Combating abysmally low response rates for political polling

One pollster describes the difficulty today in reaching potential voters:

Photo by Breakingpic on Pexels.com

As the years drifted by, it took more and more voters per cluster for us to get a single voter to agree to an interview. Between 1984 and 1989, when caller ID was rolled out, more voters began to ignore our calls. The advent of answering machines and then voicemail further reduced responses. Voters screen their calls more aggressively, so cooperation with pollsters has steadily declined year-by-year. Whereas once I could extract one complete interview from five voters, it can now take calls to as many as 100 voters to complete a single interview, even more in some segments of the electorate…

I offer my own experience from Florida in the 2020 election to illustrate the problem. I conducted tracking polls in the weeks leading up to the presidential election. To complete 1,510 interviews over several weeks, we had to call 136,688 voters. In hard-to-interview Florida, only 1 in 90-odd voters would speak with our interviewers. Most calls to voters went unanswered or rolled over to answering machines or voicemail, never to be interviewed despite multiple attempts.

The final wave of polling, conducted Oct. 25-27 to complete 500 interviews, was the worst for cooperation. We could finish interviews with only four-tenths of one percent from our pool of potential respondents. As a result, this supposed “random sample survey” seemingly yielded, as did most all Florida polls, lower support for President Trump than he earned on Election Day.

After the election, I noted wide variations in completion rates across different categories of voters, but nearly all were still too low for any actual randomness to be assumed or implied.

This is a basic Research Methods class issue: if you cannot collect a good sample, you are going to have a hard time reflecting reality for the population.

Here is the part I understand less. This is not a new issue. As noted above, response rates have been falling for decades. Part of it is new technology. Some of it involves new behavior, such as ignoring phone calls or distrust of political polling. The amount of polling and data collection that takes place now can lead to survey fatigue.

But, it is interesting that the techniques used to collect this data are roughly the same. Of course, it has moved from land lines to cell phones and perhaps even texting or recruited online pools of potential voters. The technology has changed some but the idea is similar in trying to reach out to a broad set of people and hope a representative enough sample responds.

Perhaps it is time for new techniques. The old ones have some advantages including the ability to relatively quickly reach a large number of people and researchers and consultants are used to these techniques. And I do not have the answers for what might work better. Researchers embedded in different communities who could collect data over time? Finding public spaces frequented by diverse populations and approaching people there? Working more closely with bellwhether or representative places or populations to track what is going on there?

Even with these low response rates, polling can still tell us something. It is not as bad as picking randomly or flipping a coin. Yet, it is not accurate enough in recent years. If researchers want to collect valid and reliable polling data in the future, new approaches may be in order.

From Brookings: Biden wins through suburban voters

William Frey looks at presidential voting by geography and concludes that suburban voters gave Biden his victory:

In the 2016 election, rural and nonmetropolitan America gave Donald Trump enough of a margin to beat Hillary Clinton in seven key states. Ahead of the 2020 election, Republicans worried that Trump would lose his rural edge, in light of reduced support there in the 2018 midterm elections. But this was not the case. Instead, Trump’s loss to Joe Biden was due mostly to voters in large metropolitan suburbs, especially in important battleground states…

However, large suburban areas in 2020 registered a net Democratic advantage for the first time since Barack Obama’s victory in 2008. This is significant because more voters reside there than in the other three categories. In terms of aggregate votes in these large suburban counties, there was a shift from a 1.2 million vote advantage for Trump in 2016 to (at last count) a 613,000 vote advantage for Biden—a nearly 2 million vote flip. In addition, Biden benefitted from more modest Republican margins in small metropolitan areas. These advantages for the President-elect were even greater in key battleground states…

The three northern battleground states of Michigan, Pennsylvania, and Wisconsin—which flipped to Trump in 2016—again entered the Democratic fold in 2020. Here, even more than in the national analysis, the 2016 to 2020 suburban shifts to either greater Democratic or smaller Republican support were instrumental in Biden’s victory…

Suburban voting patterns also made a difference in the Sun Belt, especially in large southern states where suburbanization has been rampant. The focus here is on two such states: Georgia, where Biden is ahead and a recount has been announced; and Texas, which Trump won, but where urban and suburban voting patterns closed the longtime Republican-Democratic gap.

