Abstract

Data on the health and social determinants for Native Hawaiians and Pacific Islanders (NHPIs) in the United States are hidden, because data are often not collected or are reported in aggregate with other racial/ethnic groups despite decades of calls to disaggregate NHPI data. As a form of structural racism, data omissions contribute to systemic problems such as inability to advocate, lack of resources, and limitations on political power. The authors conducted a data audit to determine how US federal agencies are collecting and reporting disaggregated NHPI data. Using the COVID-19 pandemic as a case study, they reviewed how states are reporting NHPI cases and deaths. They then used California's neighborhood equity metric—the California Healthy Places Index (HPI)—to calculate the extent of NHPI underrepresentation in communities targeted for COVID-19 resources in that state. Their analysis shows that while collection and reporting of NHPI data nationally has improved, federal data gaps remain. States are vastly underreporting: more than half of states are not reporting NHPI COVID-19 case and death data. The HPI, used to inform political decisions about allocation of resources to combat COVID-19 in at-risk neighborhoods, underrepresents NHPIs. The authors make recommendations for improving NHPI data equity to achieve health equity and social justice.

For decades Native Hawaiian and Pacific Islander (NHPI) leaders in the United States have advocated for disaggregated data that will allow social and health issues to no longer remain invisible in the public eye (Chang, Penaia, and Thomas 2020; Office of Hawaiian Affairs 1982; OMB 1997). NHPIs are diverse, with origins ranging across the Pacific regions of Polynesia, Melanesia, and Micronesia (Hixson, Hepler, and Kim 2012). In the United States, NHPIs account for 0.4% of the population (about 1.4 million people) and are one of the fastest-growing populations (US Census Bureau 2020). Yet data for NHPIs are often hidden as a result of gaps in data collection and reporting (Kana'iaupuni 2011; Panapasa, Crabbe, and Kaholokula 2011; Taualii et al. 2011).1 The result is that the issues that need attention in NHPI communities are made invisible.

The lack of collected and reported NHPI data equates to a form of structural racism that disproportionately harms NHPI communities (Morey et al. 2020). Structural racism is defined as the ways in which society fosters racial discrimination via macrolevel systems, institutions, ideologies, and processes that result in the reinforcement of discriminatory values, beliefs, and distribution of resources throughout history (Bailey et al. 2017; Gee and Ford 2011). Often supported by interconnected institutions and policies, structural racism does not need to be initiated by a particular individual or group of individuals with racist intent. Rather, structural racism can result from subconscious or automatic disparate treatment that results in harm to historically oppressed people of color (Reskin 2012). Historical and continued oppression of NHPIs can be attributed to settler colonialism—the occupation of Indigenous lands by a society of settlers through the forcible removal of Indigenous peoples—which results in the continued erasure of these populations in public discourse (Tuck and Yang 2012).

Structural racism and settler colonialism manifest to harm NHPI communities through data gaps and limitations. With limited data on health disparities, public health efforts to support NHPI health are underresourced (Samoa et al. 2020). NHPI policy advocates experience decreased political power because of lack of data on social and health inequities, limiting their ability to advocate for policy changes (Morey et al. 2020). The complete omission of NHPI data or aggregation of it with other racial groups, often with Asian Americans, reinforces the marginalization that NHPIs experience in US society (Chang, Penaia, and Thomas 2020; Kaholokula et al. 2020). In this article, we contend that social and health equity for NHPIs can be achieved when there is equity in the collection and reporting of data, especially in conjunction with community-based mobilization of health-promoting measures informed by those data.

Background

On October 30, 1997, the OMB announced revisions to the standards for the classification of federal data on race and ethnicity (OMB 1997). This notice, which revised the initial classifications provided by Statistical Policy Directive Number 15 (also known as OMB 15), created 30 years prior, included three major modifications: (1) treating the Asian or Pacific Islander category as two separate categories—“Asian” and “Native Hawaiian or Other Pacific Islander,” (2) changing the term “Hispanic” to “Hispanic or Latino,” and (3) allowing more than one self-identified race.2 The implementation of these revisions represented a major milestone for those identifying as Native Hawaiian or Other Pacific Islander, reflecting hard-fought efforts to advocate for changes to the standard race classifications that previously aggregated Asians and Pacific Islanders together.3

The aggregation of NHPIs with Asians rendered NHPI health and social inequities invisible, because Asians represent a significantly larger population that is more socioeconomically advantaged on average (Kana'iaupuni 2011; Panapasa, Crabbe, and Kaholokula 2011; Taualii et al. 2011). Compared to Asian American populations as an aggregate group, NHPI populations experience higher rates of chronic and infectious disease and have very different profiles regarding the social determinants of health leading to such health inequities, including lower educational attainment, higher rates of poverty, and limited access to preventive health care (Hixson, Hepler, and Kim 2012; US Census Bureau 2020). Disaggregating NHPIs from Asians acknowledges these experiences, which also reflect differences in the histories, cultures, languages, and ancestries of these groups (Hosaka, Castanera, and Yamada 2021).4 The OMB's racial and ethnic categories are important because they set the minimum standard for federal data on the classification of race/ethnicity used to produce demographic data as well as to monitor civil rights enforcement and inform program implementation.

The revisions to the OMB 15 race and ethnic classification standards in 1997 did not arise automatically. The process began in 1993, when the OMB underwent a comprehensive review of the categories used to measure race and ethnicity (OMB 1997). This occurred after the OMB received criticism following the 1990 US Census from members of the public who felt that the minimum categories inadequately reflected the diversity of the nation's population. The comprehensive review of the OMB racial/ethnic classifications included hearings, testimony, and a research agenda by the Interagency Committee to evaluate the effect of possible changes to the racial and ethnic categories. In 1997, the OMB released a Federal Register notice (62 FR 36874—36946) requesting public comment on the Interagency Committee's Report to the OMB on the Review of Statistical Policy Directive No. 15.

