UMass Poll Director Tatishe Nteta on Public Polling, a Diverse Electorate, and Lessons from the 2016 Election [Video]
Conducting public opinion polls is not an easy job. People increasingly look to polls to predict the outcomes of elections and ballot initiatives. But when the results are not what they expected, it can feel like the polls misled them.
"There needs to be a recognition of the true diversity of our population, and that having a representative sample on things like age, gender, race, ethnicity is in fact imperative..."
“What we’re trying to do is to provide accurate reflection of where the public is when we poll.” says Tatishe Nteta, Director of UMass Poll. “Now, things can change; but the point of polling is to provide that reflection at that moment in time to the best of our ability.”
Given the polarized state of contemporary politics and the residue of suspicion left by the 2016 polls (which showed Hillary Clinton leading for much of the race), Nteta and his colleagues recognize that people will turn to and scrutinize polls to help them make sense of the elections ahead. In the run up to primary season,Nteta broke down for us how UMass Poll works, why it is important, and how it has learned and evolved from 2016.
This interview has been edited for clarity and brevity.
Tatishe, what exactly is UMass Poll?
UMass Poll is a polling organization that’s run out of the UMass political science department with my colleagues Ray La Raja and Jesse Rhodes. Our purpose is to assist with the civic engagement in the Commonwealth of Massachusetts, but New England in general as well. We think that public opinion polling is a net good; that having more information about how the citizens of the commonwealth -- of New England, of the nation -- think about politics and think about their own everyday lives is an important piece of information not just for citizens in these areas but also for lawmakers.
How has polling changed with a more diverse electorate?
In public opinion polling, the “opinion” is important but the “public” is also important. The first thing to note is, since the 1960s the US has become more democratic, more encompassing, more reflective. This is a result of legislation like the Voting Rights Act, Civil Rights Act, and Fair Housing Act. With a change in the racial-ethnic demographics in the US, we have as pollsters attempted to capture that increasing diversity. Previous to the 1990s, polls did not poll in Spanish, for instance; we’re doing a better job of capturing the largest racial minority in the US, being Latinx. Polls did not do a very good job contacting African-Americans but we’re doing a better job and there's much more attention focused on this particular subset of the country. Today we’re doing a much better job than we did even ten years ago.
And it’s not just about gender, race, and ethnicity, it’s also about class. It’s difficult to truly capture the attitudes of individuals who live in rural areas and are on the periphery economically. With the ability to poll folks on the Internet, we are dealing with some of those problems. Ninety percent of Americans are on the Internet and so Internet polling has led to an ability for us to deal with some of the problems that have plagued polling previously.
How do you conduct a poll?
We usually poll around elections, whether it’s before an election or after. What we’re interested in if we poll before the election is trying to get a sense of which candidates, which questions, which policies, are going to be supported by a majority of the Commonwealth’s residents. If we poll after an election, it’s about trying to explain what just happened and trying to provide a folks for a roadmap to the next election. Elections are an important factor in timing our polls.
In terms of questions, it’s about keeping a finger on the pulse of the region, of the state, of the nation, as to what the key and most important issues are. In designing our questions, we’re focused primarily on addressing what people care about, and thinking also if this is a tool lawmakers can use in crafting better policies, more engaging policies, more inclusive policies.
Speaking of elections, what made 2016 different? Did polls fail us in 2016?
No. The first thing to recognize is, if you look at the polls of the nation on average they had Clinton winning by three percentage points. Clinton won by two percentage points when you look at the popular vote. The problem people had was with polls in particular states: if you look at a state like Pennsylvania, Wisconsin, Michigan, or even Florida, a lot of the polls had Clinton leading and the argument is the polls got it wrong and in some ways that could be true; but in many of those polls, Clinton’s lead was in the margin of error. It was a unique election in that a non-trivial portion of voters made their decisions in the last week, and that non-trivial portion of voters overwhelmingly went for Trump -- that’s what we came to find.
Another issue is that the dynamics of the campaign were unique in that we had these monumental events which occurred at different periods of time. You had the Access Hollywood tape in November, you had former FBI Director Comey re-instigating an investigation which many thought was dead, you had all these events which literature has shown does affect voters’ calculations.
That being said, a lot of the state polls and even a lot of the national polls did make mistakes. One of the central mistakes they made was not weighting education. Highly educated individuals are more likely to take polls and those with less education are more wary of polls. There needs to be a recognition of the true diversity of our population, and that having a representative sample on things like age, gender, race, ethnicity is in fact imperative, but the role of education particularly in our politics is really important in understanding who turns out to vote and who they vote for.
I always say we should press the brakes on the notion that polls did not accurately predict what happened, but also to recognize the areas of improvement that polls should focus on and rectify going forward. Looking at 2018, they’re a lot more accurate; I think in 2020, we’ll have even more accurate polls.
What is UMass Poll working on now?
Well we have a presidential election in 2020. We have a Democratic Primary here in the State of Massachusetts in 2020. We’re trying to get a sense of where Democrats and Independents in the state are participating in the primary, where they’re leaning towards. We now have two candidates with Massachusetts ties that are in the Democratic Primary in Senator Elizabeth Warren and former governor Deval Patrick. We have a senator from a neighboring state in Bernie Sanders who has also done very well in this state previously in the 2016 campaign. There’s a New England story to be told here and we hope to be the ones telling that story as it pertains to the 2020 election. And of course the Senate primary between [Representative Joseph P.] Kennedy and [Senator Edward J.] Markey is going to be really interesting. It’s a generational story and we’ll see which candidate succeeds.
As the campaign develops and matures, we’ll have a better sense of where the wind is blowing as it pertains to those races. But what I do know is that it’s going to be an exciting year for polling.