Why racial oversamples in polling make sense

Additional data uncovers differing perspectives on policy

A RECENT AND SUDDEN change to polling approaches has greatly expanded what we know about public opinion here in Massachusetts. The rapid expansion of “oversampling” happened during the nationwide upheaval following the killing of George Floyd by a Minneapolis police officer. In essence, oversampling means collecting extra respondents from a relatively smaller group to ensure their views can be examined separately. Rather than quickly fading, it has continued unabated over the last year. This has given us a much better understanding of public opinion among black, Latino, and Asian residents than we have ever had before.

We at the MassINC Polling Group do a pretty large share of the public polling in Massachusetts, so we have a good sense of how things are changing here. In the past, we did a handful of oversamples by race over the course of our 11-year history. Today, the large majority of our polls include oversamples by race. This extends far beyond issues that have traditionally been viewed as “issues of race.” For equity to be central to our state’s policies, we must understand the realities directly from our communities of color. Representative data is a critical place to start.

The population of Massachusetts is 71 percent white, 9 percent black, 12 percent Latino, and 7 percent Asian, with the remainder being other races. This means a poll of 500 or 600 respondents might only include a few dozen interviews from groups other than white residents if the poll is only representative of the majority of the state’s population. Oversamples will add a few hundred to these few dozen interviews, radically expanding what we can learn about the views of people of color across the state. The ability to access representative data enables us to disaggregate within racial groups as well, allowing a much more nuanced picture to emerge.

Here we have to note this is not primarily our doing, but rather the new interest and commitment from poll sponsors to gathering diverse viewpoints and examining them in detail. This is not a free approach, but involves conducting (and paying for) hundreds of additional interviews to reach adequate oversamples. The massive expansion in the practice emerges from a growing embrace of the fact that every political and policy issue is closely related to race. State budgets are related to race, as is education, transportation, policing, income inequality, climate change, housing, and more.

Since the start of the pandemic, the MassINC Polling Group has surveyed Massachusetts residents on issues ranging from education, telehealth, and transportation to the state budget and small business considerations, among many others. Zeroing in on a few highlights shows how details of public opinion on these policy questions have remained hidden from view until now.

On the state budget, for instance, 60 percent or more white voters identified just two policy ideas as “very important.” Zooming in on black and Latino voters shows a much longer priority list. Black voters call 12 different policy areas “high priority,” while Latino voters identify 10. These two groups are also more than twice likely to prioritize “investing in communities of color” and “increasing state contracts with women and minority owned businesses.” To put it plainly, black and Latino voters see a much longer to-do list for state leaders than white voters do, which we would not see just from the overall numbers.

Other polling documented the devastating impact the pandemic has had specifically on lower income black and Latino residents and specifically women of color. Surveys of small business owners found those companies founded and owned by women and people of color had suffered far deeper impacts. This exploration helped shape the state’s small business assistance programs.

Oversamples also allowed us to document the massive gaps in educational experiences during the pandemic. Our K-12 parent poll last year found major differences in technology and internet access by race and ethnicity, resulting in large gaps in students’ ability to participate in online school. This helped spur a major focus on technology access over the summer. Oversampling also showed the massive differences by race between in-person, hybrid, and remote schooling, as well as the reasons parents of color remain skeptical of a quick return to the classroom.

Another poll outlined the shape of vaccine hesitancy and the specifics by race and ethnicity months before vaccines even rolled out. Latino residents were the most hesitant even then, which has continued as millions of residents queue up for their shots. This has also let state leaders take direct aim at the most hesitant groups, putting Massachusetts at the top of the list on several vaccine equity measures. More recent polling of parents has found large gaps within race groupings when it comes to vaccinating children. While Latino residents with degrees are comparatively eager to get their children vaccinated, those without degrees are far less so. Such granularity is another benefit of larger samples by race.

Polling is just one part of the political process. When it comes to political leadership, we have a long way to go to ensure voices are not only heard but equally represented. Equity across policy issues is even further away, and would require systemic change to not just which policies are pursued but how policy is formed. Viewing public opinion data that is representative of more communities across the state is a step toward equity, and one worth pausing to celebrate.

Zayna Basma is a research associate and Steve Koczela is the president of the MassINC Polling Group, which is a for-profit subsidiary of MassINC, the parent company of CommonWealth.