Why the federal government needs to change how it collects data on Native Americans

Waving flags of Navajo American Indians. Flag of the Navajo Nation in US. 3D rendering

This January, the White House released proposals for reclassifying racial data collection in the 2030 census. Most notably, the proposals include combining race and ethnicity into a single question, as well as creating a new racial category for Middle Eastern and North African people.

While these changes are welcome and will go a long way toward helping create a more accurate picture of demographic identity in the U.S., they don’t improve data representation for a group that has long been misrepresented: American Indians and Alaska Natives (here collectively referred to as “Native Americans”). In particular, the way the U.S. government currently collects, aggregates, and publishes race and ethnicity data can lead to the exclusion of more than three-quarters of Native Americans from some official data sets. These practices may bias research, contribute to negative policy impacts, and perpetuate long-standing misunderstandings about Native American populations.

The concept of “race” can be a problematic descriptor for Native American identity, which is a political and legal identity in addition to a racial one. By rethinking how the federal government collects and publishes data on Native Americans, we can begin to better assess the challenges they face and ensure the population is accurately represented when that data is used in research and policy.

The complexities of US race data collection for Native Americans

In 1977, the White House Office of Management and Budget (OMB) issued Directive No. 15, “Race and Ethnic Standards for Federal Statistics and Administrative Reporting,” which guides government collection and aggregation of race and ethnicity data. In 1997, Directive No. 15 received a major update—the most recent until last year, when the OMB announced it was again reviewing the directive in preparation for the 2030 census and beyond.

Today, under guidance from Directive No. 15, race data is typically aggregated into five major racial categories: white, Black, Asian, American Indian or Alaska Native, and Native Hawaiian or Other Pacific Islander. The additional major category is ethnicity: Hispanic or Latino, which is currently not considered a race (an at times confusing distinction that contributed to the most recent proposed changes).

This system has historically worked for most Americans, because the vast majority of the three largest racial groups in the United States—white, Black, and Asian—identify as one race alone. However, Native Americans identify as two or more races at significantly higher rates than those larger groups.

Racial classification on the 2020 Decennial Census

Racial data captured in the 2020 census shows that 87% of white Americans, 88% of Black Americans, and 83% of Asian Americans are classified as one race alone. In contrast, just 39% of American Indians and Alaska Natives are classified as one race alone. This is a legacy of the complex effects that hundreds of years of colonization have had on the identities of Native Americans, as well as the modern-day technical processes that the Census Bureau uses to code individual survey responses. The same factors help explain why just 43% of Native Hawaiians or Other Pacific Islanders, who are also Indigenous, are classified as one race alone.

This is problematic because government agencies and non-governmental researchers often choose to aggregate all multiracial individuals into a single “two-or-more-races” category. When that happens, it removes a majority of Native Americans and lumps them into a catch-all category with groups that have significantly different backgrounds and life experiences. While government agencies and non-governmental researchers are allowed to provide more detailed race data than just the minimum categories outlined by Directive No. 15, they frequently don’t. Many U.S. government data sets and publications use only single-race categories, including the Bureau of Labor Statistics’ (BLS) monthly unemployment report, the Department of Education’s higher education data, and the Centers for Disease Control and Prevention’s (CDC) life expectancy data.

The problem is further compounded by Latino or Hispanic identity, which the federal government currently considers an ethnicity, not a race. In recent years, government agencies and non-governmental researchers have moved to treat Latino or Hispanic identity as a de facto racial category (prompting, in part, the move to officially combine Latino or Hispanic identity and racial identity into the same question). Today, government agencies and non-governmental researchers frequently break out Latino or Hispanic people into their own, mutually exclusive category, and show only non-Latino or Hispanic individuals in other racial groups (e.g., non-Hispanic white, non-Hispanic Black, etc.). While this process is important for ensuring Latino or Hispanic representation in data, it further erodes the accuracy of data around Native people, because single-race Native Americans are classified as Latino or Hispanic at a higher rate than single-race individuals in the other four major racial categories.