This is the most detailed analysis I have seen thus far. The predictions were right: the 2020 presidential election depended on the suburbs!

It also brings several other features of American and political life into relief:

  1. Joe Biden was nominated in part because of his electability. In the long run, his electability in one particular kind of place was particularly important: suburbs. Trump, to some degree, knew this but his approach was more combative and did not have the appeal he hoped.
  2. While political analysis suggests middle suburbs are battleground areas, I wonder if this signals that these suburbs are also in the middle of all sorts of other trends including demographic changes, cultural tastes, and suburban inequality. To build on earlier posts, perhaps finding middle America right now involves going to a Walmart in a middle suburb or an emergency room in a middle suburb.
  3. Many people have discussed the electoral college in recent years. Here is a crazier proposal based on more recent trends: instead of the electoral college by states, how about an electoral college by cities, suburbs, and rural areas? With concerns on either side that cities or rural areas are controlling political outcomes, could there be some way to weight the results such that all three geographies could influence the outcome? Grouping votes by states obliterates any distinctions between places.

Suburban shift: 49-45 for Trump in 2016, 51-48 in 2020 for Biden

In 2020, the tight suburban vote swung opposite of 2016 and slightly in favor of Joe Biden:

Photo by Element5 Digital on Pexels.com

While there was much attention paid to late-counted mail ballots in urban areas that put Biden over the top in states such as Michigan, Pennsylvania and Wisconsin, the bigger shift was in the suburbs. In fact, there’s evidence that Trump actually improved his lot slightly in urban areas, which made the suburbs crucial for Biden.

In 2016 exit polls, Trump won the suburbs by four points, 49 to 45. This time, Biden won them 51 to 48 — a seven-point shift in the margin. Biden also joins Obama as the only Democratic presidential candidate to carry the suburbs since 1992, if the exit polls don’t shift from now. The New York Times has a great visualization of the shift from 2016.

Democrats’ suburban edge was also slightly bigger than in the 2018 election in which they won the House, when those areas split about evenly. And given that the suburbs account for about half the votes these days — and growing — Democrats will want to keep that going.

The only problem for them — and it’s a big split between the presidential race and down-ballot — was that this performance didn’t stretch to the conservative-leaning suburban seats Democrats were hoping to take from Republicans. So Democrats need to ask themselves whether this is really about the new reality or whether it was just about Trump, who underperformed his party in many ways, but particularly in these areas.

Two takeaways from these four paragraphs:

  1. The suburbs will continue to be a battleground in presidential, congressional, state, and local races.
  2. There appears to be a shift in 2020 compared to 2016 but this was a relatively minor shift. What happens in subsequent elections will help indicate whether a suburban shift is longer-lasting.

In other words, expect more of the same in 2022 and 2024 with campaigns and pundits focused on suburban locations in the swing states of the 2016 and 2020 elections as well as some other possible locations (like Texas). And county level analysis will be helpful but more fine-grained study will reveal where some of the important divides or pockets of different voters are in sprawling geographic areas.

Getting the categories right in continued urban/rural-plus-suburbs-and-exurbs-in-the-middle divide

Numerous outlets have commented on the continued presence of an urban/rural divide in 2020 voting. Here is another example:

Photo by Pixabay on Pexels.com

Rather than flipping more Obama-Trump counties, Biden instead exceeded previous Democratic win margins in Wisconsin’s two biggest cities, Milwaukee and Madison.

That pattern extended to Michigan and other battleground states, with Biden building upon Democrats’ dominance in urban and suburban jurisdictions but Trump leaving most of exurban and rural America awash in red.