However, the Interagency Committee recommended that data on Native Hawaiians continue to be classified in the “Asian or Pacific Islander” category. In response, the OMB received approximately 300 letters and 7,000 individually signed and mailed preprinted yellow postcards specifically on the issue of classifying Native Hawaiian data separately from Asians. The OMB additionally received about 500 signed form letters from members of the Hapa5 Issues Forum in support of reporting multiple races. More than half of NHPIs identify as multiracial (Hixson, Hepler, and Kim 2012; US Census Bureau 2020). The 7,000 individuals who signed and sent preprinted yellow postcards, the Hawaiian congressional delegation, the departments and legislature of the government of the state of Hawai‘i, Hawaiian organizations, and individual advocates strongly opposed the Interagency Committee's recommendation. Their arguments supported reclassifying Native Hawaiians with American Indians or Alaska Natives, given their identification as the original inhabitants of Hawai‘i. Their comments further expressed that disaggregated data were needed to monitor the socioeconomic conditions of Native Hawaiians and to address systematic discrimination against this population in housing, education, employment, and other sectors. At the time, Native Hawaiian advocates did not request a separate category for Native Hawaiians, because the Interagency Committee had expressed opposition to adding more race categories to the original four OMB 15 race categories (American Indian or Alaska Native, Asian or Pacific Islander, Black, and white). In the end, the OMB decided to add the fifth category, splitting the “Asian or Pacific Islander” category into two: “Asian” and “Native Hawaiian or Other Pacific Islander.” The latter was defined as a “person having origins in any of the original peoples of Hawai‘i, Guam, Sāmoa, or other Pacific Islands.” At the time, it was estimated that about 60% of the NHPI category would consist of Native Hawaiians, but it would also include Carolinians, Fijians, Guamanians (Chamorros), Kosraeans, Melanesians, Micronesians, Northern Mariana Islanders, Palauans, Papua New Guineans, Ponapeans (Pohnpelans), Polynesians, Sāmoans, Solomon Islanders, Tahitians, Tarawa Islanders, Tokelauans, Tongans, Trukese (Chuukese), and Yapese. The revised race and ethnicity OMB standards reflect a federal review process that was shaped by the urgent desires of NHPI community members and organizations.

By treating NHPIs as a separate race category, the 1997 revised OMB 15 standards allowed for greater attention to be paid to the health, social, and economic issues that would affect NHPI populations in the future. The important implications of this disaggregation of NHPIs from the “Asian Pacific Islander” category become apparent in times of crisis, including during the COVID-19 pandemic. Initial COVID-19 disaggregated data in the states of Arkansas, California, Colorado, Hawai‘i, Oregon, Utah, and Washington (some of the first states to report COVID-19 data by race for NHPIs) in the spring of 2020 revealed that NHPIs were experiencing the highest rates of COVID-19 cases and deaths of any other racial/ethnic group in those states (Chang, Penaia, and Thomas 2020). These early reports led a coalition of NHPI community leaders to form the National Pacific Islander COVID-19 Response Team (NPICRT) (Samoa et al. 2020). The NPICRT championed the formation of the NHPI Data Policy Lab—housed at the UCLA Center for Health Policy Research—comprising researchers, data analysts, and policy advocates who would consolidate and represent NHPI data to inform COVID-19 advocacy efforts from the local to the national levels. However, as members of the NHPI Data Policy Lab quickly learned, there was and continues to be inconsistent collection and reporting of NHPI COVID-19 case and death data across states and localities, obstructing grassroots efforts to respond to NHPI community needs in those areas during the pandemic (Chang, Penaia, and Thomas 2020).

Given this context, an up-to-date review of compliance with the 1997 revised OMB 15 standards is warranted. In 2011, Panapasa, Crabbe, and Kaholokula (2011) reviewed data sources from federal agencies for compliance with the 1997 revised OMB 15 standard on the collection and reporting of NHPI data. They found that while these data sources were collecting disaggregated NHPI data appropriately, the vast majority of the data sources were not reporting NHPI data. Panapasa, Crabbe, and Kaholokula highlighted the ongoing problems with data reporting for NHPIs that are the result of inadequate sample sizes or inappropriate reaggregation of NHPIs into “Asian American or Pacific Islander” or “Other race” groups. The authors made recommendations to increase efforts to oversample NHPI populations, create reliable data estimates, and partner with NHPI communities in federal data sources. Nevertheless, special surveillance efforts are often needed. An example is the 2014 NHPI National Health Interview Survey (NHPI NHIS), the first and largest nationally representative survey of NHPI health conducted by the Center for Disease Control and Prevention's National Center for Health Statistics (National Center for Health Statistics 2014).6

NHPIs continue to be systematically missed in efforts to achieve health equity (Morey et al. 2020). In recent years, more attention has been given to issues of neighborhood inequity, including environmental injustices that are the result of the overlapping issues of residential segregation, concentrated poverty, decreased political power, disproportionate pollution burden, poor health infrastructure, lack of green space, unsafe built environments, and more (Diez Roux and Mair 2010; Pastor and Morello-Frosch 2014). While these are important issues, NHPIs have often been excluded from efforts that mitigate neighborhood injustices (Morey 2014). It is more common now to rely on indices that calculate neighborhood social disadvantage and disease risk in plans for the distribution of limited resources (Maizlish et al. 2019). Unfortunately, NHPI community members report that these neighborhood measures often miss NHPI populations. Therefore, policies that rely on these widely used neighborhood indices may systematically exclude NHPIs—another example of structural racism.

In the current study, we assess data equity for NHPIs as structural racism in three ways. First, we reassess the federal data sources reviewed by Panapasa, Crabbe, and Kaholokula (2011) 10 years ago for compliance with the revised OMB 15 standards for collecting and reporting NHPI data, adding some additional data sources that are relevant for understanding health and social determinants of health for NHPI populations nationally. Second, using the COVID-19 pandemic as a case study, we review the public availability of NHPI case and death data for COVID-19 by state. Third, we evaluate within the state of California the use of a health equity metric—the California Healthy Places Index—as an indicator of neighborhood disadvantage, to determine whether NHPIs and other communities of color are underrepresented in “high risk” neighborhoods. The goal of these three steps is to demonstrate how data inequity operates on a national, state, and local level, with implications for health equity and social justice efforts for NHPI populations.

Methods

Review of Federal Data Sources' Collection and Reporting of NHPI Data

In the first analysis, we reviewed national data based on those datasets first reviewed by Panapasa, Crabbe, and Kaholokula (2011) to determine progress in the past 10 years on the collection and/or reporting of NHPI data. The 2011 paper originally reviewed data from six federal agencies: the Department of Commerce, the Department of Health and Human Services, the Department of Education, the Department of Agriculture, the Department of Housing and Urban Development, and the Department of Justice. We reviewed 19 of the original 20 data sources.7 In addition, we selected 10 other national datasets to review, based on the criteria originally used to select datasets: (1) accessibility, (2) degree of national coverage of the US population, and (3) potential source of information for policy and intervention. We added a fourth criterion: collection of data is current and ongoing. In total, we reviewed 29 national datasets. The data sources are not an exhaustive list, but they represent datasets that collect and report race/ethnicity that could be useful for informing future policy decisions or to conduct research illuminating health disparities and their underlying social determinants. For each data source, we examined the public websites to determine whether NHPI data were being collected and reported, and if so, how these data were being collected and reported.8 At least two authors examined the public websites for each data source for evidence (i.e., text descriptions of available data, links to datasets, data outputs, codebooks, questionnaires, etc.) of how data on race was being collected and reported in the survey. This allowed us to determine compliance with the revised OMB 15 standards and to assess the level of disaggregation of NHPI data. We also made note of whether NHPI data collection or reporting had changed from 2011 to 2021 in the data sources previously reviewed.