American Indian and Alaska Native population by race and ethnicity

The implications of removing Latino and Hispanic individuals from Native American data are huge. On the 2020 census, just 23% of all Native Americans were classified as single race and non-Latino or Hispanic. In other words, one of the most commonly used racial categorizations of Native Americans— single-race, Non-Hispanic American Indian or Alaska Native—excludes more than three-quarters of the total Native American population. In contrast, nearly 3 million American Indian or Alaska Native people (either alone or in combination with another race), accounting for 31% of the total American Indian and Alaska Native population, were classified as having Latino or Hispanic ethnicity.

Indeed, the OMB defines American Indian or Alaska Native as “a person having origins in any of the original peoples of North and South America (including Central America) and who maintains tribal affiliation or community attachment.” So, according to the federal government’s own definition, Indigenous people from Central and South America should be considered Native American. This definition further underscores the challenges of balancing Latino or Hispanic and Native American identity.

Additionally, Native Americans are the only census-defined “racial” group that is also a political and legal classification. While modern-day Latin American countries were colonized in ways similar to the United States, Indigenous Peoples from those countries often lack federal recognition in the U.S., and they don’t typically maintain a “nation-to-nation” relationship with the U.S. government. Conversely, many Latino or Hispanic Native Americans born in the United States may be enrolled tribal citizens, but may not be counted as “Native American” by many government data sets and researchers.

Given that Latino or Hispanic individuals represent nearly one-third of the Native American population, improving our understanding of how mixed-race individuals who are Latino or Hispanic and Native American are faring is vital. However, the most commonly used racial categorizations aren’t well suited to account for these complexities. And while individual researchers can choose to detail these complexities in their work, the way federal government agencies aggregate and publish race and ethnicity data often doesn’t facilitate them doing so.

Share of U.S. population classified as two-or-more races by Decennial Census

While these challenges affect Native Americans most acutely, they may soon become a challenge for other groups as well. Between 2010 and 2020, the number of Americans classified as two or more races more than tripled, driven in large part by a more than 560% increase in the number of Latino or Hispanic individuals coded as two or more races under the Census Bureau’s revised coding process. As a result, today, more than one in 10 Americans are classified as multiracial—meaning over 10% of Americans may be getting kicked into the catch-all “two-or-more-races” bucket. Likewise, these trends may continue to accelerate if Latino or Hispanic identity is treated as a racial category in the 2030 census. In this regard, the current practice of aggregating data into mutually exclusive, single-race categories may become less and less descriptive of the realities of the U.S. population in the coming years.

Native American misrepresentation in data can have serious implications in research and politics

Two recent publications illustrate how researchers’ handling of Native American identity can affect research outcomes—either by suppressing Native American inequities or highlighting actionable realities.

The first example is a study published in January by the academic journal The Lancet, which showed that Native Americans are the group most likely to be affected by “deaths of despair” such as suicides, drug overdoses, and deaths from alcoholic liver disease. These findings cut directly against a 2015 study by Princeton researchers showing that deaths of despair were highest for non-Hispanic white Americans. Both studies appear to rely on standard CDC racial classifications, which are based on OMB’s single-race categories and use a complex process called “bridging” to sort multi-racial people into individual racial categories. But the 2015 paper did not include Native Americans in the data at all, effectively erasing them—a common occurrence when data is aggregated in a way that excludes many Native Americans and treats them as a small enough portion of the population to be ignored entirely.

A second study shows how researchers can effectively demonstrate the nuances of Native American identity to produce interesting and actionable findings. Jeffrey Burnette at Rochester Institute of Technology and Matthew Gregg at the Minneapolis Fed’s Center for Indian Country Development looked at the effects of COVID-19 on Native American life expectancy. As the authors note, a previous CDC publication found that Native American life expectancy dropped by nearly seven years from 2019 to 2021—a devastating effect on individuals and communities. However, that CDC study only looked at individuals who identified as single-race, non-Latino or Hispanic American Indian or Alaska Native. Burnette and Gregg find that when individuals who are classified as American Indian or Alaska Native alone or in combination with another race are included, the drop in life expectancy was closer to four years, and overall life expectancy for Native Americans of all racial backgrounds is nearly eight years longer than single-race, non-Latino or Hispanic Native Americans. As the authors write:

“[T]he size of the decline in life expectancy in the overall American Indian/Alaska Native population seems to be driven by deterioration in health of the non-Hispanic single-race population…It is possible that individuals who identify as single-race Native American are also more likely to share other characteristics that differ from the larger multi-racial Native American population and put them at risk for a disproportionate pandemic impact.”

