The popular perception of Asian Americans is that they are “doing well” in terms of income, education, and health.

In fact, Asian Americans and Pacific Islanders (AAPI) are a highly diverse group representing more than 30 countries of origin and many ethnicities, who certainly do experience social problems, especially at the bottom of the economic pyramid. Yet those inequities often get lost in the data, creating a harmful misperception that AAPIs don’t need help.

In the U.S., Asians have displaced Black Americans as the racial group with the greatest income inequality, according to Pew Research in 2018. The gap between the poorest and richest Asians has widened dramatically in recent years. But the plight of vulnerable Asians—low-income, limited English, immigrants, elderly, and undocumented—usually remains invisible.

The Covid-19 pandemic has especially hurt lower-income AAPIs, who suffer from high unemployment, high Covid-19 death rates, and other disparities. By last summer, 83% of Asian Americans in California with a high school education or less had filed unemployment claims, compared with 37% for the general California labor force with the same level of education, according to a UCLA analysis.

Vulnerable Asians also bear the brunt of the recent surge in anti-Asian racism and violence, because many work in service industries such as restaurants and hospitality, can’t shelter at home, and are more likely to take public transportation. This spring a recent immigrant from China who was collecting cans in New York City was beaten into a coma. The man had lost his job at a restaurant when it closed down during the pandemic. And near Atlanta, a gunman murdered Asian women spa workers on March 16. Three of them were in their 70s and 60s.

There have been many other tragic attacks, as well as more than 9,000 incidents of anti-Asian harassment and racism during the pandemic, according to an Aug. 12 report by Stop AAPI Hate, a nonprofit coalition. And on Aug. 12, Governor Gavin Newsom said he would deploy California Highway Patrol police to Oakland, which has seen a spike in violent crime, in response to requests from Mayor Libby Schaaf and Oakland Chinatown community leaders.

But many of these struggles don’t get attention. One key to addressing them is getting better data about AAPIs. Social problems go unnoticed partly because high incomes of well-off Asians mask the challenges of the vulnerable when they’re averaged together. When data about AAPI are disaggregated, a very different picture emerges. For example, in New York City, 22% of Asian Americans live in poverty. In the U.S., more than 34% of Cambodian, Laotian, and Hmong students do not complete high school, compared with 13% of the general population. Chinese and South Asians had the highest Covid-19 death and hospitalization rates, respectively, in New York’s largest public hospital system from March 1 to May 31, 2020.

The experience of the Marshallese in northwest Arkansas shows what better data can accomplish. Last year, Covid ravaged this unlikely community of Pacific Islanders. Although they’re 2.5% of the region’s population, they accounted for 19% of Covid-19 cases and 38% of deaths. “At one point, we were going to funerals every weekend,” says Sheldon Riklon, a Marshallese physician and professor at University of Arkansas for Medical Sciences.

Arkansas is home to about 12,000 Marshallese, the largest population outside the Marshall Islands, where the U.S. tested nuclear weapons in the 1940s and 1950s. The nuclear tests devastated the environment and food supply, which led to long-lasting health problems. In Arkansas, many Marshallese work in poultry processing plants and are more likely than the general population to be low-income and uninsured. Pacific Islanders in the U.S. also have greater underlying health risks such as diabetes, hypertension, and asthma, which make them more vulnerable to Covid-19.

As Covid-19 spread last year, Arkansas’s health department and local clinics raised alarms about the severe impact on Marshallese. The Centers for Disease Control dispatched a team to the state. The crisis among the Marshallese was recognized because health data on Pacific Islanders is separated from other Asian Americans, in keeping with 1997 federal mandates. And importantly, Arkansas’s local health department and clinics are attuned to the Marshallese community and collect data on them specifically.

Data disaggregation “played a huge role. We need to have that,” says Riklon.