Review of NHPI US COVID-19 Case and Death Data in States

To assess COVID-19 data in states, we used data from the NHPI COVID-19 Data Policy Lab Dashboard (UCLA Center for Health Policy Research 2021). This dashboard systematically reports NHPI COVID-19 case and death rates in states that disaggregate NHPI data. The dashboard collected counts of COVID-19 cases and deaths from the COVID Tracking Project and the Hawai‘i COVID-19 Dashboard, and it calculated rates using American Community Survey 2015–2019 five-year population estimates (“The Covid Tracking Project” 2021; Hawai‘i Department of Health 2021; US Census Bureau 2020). Of states that do not report disaggregated NHPI data, the dashboard provides information on how NHPI data are being treated in those states. We used these data to calculate the number and percentage of states in each of these categories separately for COVID-19 cases and deaths: (1) reports disaggregated NHPI data, (2) uses the obsolete panracial “Asian Pacific Islander” category, (3) specifies NHPI data under the “Other race” category, (4) does not specify an NHPI reporting practice, (5) does not report any race/ethnicity data, or (6) does not disaggregate NHPI death data (for COVID-19 death rates only). Data were up to date as of February 21, 2021.

Evaluation of the California Healthy Places Index in Representing NHPIs in California Census Tracts

On August 28, 2020, California Governor Gavin Newsom announced the “Blueprint for a Safer Economy,” which included a health equity metric—the California Healthy Places Index (HPI)—that would be used to determine which counties could move to less restrictive reopening tiers.9 The stated purpose of applying the health equity metric was to incentivize a reduction in disease transmission for all communities, especially those disproportionately impacted by COVID-19. We downloaded HPI data for California census tracts from the Public Health Alliance of Southern California's website (Public Health Alliance of Southern California 2021). The HPI provides an index score for all 2010 California census tracts with a population of 1,500 or more. The HPI includes 25 different community characteristics combined into a single score at various geographic levels. The 25 characteristics fall into eight policy action domains, including economic (e.g., income), social (e.g., two-parent households), education (e.g., bachelor's degree or higher), transportation (e.g., automobile access), built environment (e.g., park access), housing (e.g., homeownership), clean environment (e.g., ozone), and health care (e.g., insured). Notably, the HPI does not include measures of race or ethnicity to allow state agencies to remain compliant with California Proposition 209, which prohibits the use of race or ethnicity for allocating public resources.10 Each included domain was weighted, contributing to an overall HPI score. Based on the distribution of the HPI score across California census tracts, the HPI places census tracts into four numbered quartiles of neighborhood disadvantage, with quartile 4 indicating the highest level of neighborhood disadvantage (i.e., bottom quartile). These quartiles are used to make public policy decisions about which neighborhoods are most disadvantaged, with the bottom-quartile neighborhoods representing those that may be identified for additional resources during public health emergencies.

We then determined whether racial/ethnic groups are underrepresented in these most disadvantaged neighborhoods as defined by the HPI. We used American Community Survey 2015–2019 5-year data to determine the population of each of six OMB single racial/ethnic groups (Hispanic/Latino, white, Black, American Indian/Alaska Native, Asian, or NHPI) in counties and census tracts (US Census Bureau 2020). To determine whether the HPI underrepresented each racial/ethnic group, we calculated whether the total percentage of the racial/ethnic group residing within bottom-quartile (most disadvantaged) census tracts in a county according to the HPI was lower than the percentage of the racial/ethnic group in the county's total population. Using this standard, of the total 43 counties in California with census tract level HPI, we identified the number of counties where each racial/ethnic group is underrepresented by the HPI in bottom-quartile census tracts. We then calculated the percentage of counties that underrepresent that group by dividing the number of counties by 43 and multiplying by 100. We listed the California counties with a population greater than 150,000 where the HPI underrepresents communities of color in bottom-quartile census tracts (29 of the total 43 California counties have populations greater than150,000). Underrepresentation was conceptualized this way since the California Blueprint for a Safer Economy would likely incentivize resources for these census tracts in the lowest HPI quartile. However, it was unclear whether these resources would help NHPI communities or underrepresent them despite these communities having the highest statewide COVID-19 case rates of any racial/ethnic group.

Results

Review of Federal Collection and Reporting of NHPI Data

Table 1 displays the results of our review of 29 sources of federal data for compliance with the 1997 revised OMB 15 standards for collecting and reporting NHPI data. Of the 29 federal data sources, the majority (26, or 90%) are collecting data for NHPIs as a separate race category. The three data sources that are not in compliance with the revised OMB 15 standards are either collecting data inconsistently by state, are erroneously collecting data using the panracial Asian or Pacific Islander category, or are no longer collecting race data. Of the 29 federal data sources, 19 (66%) are reporting data for NHPIs as a separate race category. When data for NHPIs are not being reported separately, it is usually the result of NHPI data being reported in aggregate with the Asian race category or “other” race category, or it is unclear how NHPI data are being treated.

There was an improvement in the reporting of NHPI data from 2011 to 2021. Of the 19 data sources originally reviewed by Panapasa, Krabbe, and Kaholokula (2011), nine (47%) improved their data-reporting practices and now report NHPI data as a separate race category. In some of these cases, the public data are available but require some downloading of public-use data files and statistical software to access the NHPI data. On the other hand, 6 of the 19 originally reviewed data sources (32%) fail to provide disaggregated NHPI data 10 years later.

Review of State Reporting of NHPI COVID-19 Cases and Deaths

Figure 1 presents maps created by the NHPI COVID-19 Data Policy Lab Dashboard, showing US COVID-19 NHPI cases and deaths by state. As of February 21, 2021, there were 52,695 reported NHPI cases (Figure 1A) and 798 reported NHPI deaths (Figure 1B) in the United States. At that time, the states with the highest NHPI case rates were Louisiana, Alaska, Iowa, Illinois, Idaho, and Minnesota. The states with highest NHPI death rates were Louisiana, Iowa, Illinois, Arkansas, California, and Minnesota.