By understanding the nuances of Native American racial data, the authors were able to give a clearer picture of the realities of Native American well-being in the U.S. and discuss specific challenges to well-being that some groups of Native Americans face while others don’t.

Aggregating Native Americans into a monolithic racial group can have problematic implications beyond just research. Not only are Native Americans racially diverse, but, as mentioned above, they are also the only census-defined racial identity that is simultaneously a political identity. Today, there are 574 federally recognized Native American tribes in the United States, as well as hundreds of additional state-recognized and unrecognized tribes. Each represents the continued existence of Native nations that have inhabited the North American continent for tens of thousands of years and that often maintain treaties and political agreements with the U.S. government. Indeed, Native American identity is typically defined by citizenship in a tribe or Native nation, regardless of an individual’s racial background.

In this regard, the continued treatment of Native Americans as a “race” rather than a diverse group of nations that predate the United States may have significant downstream political implications. One example is the currently pending Supreme Court case Haaland v. Brackeen, which deals with the constitutionality of the Indian Child Welfare Act (ICWA). One of the most contested ICWA provisions prioritizes placing Native American children in the adoption and foster care system with family members, fellow tribal citizens, and Native American families before they are placed with non-Native American families. A large part of the case hinges on whether those provisions are constitutionally protected congressional actions reflecting the nation-to-nation relationship between the United States and Native nations, or whether they are instead an unconstitutional racial preference for Native Americans that other racial groups do not get. While federal government data collection has no direct bearing on how the Supreme Court will rule in the case, the broader treatment of Native Americans as a “race”—which is underscored in government data collection and reporting—may influence how non-Native people perceive or misunderstand Native American identity.

Recommendations for changing data collection on Native Americans

The problems outlined above suggest a need for the U.S. government to rethink how it collects and publishes data about Native Americans. As the OMB continues the process of reworking racial classifications for the 2030 decennial census and beyond, policymakers should explore alternative models of how to most accurately collect and aggregate data on Native Americans.

Other countries have taken steps to separate data collection around Indigenous populations from racial identity. For example, both Canada and Australia ask about Indigenous identity in a separate census question, rather than as only one option among several races. Indigenous peoples in the United States share a similar position to Indigenous peoples in Canada and Australia, especially given those countries’ shared history of European (in particular, British) settler colonialism, and respective modern perceptions as western, developed nations. Creating a separate question or questions to ask about Native American identity—with a write-in for tribal affiliation—in addition to the current racial question could be one way to begin delinking the concept of indigeneity from “race,” as well as get a more accurate picture of the total Native American population.

In the more immediate term, the OMB could take steps to promote the proliferation of more comprehensive data around Native American populations. One way to do so would be by encouraging federal government agencies to publish public data on American Indian and Alaska Native populations alone and in combination with other groups, in addition to just single-race American Indian and Alaska Native data. One example of a federal government-affiliated organization that is already doing so is the Minneapolis Fed’s Center for Indian Country Development, which publishes labor market data about individuals classified as American Indian or Alaska Native alone or in combination with another race through its Native American Labor Market Dashboard. This tool contrasts with the BLS’s official monthly employment data releases, which report data only on single-race American Indian or Alaska Native individuals.

Finally, the U.S. government can and should do more to empower tribes to collect and manage data about their own populations and territories. The growing data sovereignty movement emphasizes that tribes themselves are sovereign entities that share a nation-to-nation relationship with the United States government, and therefore should be empowered to lead on data collection about their own lands and citizens. Partnerships to support greater data sovereignty could be another avenue for tribes and other Indigenous groups to provide input into how government and non-government data can better meet their own needs and experiences, rather than just the needs of the federal government.

Regardless of how the U.S. government chooses to proceed, it’s clear that the current practice of measuring Native Americans using mutually exclusive, single-race data is not working well. Moreover, the growing population of mixed-race Native Americans may foreshadow broader demographic trends in the country as a whole. Given that, it is time for both the federal government and non-governmental researchers to rethink how they measure Native American identity, as well as reconsider the broader use of mutually exclusive single-race categories upon which U.S. data publication has long relied.