Special Covid-19 initiatives focused on the Marshallese were launched in northwest Arkansas. Information about Covid-19 in Marshallese language was conveyed on Facebook and local Marshallese radio and newspapers. Trusted messengers such as church pastors, the Marshallese Consulate in Springdale, Ark., and community nonprofits urged testing and prevention. The efforts helped: By February 2021, Covid-19 cases among the Marshallese dropped to 6.3%.

Poultry company Tyson Foods, which employs many Marshallese in Arkansas, provided Covid information in Marshallese and offered on-site vaccinations to all staff. Earlier this year, Gerald Zackios, Marshall Islands’ ambassador to the U.S., even flew from Washington, D.C., to Arkansas and got vaccinated at a Tyson plant to encourage Marshallese employees to do the same.

“It’s a trust issue. They need to see us there,” says Riklon, who is also a doctor at a clinic in Springdale. “They need to know we got their backs and we support them.”

Mainstream media and policymakers have paid little attention to high death rates of AAPI Covid patients. That neglect continues a cycle of inequity that has only spiraled because of virulent anti-Asian racism during the pandemic, which likely hindered access to health care among vulnerable Asians, and might explain the outsize rates of Covid-19 deaths and low testing among Asians. Decision makers and media also rely too much on national surveys that lack nuance. Data is not just big numbers; smaller, more focused surveys and input from communities are also valuable.

AAPIs are about 6% of the U.S. population and are the country’s fastest-growing racial group. A population of more than 20 million is expected to grow to 35 million by 2060. The Asian population in the U.S. grew 35.5% from 2010 to 2020, according to U.S. Census data released last week. And in San Francisco and New York, Asians are more than 36% and 14% of the population, respectively.

Big differences in income and education often depend on diverse immigration histories. Indian Americans have the highest incomes among Asian Americans, and 70% have college degrees or higher. Many came to the U.S. on professional H-1B visas, so by nature they have higher education levels, says Neil Ruiz, Pew Research’s associate director of race and ethnicity research. Southeast Asians who came to the U.S. as war refugees, such as Hmong and Cambodians, tend to have lower education and incomes. Nepalese and Burmese earn the least.

AAPIs also often go uncounted. Many states don’t collect Covid-19 data for Asians. Instead, they are often relegated to “Other” or “Unknown” categories. Researchers found that about 4,000 Asian patients were miscategorized as “Other” in New York City’s public hospital system last year. Only around 20 states provided Covid-19 data about Native Hawaiian Pacific Islanders (NHPI) last year in spite of federal mandates to separate their data.

In California, where Native Hawaiians and Pacific Islanders (NHPI) data is collected, NHPIs had the highest death rates of any racial and ethnic group, according to the NHPI Covid-19 Data Policy Lab at UCLA. Yet government health officials weren’t overly concerned with NHPI Covid cases, says Calvin Chang, co-founder of UCLA’s Native Hawaiian and Pacific Islander Data Policy Lab. Their response was “you guys should be happy, your numbers are really low,” says Chang.

And while absolute numbers of NHPI Covid deaths were low, the proportion was very high for a smaller population. Deaths of elders were especially traumatic because “cultural knowledge and history is potentially lost forever,” says Chang, who is Native Hawaiian. Each death “was like watching a wing of the Library of Alexandria burn.”

Filipinos accounted for 31% of Covid-related nurse deaths last year, although they’re just 4% of nurses in the U.S., according to National Nurses United. And Vietnamese in Northern California’s Santa Clara County had the highest Covid-19 case rates compared with other AAPIs. Vulnerable AAPIs tend to work in service industry jobs and live in multigenerational homes that increase Covid exposure.

Doing research in Asian languages is vital to collecting better data. Most large surveys are conducted in English and Spanish, usually online or by phone. But Asians with limited English, which is linked to lower education and income, are more vulnerable to poverty and health problems. Surveys in English tend to capture experiences of Asian Americans with higher levels of education and income. “That just serves to reinforce model minority myths rather than surfacing the challenges that exist within our own communities,” says Howard Shih, director of research and policy at advocacy group Asian American Federation.