Table 2 shows that of the 50 states, only 20 (40%) are reporting disaggregated NHPI case data, and only 16 (32%) are reporting disaggregated NHPI death data. Of those not reporting disaggregated NHPI case and death data, nine states (18%) are using the obsolete panracial “Asian Pacific Islander” category, while five states (10%) are including NHPI data in the “other race” category. For the remaining states, it is unclear how NHPI data are being treated, or the states are not reporting COVID-19 data for NHPIs or by race/ethnicity at all. Of the states that are properly reporting disaggregated data, the NHPI rates per 100,000 population rank the highest of any racial group in 16 of 20 (80%) for COVID-19 cases and 11 of 16 (69%) for COVID-19 deaths.

Evaluation of the HPI in Representing NHPIs in California

Table 3 shows for each OMB racial/ethnic group the number and percentage of the 43 California counties where that group is considered underrepresented in the most disadvantaged (4th quartile) census tracts according to the HPI. Results show that the HPI underrepresented certain populations by race in California counties. Of communities of color, Asian Americans were most affected, with 34 (79%) of 43 counties underrepresenting Asian populations compared to the county's total population's percentage that reside within bottom-quartile census tracts. NHPIs were the second most affected, with 22 (49%) of 43 counties underrepresenting NHPI populations in bottom-quartile census tracts. American Indian/Alaska Native populations are underrepresented in 16 (37%) of 43 counties in the bottom-quartile census tracts. Latino/Hispanic populations were generally overrepresented in the majority of 4th-quartile tracts ranked by the HPI. Table 4 lists the California counties that underrepresent communities of color in the 4th quartile of the HPI, out of the 29 counties with populations greater than 150,000 people. The counties that underrepresent Asian, NHPI, American Indian/Alaska Native, and Black populations are listed separately for each racial group.

Discussion

Members of the NHPI community have long advocated for greater representation in data as an issue of data equity (Chang, Penaia, and Thomas 2020; Office of Hawaiian Affairs 1982; OMB 1997; Panapasa, Crabbe, and Kaholokula 2011). Disaggregated NHPI data are instrumental in supporting program implementation and policy advocacy to address long-standing social and health inequities. On the other hand, omissions of NHPI data through data-collection gaps or inappropriate aggregation of data in reporting are a form of structural racism and an extension of settler colonialism that stymies the passage and implementation of more inclusive public policies (Morey et al. 2020; Tuck and Yang 2012). This article represents a review of publicly available data at the national, state, and local levels that could support public health and public policy efforts intended to benefit NHPI populations through data disaggregation in accordance with the revised OMB 15 standards. By reviewing these data, we aimed to evaluate the current state of data equity for NHPIs.

Our analysis of US federal data compliance with the revised OMB 15 standard for reporting NHPIs separately from Asian Americans found that there has indeed been progress since these same datasets were reviewed 10 years ago by Panapasa, Crabbe, and Kaholokula (2011). Our findings that some federal datasets that were not previously reporting disaggregated NHPI data are now in compliance with the revised OMB 15 standards indicate the success of many years of advocacy efforts by NHPI community members. Nevertheless, there is still work to be done. This is especially true of health data from the Department of Health and Human Services. Seven out of 13 federal health data sources are not reporting NHPI data separately from other racial/ethnic groups. One health data source, the Web-based Injury Statistics Query and Reporting System (WISQARS), is not collecting NHPI data in accordance with the revised OMB 15 standards. The remaining gaps in reporting are likely the result of insufficient sample sizes among the datasets that are collecting disaggregated NHPI data but not reporting the data. Many population-based samples, especially for health surveys, are limited in their reporting of NHPI data because they do not collect large enough samples of NHPIs to report the data publicly (Panapasa, Crabbe, and Kaholokula 2011).

Small sample sizes among NHPIs are a long-standing problem, as statistical estimates resulting from these small samples are often unstable. Confidentiality is a potential problem that limits the release of data for a small number of people who may be identifiable. At times, sufficiently large sample sizes can be obtained for NHPIs by pooling data across multiple years of data collection (Subica et al. 2017). However, such efforts often require accessing the restricted data files for these federal datasets. Accessing restricted data is not easy and is at times impossible because of confidentiality concerns. There are financial, time, geographic, and skill set barriers that prevent most researchers from enduring the arduous process of accessing restricted federal data. Therefore, researchers and data analysts studying relatively small and underresourced populations, such as NHPIs, must pay greater penalties to access the data they need, which may also underrepresent the needs of the population. There have been efforts to mitigate this problem. For example, the NHPI NHIS in 2014 collected a separate nationally representative NHPI sample to estimate the prevalence of disease in this population for the first time (National Center for Health Statistics 2014). We recognize that there are greater financial costs of oversampling smaller populations. Nevertheless, these costs are outweighed by the health, societal, and financial costs associated with overlooking inequities for minoritized populations, which become compounded over time. More efforts are needed to make sure NHPIs are included in nationally representative surveys and that collected data are made available for the public to access to inform policy decisions.

Timely and transparent data are extremely important to inform public health efforts, especially during a global pandemic. As the COVID-19 pandemic has shown, the numbers are constantly changing, as are the corresponding scientific and policy recommendations. From the early days of the pandemic, states and counties were reporting extremely high rates of COVID-19 cases and deaths among NHPI populations (Chang, Penaia, and Thomas 2020; Morey et al. 2020). The formation of the NHPI Data Policy Lab allowed for these data to be consolidated and disseminated, supporting local, state, and national efforts to garner resources to address the disproportionate effects of COVID-19 on NHPI populations (Samoa et al. 2020). As the NHPI Data Policy Lab Dashboard shows, NHPI populations continue to be greatly impacted by the pandemic, with the highest rates of COVID-19 cases and deaths among all racial/ethnic groups in the majority of states that report NHPI data (UCLA Center on Health Policy Research 2021). Although NHPI populations are found in all 50 states, most states are not disaggregating NHPI case and death data. It is unclear how NHPI data are being specified in some states. Eight states are reporting NHPI data with Asian data in a panracial “Asian Pacific Islander” category, while five states are consolidating NHPI data in the “other race” category.

