was driven, at least in part, by the improvement in the performance of the marginalized group. 2) A disparity improved (again, it shrank), but this was driven, at least in part, by worsening outcomes for the non-marginalized group, even as the performance of the marginalized group declined. 3) The disparity worsened (that is, it grew larger), even as the performance of the marginalized group nonetheless improved in an absolute sense. 4) The disparity worsened (again, grew larger), but the performance of the marginalized group worsened in an absolute sense. We originally planned to code each disparity according to this typology, but abandoned this typology after we realized that determining whether a disparity improved or worsened was not straightforward, because it could get worse in an absolute sense, but shrink in a relative sense (see Median Home Values). Thus, this typology broke down when looking at the details.
  • 3 It may seem paradoxical to feature an analysis on disparities, but then suggest that reducing disparities should not the primary policy goal, but there are many problems with a disparities focus. First, disparities can normalize the performance, outcomes or ownership of the non-marginalized group. In many cases, that means setting as the bar white outcomes. There are reasons we shouldn't do that, not just normative reasons. Second, and relatedly, the performance of the non-marginalized group can be declining, as we see with declines in life expectancy and maternal mortality among white Americans. If we are simply and narrowly focused on reducing a disparity, then that can theoretically happen even if the performance of the marginalized group does not improve. But we don't want a society where the goal is bringing down or level down the performance of another group, through retrogression. Rather, we want to raise all groups up. Third, and related to the second point, disparities can grow larger even if absolute progress has been enormous (see Bachelor's Degree attainment). We can and should celebrate progress, even if disparities persist. Fourth, and related to this last point, as this project shows, determining whether a disparity is growing or shrinking is not easy or straightforward. It can grow in an absolute sense even as it shrinks in a relative sense, and is on track to eventually close. This is paradoxical, but true. Fifth, the reduction in disparities does not necessarily entail a just society. Even if we eliminated disparities between groups, there could be enormous disparities within groups that are masked by a disparity focus. Sixth, a disparity lens is inherently a deficit model. We want to bring groups up, not focus on what groups lack. Relatedly, a deficit frame can often make it hard to build coalitions to solve problems because it creates a zero-sum mentality, even where the issue is not zero-sum.