One model is the California Health Interview Survey (CHIS), launched in 2001 and conducted by UCLA’s Center for Health Policy Research. It “is the only population-based data that I trust,” says Tung Nguyen, professor of medicine at University of California San Francisco. “Anything else I don’t bother reading. It’s garbage in, garbage out.”

In addition to English and Spanish, the CHIS, the nation’s largest statewide health survey, is conducted by phone or online in several Asian languages: Chinese (Mandarin and Cantonese), Korean, Vietnamese, and Tagalog.

This sheds light on urgent trends. For example, the CHIS shows big health disparities among all Filipinos, not just health workers, compared with the general population. And Ninez Ponce, director of the UCLA Center for Health Policy Research, points out that Asian health-care workers overall have the highest Covid-19 case fatality rate—deaths per diagnosed cases—in California, according to restricted data from the state’s Department of Public Health.

While there are many Asian ethnicities, the six largest subgroups account for 85% of all Asian Americans. Surveys in the main Asian languages would reach a large swath. It is more expensive to do surveys in Asian languages. But cost is “a miniscule investment compared to the rest of our budget,” says Ponce.

Nguyen says that Asian language surveys are “not as expensive or hard for organizations that know how to work with Asian American communities.”

And working with trusted community organizations also helps reach vulnerable Asians, especially if questionnaires are about sensitive issues such as racism or mental health. “It’s not just, ‘Please fill out these surveys’. But if a group is helping someone get food stamps or taking care of their children, they will do it,” says Quyen Dinh, executive director of the advocacy group Southeast Asia Resource Action Center (SEARAC).

She notes that for a recent survey of elderly Cambodians in California, community groups did three rounds of edits to simplify the language and then verbally helped seniors with questions. “A paper form is not enough,” she says.

Media and policymakers usually favor large national surveys and studies from big institutions even though they can be skewed, incomplete. or lack nuance. Under-resourced grassroots organizations must then work harder to get attention for communities in need.

“It was a strategy to work with researchers at White institutions. That’s how you got people to take your issue seriously,” says Giles Li, former head of a Boston nonprofit that works with Asians and immigrants. “We never had to confirm that we knew what we were talking about. But we understood the rules of that game.”

Notions of what constitutes valuable data must be expanded beyond big numbers and famous institutions. Community-based surveys and smaller reports are also insightful and should complement larger data sets.

Input from communities brings statistics to life. For example, a 2018 report from SEARAC and other Southeast Asian nonprofits found that half of AAPI youth surveyed in five California cities had been bullied in school. The same survey had other powerful insights from 813 responses. More than 90% of Samoan, Cambodian, Hmong, and Laotian participants said their families are worried about having enough money to pay bills.

“It’s not about getting a gigantic large-scale survey,” says Dinh, “It’s about connecting with community members.” She adds that many youths wanted to share their experiences: “They want to be part of the solution,” she says.

Academic research has other challenges. Studies can take years to be published, if at all. And communities have historically experienced unethical behavior from researchers who “harvest what they need and leave,” says Raynald Samoa, lead of the National Pacific Islander Covid-19 Response Team.

Sometimes researchers don’t share results with communities. And studies can be more about advancing the careers of researchers than helping people.

“Research should improve the quality of life of people who are suffering,” says Carolyn Leung Rubin, assistant professor at Tufts University School of Medicine who has worked with Asian non-profits in Boston. “If there is no direct community benefit, we are just using them for our career development.”

In Arkansas, working with the Marshallese was critical to helping and learning from them. “We should be of the mindset that they are experts in their culture, customs, history, and they will teach us how to work with them,” says Riklon.

Nguyen of UCSF puts it more starkly: “Unless we address these issues that make us invisible to other people, they won’t see us as human beings.”

Amy Yee is an award-winning journalist who has written for The New York Times, The Economist, Wall Street Journal, Washington Post, and National Geographic, among other publications.

Co-published with Bloomberg News.