Using the panracial “Asian Pacific Islander” category violates the revised OMB 15 standards and inflicts harm on NHPI communities (OMB 1997; Panapasa, Crabbe, and Kaholokula 2011). Although we recognize that NHPI and Asian panracial coalitions continue to collaborate to achieve common goals, when it comes to directing public resources to address social and health problems, more data disaggregation for NHPI and Asian subpopulations is crucial. In states where the majority of Asian Americans are experiencing lower COVID-19 case and death rates and also make up a larger proportion of the population than NHPIs, the plight of NHPIs is obscured, hiding disparities (Chang, Penaia, and Thomas 2020; Ponce, Shimkhada, and Tulua 2021). In North Carolina, data were showing that NHPIs were experiencing the highest COVID-19 death rates in the state. However, for reasons unknown, the state began aggregating NHPIs with Asian Americans, thus hiding the disparity within the racial group currently experiencing the lowest death rates in the state (UCLA Center for Health Policy Research 2021). Therefore, aggregating NHPIs with Asian Americans commits harm against NHPI communities, limiting their ability to advocate for resources to combat the pandemic. In a situation as dire as the COVID-19 pandemic, disaggregated NHPI data are desperately needed to mobilize efforts to save lives. This is why even though statisticians and epidemiologists have cited problems with small numbers, including potential anonymity issues, NHPI advocates have been calling for the release of NHPI COVID-19 case and death data as a separate race category, regardless of the size of the numbers (Morey et al. 2020; Samoa et al. 2020). The handling of NHPI COVID-19 data influences life-and-death decisions about whether NHPI communities are included in plans for equitable COVID-19 response.

The exclusion of NHPIs in equity plans to combat the COVID-19 pandemic becomes clear at the local level. In the state of California, the HPI is being used to inform the distribution of COVID-19 resources—including vaccines—to the neighborhoods considered most disadvantaged (Lin II, Money, and Shalby 2021). However, our analysis shows that the HPI underrepresents NHPI populations, even while NHPI populations are experiencing the highest COVID-19 case rate (10,572 per 100,000) and death rate (204 per 100,000) in California compared to all other race and ethnic groups. While the HPI by design does not include neighborhood data on race/ethnicity because of Proposition 209, its purpose is to allocate resources to the areas most affected by the pandemic. In the case of NHPI populations who are suffering in the pandemic, the HPI underrepresents them. This may be due in part to NHPI and other smaller populations such as American Indians/Alaska Natives being more spread out and less concentrated than larger minoritized populations experiencing residential segregation. Furthermore, place-based measures may systematically bias against socioeconomic dimensions of household composition. The HPI measures socioeconomic status using median household income, which is artificially inflated for NHPI households, which tend to be large, multigenerational, and multifamily (Delaney et al. 2018). This unintended bias against NHPI communities embedded in neighborhood socioeconomic measures is a form of structural racism. In the absence of allowing race data in California to be considered as part of a plan for equitable distribution of resources because of Proposition 209, other metrics might be more relevant to capture neighborhood risk for groups such as NHPIs in equity metrics such as the HPI. For example, per capita income can be used instead of median household income, which will address the problem of underestimating the socioeconomic needs of families living in large multifamily homes with several income earners.

The underestimation of NHPI needs by only focusing on the most disadvantaged neighborhoods identified by area-level metrics like the HPI demonstrates a form of structural racism that often persists unnoticed (California Pan-Ethnic Health Network 2021). Increasingly, health organizations in the government, nonprofit, and for-profit realms are relying on similar metrics and thresholds to make decisions about how to target resources. The availability of big data allows for these types of metrics to be created and used widely, with concrete consequences. Recent research has shown decisions that are “race neutral” on the surface can end up unjustly disadvantaging communities of color and propagating societal biases (Obermeyer et al. 2019; Zou and Schiebinger 2018). Although California Proposition 209 was originally marketed as a civil rights initiative to make public policy decisions in a “color blind” way, evidence from this study and others has shown that when public and private institutions ignore race/ethnicity completely, conscious and unconscious biases against communities of color proliferate (Kidder 2013). Most of the time, the average consumer is unaware of how these metrics work or how they were developed, even though they may have serious implications for health and social equity. Therefore, we shed light on the potential weaknesses of the large-scale use of some metrics to determine the allocation of resources that may disadvantage the communities of color that need the most resources to combat racial injustices, including NHPIs. We recognize that there is great potential with the increasing use of big data to make program and policy decisions using health equity metrics. At the same time, we caution that these metrics must be applied with transparency and with deliberate attention paid to the problems of racial inequity that can result from their use (Green 2020).

Recommendations

Having demonstrated the importance of data equity as fundamental to achieving social and health justice for NHPIs, we provide the following recommendations. First, at the national, state, and local levels data must be collected and reported in accordance with the revised OMB 15 guidelines set in 1997 wherever this is not already occurring. All systems currently collecting race data should be collecting data for NHPIs separately from Asian Americans and from other race categories. This is the same call to action that has been ongoing for decades (Chang, Penaia, and Thomas 2020; Office of Hawaiian Affairs 1982; Panapasa, Crabbe, and Kaholokula 2011). Every effort should be made to report disaggregated NHPI data and to make these data easily and publicly accessible, in accordance with the revised OMB 15 standards. As a majority of NHPIs are multiracial but also strongly identify being NHPI, we recommend that statisticians consider including multiracial NHPIs in a separate “multiracial NHPI” category or in the larger NHPI category. This should be done with transparency, noting how data for multiracial people are being treated. Although epidemiologists and statisticians often hesitate to report the small numbers for NHPIs because of unstable rates or not reaching a certain statistical threshold, we contend that the data should be reported anyway, with caveats outlining the limitations of the data. As in the case of the COVID-19 pandemic, these numbers were essential in the early days of the crisis to mobilize grassroots community responses to the spread of the virus, even when the initial numbers were low (Samoa et al. 2020). Making data more transparent allows communities to make informed decisions and to understand how their data are being treated. The agencies collecting population data should realize the power that they wield when making decisions about which data to make publicly available. NHPIs and other relatively smaller populations have a higher transaction cost to access their own community's data. Therefore, agencies should make efforts to lower these costs for communities who, like NHPIs, are underrepresented so that community researchers have equitable and ethical access to data.

Second, recognizing that the sample sizes of NHPIs collected at the federal level are often not large enough to be reportable or to inform decisions, we recommend a second round of NHPI NHIS data collection. The 2014 NHPI NHIS has been extensively used to report on NHPI disparities (Narcisse et al. 2020). A subsequent iteration of the NHPI NHIS will bolster the sample size and provide more accurate surveillance of NHPI health nationally.

