[ad_1]
The week after eight people were killed in a string of mass shootings at Asian spas around Atlanta, I spoke to a group of college students about the year in data journalism, from Covid-19 trackers to US election results pages.
I told them that when talking about data visualisation, some of the most powerful graphics allow audiences to see themselves. But then I thought about how many times I made a chart of US racial data that either excluded Asians or relegated them to the “other†category.
“We’re missing from the history books; we’re also missing from the budget books,†Grace Meng, a New York congresswoman, said during a recent rally in Manhattan’s Chinatown, referring to America’s often forgotten history of anti-Asian violence and the disproportionately low amounts of philanthropic dollars going towards Asian-American and Pacific Islander (AAPI) causes. “We have been invisible for way too long.â€
In data, AAPI communities are also overlooked. Some of this is out of statistical necessity. US public opinion surveys typically have about 1,000 respondents. Since Asian Americans make up 5.9 per cent of the population, a nationally representative poll would include just 59 of them — too small a sample size from which to draw inferences. Breakdowns by gender or age would be even smaller.
But during presidential elections, this lack of representation in national surveys means that the voices of Asian voters, though they are the fastest-growing group in battleground states such as Georgia and Texas, are disadvantaged in terms of news coverage.
Asian Americans are extraordinarily diverse, consisting of more than 48 different ethnicities. Conducting a survey with adequate sample sizes for each of the major subgroups with language support for respondents could cost upwards of $1m, says Karthick Ramakrishnan, founder of AAPI Data. Contrast that with the $50,000 he says it costs for a nationally representative survey of white, black and Latino populations and it’s easy to see why Asians are so often grouped as “otherâ€.
As a data reporter, I understand the need for this “othering†when there is no data. But I can’t help but wonder if Asian-Americans’ relatively low voter registration and turnout rates have anything to do with their invisibility in political and public opinion data.Â
An even more insidious form of erasure happens when “Asian American†is treated as a monolithic identity, with differences between ethnic groups averaged away. Aggregate statistics can mask disparities and perpetuate the “model minority†stereotype. Asian Americans have the highest median income of any racial or ethnic group in the US, but also the highest gap between the highest and lowest earners.
Aggregate data can also hide health disparities. After community advocacy, the US Census Bureau began counting Native Hawaiian and Pacific Islanders and Asian Americans separately in 2000, but most states still combine them in their Covid-19 reporting. As a result, the high coronavirus death rates among the relatively smaller Native Hawaiian and Pacific Islander populations disappear in the overall statistics.
Identity is complex. Terms such as “AAPI†or “Asian American†can be too broad in some ways, but powerful in others, such as in response to anti-Asian attacks. At the end of my talk, I cited police-reported hate crimes as an example of when data might be missing or unreliable, and discussed how we can address these problems.
We should of course push for better data: more accurate reporting of hate crimes, more details on race and ethnicity, more funding for large-scale studies like the National Asian American Survey (which Ramakrishnan says was not conducted in 2020 due to lack of funds).
But we should also recognise the limitations of data analysis. It can be a blunt tool that flattens lived experiences into categories to make them easier to study. Statistics like hate crimes aren’t, and won’t ever be, precisely measured. Improving them must go hand-in-hand with telling the stories of the people behind the numbers — like the best examples of data journalism.
[ad_2]
Source link