Third, when possible, NHPI data should be further disaggregated into subpopulations given the diverse languages, cultural practices, and histories of each of the Pacific Island groups that have been impacted by settler colonialism, militarization, and migration in ways distinct from one another. For example, Native Hawaiians have experienced the historical trauma of having lands and culture stripped away by the US government (Dougherty 1992; Kaholokula et al. 2020; Ong 2009). Pacific Islanders from Federated States of Micronesia, Republic of Marshall Islands, and Republic of Palau who are under the Compacts of Free Association (COFA) were subjected to severe health consequences and loss of land as a result of the US government's nuclear testing on the islands from 1946 to 1958. Although COFA migrants are allowed to live and work in the United States,11 they were denied access to Medicaid under the 1996 Personal Responsibility and Work Opportunity Reconciliation Act until Congress restored Medicaid access in December 2020 (Asian and Pacific Islander American Health Forum 2020; McElfish, Hallgren, and Yamada 2015). Chamorros, the Indigenous people of Guam, have survived centuries of settler colonialism, first by Spain and then by the United States as an unincorporated territory, and their territory is used as a military outpost.12 Although people born and living on Guam are considered US citizens, they are denied constitutional protections such as the right to vote in presidential elections. American Sāmoa has had a similar history of militarization and settler colonialism as an unincorporated US territory. American Sāmoans are considered US nationals and must go through the naturalization process to earn the rights of US citizenship, such as applying for certain jobs or voting in presidential elections (Asian Americans Advancing Justice 2014).13 Although we cannot illuminate each unique story of the many Pacific Islands here, we provide these few examples to demonstrate how various historical contexts and political forces differentially shape the well-being and social standing of NHPI subpopulations in the United States. Therefore, fine-grained data are needed to highlight the diversity within the NHPI aggregate grouping.

Fourth, stronger partnerships are needed between government, academic, and community-based organizations to increase NHPI sample sizes and to make data more useful. Institutions should listen to and learn from NHPI voices to understand the types of data outputs that will be most appropriate and to make data-collection efforts more effective. Egalitarian relationships, open communication, and sensitive outreach to NHPI community organizations will allow for the improvement of data collection and quality. Furthermore, institutions of higher education and national funders should invest in building capacity among NHPI community organizations to support the next generation of researchers and data scientists who understand the specific needs of NHPI populations. Building stronger infrastructure within the NHPI community will enable grassroots efforts to use data to inform policy and programmatic solutions. Such commitments will help to mitigate the systemic underinvestment in communities of color such as NHPIs.

Fifth, we recognize that in the era of big data and machine learning, summary metrics can be inherently biased against people of color, especially populations such as NHPIs who are underrepresented in data systems to begin with. Therefore, we recommend that future metrics must be created with a careful and deliberate focus on equity (Green 2020). The consideration of equity should apply broadly to communities of color, including NHPI populations. Metrics used to determine distribution of resources must be made transparent to allow the public to evaluate whether these metrics are truly effective and equitable. As such metrics are being created and applied increasingly widely, we recommend the purposeful evaluation of the effects of using these metrics for public policy decisions on racial equity. People creating and applying these metrics must be educated on issues of racial equity. These steps will help to ensure that policy decisions made based on these metrics do not perpetuate and exacerbate racial biases that exist in society.

Conclusion

We recognize the great progress that has been made in the collection and reporting of data for NHPI populations in the United States, largely because of the grassroots efforts and advocacy that have been ongoing from NHPI community members for decades. However, our work is far from complete. We continue to advocate for the appropriate disaggregation of NHPI data to achieve equity. By achieving data equity, our hope is that future generations will be able to achieve health and social equity for all communities of color.

Acknowledgments

This work was supported by grants from the Robert Wood Johnson Foundation (grant #79504, “A Model for Data-Driven Policy Making in the NHPI Community”) and the National Institute on Minority Health and Health Disparities (NIH R01 MD012292). We acknowledge and thank Sela V. Panapasa, Kamana'opono M. Crabbe, and Joseph Keawe'aimoku Kaholokula for their original examination of federal data compliance with the 1997 revised OMB 15 standards for collecting and reporting NHPI data.

Notes

1.

Data may be collected improperly in surveys that do not provide a separate race category for NHPI identification on forms. Data may be reported improperly by aggregating NHPIs with Asians or other race categories prior to releasing the data.

2.

Allowing people to self-identify with more than one race on forms, rather than automatically having people mark a single “multiracial” category, helps with identifying the diversity of multiracial people. This is important for NHPIs, more than half of whom are multiracial but who often strongly identify with their NHPI ancestry, as described below.

3.

Historically, NHPIs have often been aggregated with Asians to increase political and social influence, achieving common goals as a broad panethnic group (see Okamoto and Mora 2014). However, automatically using the “Asian and Pacific Islander” panethnic category beyond the purpose of these efforts has also inflicted harm on the relatively smaller NHPI grouping (see Tuck and Yang 2012).

4.

Although it is important to note that the OMB racial/ethnic standards for classification make it clear that these categories do not reflect scientific (i.e., biological or genetic) or anthropometric (i.e., phenotypic) distinctions, these categories may reflect social characteristics placed in the context of the experiences and histories of these groups (OMB 1997).

5.

Hapa is a Hawaiian word that traditionally refers to someone of mixed Native Hawaiian and foreign ancestry.

6.

Although the NHPI NHIS is nationally representative and the first of its kind, it is a stand-alone cross-sectional survey and is not incorporated into the National Health Interview Survey.

7.

One data source—the National Hospital Discharge Survey—was not included because data collection is no longer ongoing.

8.

Most surveys determine race/ethnicity by self-report, which is preferable and likely most accurate. Other data sources (e.g., death certificate data) use the report of a proxy (e.g., coroner or doctor) and may have lower accuracy.

9.

The HPI health equity metric applied to California counties with more than 106,000 residents. For a county to move to a less restrictive tier, it must meet specific COVID-19 case and test positivity rates within their lowest-quartile HPI (i.e., most disadvantaged) census tracts. The Blueprint for a Safer Economy: Equity Focus can be found at www.cdph.ca.gov/Programs/CID/DCDC/Pages/COVID-19/CaliforniaHealthEquityMetric.aspx.

10.

Proposition 209, approved by voters in 1996, is a constitutional amendment that reads: “The State cannot discriminate against or grant preferential treatment on the basis of race, sex, color, ethnicity or national origin in the operation of public employment, public education, and public contracting” (California Constitution article I, section 31). This amendment essentially bans the use of race/ethnicity or nationality as the basis of appropriating state resources, including resources to combat COVID-19.

11.

This is another example of settler colonialism operating. The land and waters were seen as a valuable military outpost to be used for nuclear testing to advance US power, while the people were treated as expendable. COFA serves to continue this legacy of settler colonialism, displacing Pacific Islanders from their native lands in exchange for the United States having continued access to the islands for military purposes, simultaneously exploiting COFA migrants' bodies and labor in the United States.

12.

Another example of militarization, Guam was long under rule by the US Navy, while Chamorros were often treated as expendable by the US government. For example, during World War II, Guam was bombed and seized by Japan only hours after the bombing of Pearl Harbor, leading to the suffering and death of many Chamorros (see Cultures of Commemoration: The Politics of War, Memory and History in the Mariana Islands by Keith L. Camacho).

13.

American Sāmoa is the only unincorporated US territory where people born there are not automatically considered US citizens. A recent (June 15, 2021) federal appeals court ruled that US citizenship should not be forced on American Sāmoans. This is in response to a lower court ruling siding with three people from American Sāmoa who sued to be recognized as US citizens. Some government leaders and community members in American Sāmoa have fought against automatic citizenship, which could disrupt traditions of communal land ownership. Still others argue that the naturalization path is too costly for American Sāmoans.

References

Asian Americans Advancing Justice
.
2014
. “
A Community of Contrasts: Native Hawaiians and Pacific Islanders in the United States
.” www.advancingjustice-aajc.org/sites/default/files/2016-09/2014_Community%20of%20Contrasts.pdf (accessed
October
19
,
2021
).
Asian and Pacific Islander American Health Forum
.
2020
. “
Medicaid Restoration for Compact of Free Association Communities
.” www.apiahf.org/focus/health-care-access/cofa/ (accessed
June
19
,
2021
).
Bailey, Zinzi D., Krieger, Nancy, Agénor, Madina, Graves, Jasmine, Linos, Natalia, and Bassett, Mary T.
2017
. “
Structural Racism and Health Inequities in the USA: Evidence and Interventions
.”
Lancet
389
, no.
10077
:
1453
63
. doi.org/10.1016/S0140–6736(17)30569-X.
California Pan-Ethnic Health Network
.
2021
. “
Nothing About Us Without Us: Can Area-Based Social Indices Effectively Advance Health Equity?
October
26
. cpehn.org/publications/nothing-about-us-without-us/
Chang, Richard Calvin, Penaia, Corina, and Thomas, Karla.
2020
. “
Count Native Hawaiian and Pacific Islanders in COVID-19 Data—It's an OMB Mandate
.”
Health Affairs Blog
,
August
27
. www.healthaffairs.org/do/10.1377/hblog20200825.671245/full/.
The COVID Tracking Project
.”
2021
.
Atlantic
. covidtracking.com/ (accessed
October
19
,
2021
).
Delaney, Tracy, Dominie, Will, Dowling, Helen, Maizlish, Neil, Chapman, Derek, Hill, Latoya, Orndahl, Christine, et al.
2018
.
“The California Healthy Places Index.” Public Health Alliance of Southern California
. healthyplacesindex.org (accessed
October
19
,
2021
).
Diez Roux, Ana V., and Mair, Christina.
2010
. “
Neighborhoods and Health
.”
Annals of the New York Academy of Sciences
1186
, no.
1
:
125
45
.
Dougherty, Michael.
1992
.
To Steal a Kingdom: Probing Hawaiian History
.
Waimanalo, HI
:
Island Style Press
.
Gee, Gilbert C., and Ford, Chandra L.
2011
. “
Structural Racism and Health Inequities
.”
Du Bois Review: Social Science Research on Race
8
, no.
1
:
115
32
.
Green, Ben.
2020
. “
Data Science as Political Action: Grounding Data Science in a Politics of Justice
.”
SSRN
,
August
29
. dx.doi.org/10.2139/ssrn.3658431.
Hawai‘i Department of Health
.
2021
. “
Hawai‘i COVID-19 Summary Metrics
.” Hawai‘i COVID-19 Data. health.hawaii.gov/coronavirusdisease2019/what-you-should-know/current-situation-in-hawaii/ (accessed
October
19
,
2021
).
Hixson, Lindsay, Hepler, Bradford B., and Kim, Myoung Ouk.
2012
. “
The Native Hawaiian and Other Pacific Islander Population: 2010
.”
US Census Bureau, Report #C2010BR-12
,
May
. www.census.gov/library/publications/2012/dec/c2010br-12.html#:∼:text=According%20to%20the%202010%20Census,one%20or%20more%20other%20races.
Hosaka, Kalei R. J., Castanera, Max P., and Yamada, Seiji.
2021
. “
Structural Racism and Micronesians in Hawaii: The COVID-19 Syndemic
.”
Asia Pacific Journal of Public Health
,
May
3
. doi.org/10.1177/10105395211012188.
Kaholokula, Joseph Keawe‘aimoku, Miyamoto, Robin E. S., Hermosura, Andrea Hepuapo‘okela, and Inada, Megan.
2020
. “
Prejudice, Stigma, and Oppression on the Behavioral Health of Native Hawaiians and Pacific Islanders
.” In
Prejudice, Stigma, Privilege, and Oppression: A Behavioral Health Handbook
, edited by T. Benuto, Lorraine, Duckworth, Melanie P., Matsuda, Akihiko, and O'Donohue, William,
107
34
.
New York
:
Springer
.
Kana'iaupuni, Shawn Malia.
2011
. “
Lots of Aloha, Little Data: Data and Research on Native Hawaiian and Pacific Islanders
.”
AAPI Nexus: Policy, Practice, and Community
9
, nos.
1–2
:
207
11
.
Kidder, William C.
2013
. “
Misshaping the River: Proposition 209 and Lessons for the Fisher Case
.”
Journal of College and University Law
39
, no.
1
:
53
127
.
Lin, Rong-GongII, Money, Luke, and Shalby, Colleen.
2021
.
“California Is Reserving 40% of COVID-19 Vaccine for the Neediest
.
Who Will Get It?” Los Angeles Times
,
March
4
. www.latimes.com/california/story/2021-03-04/california-shifts-covid-vaccine-rollout-for-neediest-groups.
Maizlish, Neil, Delaney, Tracy, Dowling, Helen, Chapman, Derek A., Sabo, Roy, Woolf, Steven, Orndahl, Christine, Hill, Latoya, and Snellings, Lauren.
2019
. “
California Healthy Places Index: Frames Matter
.”
Public Health Reports
134
, no.
4
:
354
62
. doi.org/10.1177/0033354919849882.
McElfish, Pearl Anna, Hallgren, Emily, and Yamada, Seiji.
2015
. “
Effect of US Health Policies on Health Care Access for Marshallese Migrants
.”
American Journal of Public Health
105
, no.
4
:
637
43
. doi.org/10.2105/AJPH.2014.302452.
Morey, Brittany N.
2014
. “
Environmental Justice for Native Hawaiians and Pacific Islanders in Los Angeles County
.”
Environmental Justice
7
, no.
1
:
9
17
.
Morey, Brittany N., Tulua, ‘Alisi, Tanjasiri, Sora Park, Subica, Andrew M., Kaholokula, Joseph Keawe‘aimoku, Penaia, Corina, Thomas, Karla, et al.
2020
. “
Structural Racism and Its Effects on Native Hawaiians and Pacific Islanders in the United States: Issues of Health Equity, Census Undercounting, and Voter Disenfranchisement
.”
AAPI Nexus: Policy, Practice, and Community
,
October
24
. www.aapinexus.org/2020/11/24/structural-racism-and-its-effects-on-native-hawaiians-and-pacific-islanders.
Narcisse, Marie-R., Dobbs, Page, Long, Christopher R., Purvis, Rachel S., Kimminau, Kim S., and McElfish, Pearl A.
2020
. “
Electronic Cigarette Use and Psychological Distress in the Native Hawaiian and Pacific Islander Adults Compared with Other Racial/Ethnic Groups: Data from the National Health Interview Survey, 2014
.”
Journal of Community Psychology
48
, no.
2
:
225
36
. doi.org/10.1002/jcop.22248.
National Center for Health Statistics
.
2014
. “
Native Hawaiian and Pacific Islander (NHPI) National Health Interview Survey
.” Centers for Disease Control and Prevention. www.cdc.gov/nchs/nhis/nhpi.html (accessed
October
10
,
2021
).
Obermeyer, Ziad, Powers, Brian, Vogeli, Christine, and Mullainathan, Sendhil.
2019
. “
Dissecting Racial Bias in an Algorithm Used to Manage the Health of Populations
.”
Science
366
, no.
6464
:
447
53
. doi.org/10.1126/science.aax2342.
Office of Hawaiian Affairs
.
1982
. “
Office of Hawaiian Affair's Goals, Objectives, and Policies
.” In
OHA Master Plan
, n.p.
Honolulu
:
Office of Hawaiian Affairs
.
Okamoto, Dina, and Mora, G. Christina.
2014
. “
Panethnicity
.”
Annual Review of Sociology
40
, no.
1
:
219
39
.
OMB (Office of Management and Budget)
.
1997
. “
Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity
.”
Federal Register
,
October
30
. www.federalregister.gov/documents/1997/10/30/97-28653/revisions-to-the-standards-for-the-classification-of-federal-data-on-race-and-ethnicity.
Ong, Paul.
2009
. “
Trouble in Paradise: The Economic Marginalization of Native Hawaiians
.” In
Wealth Accumulation and Communities of Color in the United States: Current Issues
, edited by Gordon Nembhard, Jessica and Chiteji, Ngina,
155
72
.
Ann Arbor
:
University of Michigan Press
.
Panapasa, Sela V., Crabbe, Kamana‘opono M., and Kaholokula, Joseph Keawe‘aimoku.
2011
. “
Efficacy of Federal Data: Revised Office of Management and Budget Standard for Native Hawaiian and Other Pacific Islanders Examined
.”
AAPI Nexus
9
, nos.
1–2
:
212
20
.
Pastor, Manuel, and Morello-Frosch, Rachel.
2014
. “
Integrating Public Health and Community Development to Tackle Neighborhood Distress and Promote Well-Being
.”
Health Affairs
33
, no.
11
:
1890
96
. doi.org/10.1377/hlthaff.2014.0640.
Ponce, Ninez A., Shimkhada, Riti, and Tulua, ‘Alisi.
2021
. “
Disaggregating California's COVID-19 Data for Native Hawaiians and Pacific Islanders and Asians
.” UCLA Center for Health Policy Research,
May
26
. healthpolicy.ucla.edu/publications/search/pages/detail.aspx?PubID=2135.
Public Health Alliance of Southern California
.
2021
. “
Data and Reports
.” healthyplacesindex.org/data-reports/ (accessed
October
19
,
2021
).
Reskin, Barbara.
2012
. “
The Race Discrimination System
.”
Annual Review of Sociology
38
, no.
1
:
17
35
. doi.org/10.1146/annurev-soc-071811-145508.
Samoa, Raynald, Kaholokula, Joseph Keawe‘aimoku, Penaia, Corina, Tupai-Firestone, Ridvan, Fa‘amoe-Timoteo, Elena, Leaelan, Melisa, and Aitaoto, Nia.
2020
. “
COVID-19 and the State of Health of Pacific Islanders in the United States
.”
AAPI Nexus: Policy, Practice, and Community
,
September
24
. www.aapinexus.org/2020/09/24/article-covid-19-and-the-state-of-health-of-pacific-islanders-in-the-united-states/.
Subica, Andrew M., Agarwal, Neha, Sullivan, J. Greer, and Link, Bruce G.
2017
. “
Obesity and Associated Health Disparities among Understudied Multiracial, Pacific Islander, and American Indian Adults
.”
Obesity
25
, no.
12
:
2128
36
. doi.org/10.1002/oby.21954.
Taualii, Maile, Quenga, Joey, Samoa, Raynald, Samanani, Salim, and Dover, Doug.
2011
. “
Liberating Data: Accessing Native Hawaiian and Other Pacific Islander Data from National Data Sets
.”
AAPI Nexus
9
, nos.
1–2
:
249
55
.
Tuck, Eve, and Yang, K. Wayne.
2012
. “
Decolonization Is Not a Metaphor
.”
Decolonization: Indigeneity, Education, and Society
1
, no.
1
:
1
40
.
UCLA Center for Health Policy Research
.
2021
. “
NHPI COVID-19 Data Policy Lab Dashboard
.” UCLA Fielding School of Public Health. healthpolicy.ucla.edu/health-profiles/Pages/NHPI-COVID-19-Dashboard.aspx (accessed
October
19
,
2021
).
US Census Bureau
.
2020
. “
American Community Survey 2015–2019 5-Year Data Release
.”
December
10
. www.census.gov/newsroom/press-kits/2020/acs-5-year.html.
Zou, James, and Schiebinger, Londa.
2018
. “
AI Can Be Sexist and Racist—It's Time to Make It Fair
.”
Nature
,
July
18
. www.nature.com/articles/d41586-018-05707-8.