The urban-rural divide illustrates the pronounced polarization evident in preliminary 2020 election results. The split underscores fundamental disagreements among Americans about how to control the coronavirus pandemic or whether to even try; how to revitalize the economy and restore jobs; how to combat climate change or whether it is an emergency at all; and the roles of morality, empathy and the rule of law in the body politic.

Four thoughts in reaction to this.

  1. The urban/rural divide is described in an interesting way above: it is cities and suburbs for Democrats and exurbs and rural areas for Republicans. This matches the patterns of this and recent elections. However, is separating the suburbs and exurbs worthwhile? Here is where county level analysis may not be fine-grained enough to see the patterns. Another way to put it might be this: there is a gradient in voting by party as the distance from the big city increases. Does it shade over to Republicans only in exurbs – which are suburbs on the outer edges? Is the 50/50 split a little before exurbs? A concentric circles approach could help though there still could be pockets that break with the overall pattern.
  2. Suburbs might just be too broad of a term to be useful in such analysis. The exurb/suburb split it one way to put it. Might it help to also think of different types of suburbs (wealthier bedroom communities, ethnoburbs/majority-minority communities, working-class suburbs, industrial suburbs, etc.)?
  3. Explaining the differences as urban/rural has a nice short ring to it and it fits the data. Introducing more categories in the middle is interesting to campaigns, pundits, and researchers but is harder to quickly describe. Perhaps the urban/suburban versus exurban/rural divide?
  4. Is the urban/rural divide one of the most fundamental aspects of polarization? Or, is it a symptom? In this story, the divide leads off the discussion of polarization on a number of fronts. But, what leads to these spatial patterns in the first place? While the geography is helpful to think about, are the real issues behind the urban/rural divide about race/ethnicity and class? Given residential segregation patterns in the United States, using the spatial patterns as an explanation covers up a lot of important social forces that led to those patterns in the first place.

First exit poll I have seen on suburban voters

In reporting exit polls yesterday, the New York Times included votes for Biden or Trump by location:

https://www.nytimes.com/interactive/2020/11/03/us/elections/exit-polls-president.html

The 2020 election, according to many pundits, was going to be decided by suburban voters.

The data above appears to follow the patterns from the last few elections. Cities go for Democrats. Rural areas and exurbs go for Republicans. The suburbs are contested ground. The splits are bigger in the table above: 9% advantage for Trump in small city or rural areas, 23% advantage for Biden in cities. And the suburbs are close with just a 3% gap.

Going even further, the suburbs of the true battleground states may be home to the truly important voters. In this election, the residents outside Atlanta, Charlotte, Raleigh-Durham, Cleveland, Columbus, Cincinnati, Pittsburgh, Philadelphia, Detroit, Grand Rapids, Milwaukee, Minneapolis, Phoenix and more might have just decided the election. That national 3% difference might not seem like much but it could matter a lot when state races are tight.

I am sure there will be more fine-grained analysis of the suburban vote by itself as well in comparisons to other demographic factors that influenced vote totals. But, this early data seems to suggest suburban voters will continue to be important in coming elections.

Trying to fit all the election results on one television screen

I watched briefly a number of election night broadcasts last night. One conclusion I came to: there is way too much data to fit on a television screen. And if you want more of the data, you need the Internet, not television.

The different broadcasts tried similar variations: flipping back and forth between a set of anchors and pundits at desks and analysts at a smart board showing election results from different states and locations. They have done this for enough election nights that the process is pretty established.

While they do this, there is often a lot of data on the screen. This could include: a map of the United States with states shaded; a chryon at the bottom with scrolling news; another panel at the bottom flipping through results from different races; and people talking, sometimes in connection to the data on the screen and sometimes. If the analyst at the smart board is on the screen, there is another set of maps to consider.

CNN broadcast, November 4, 2020

This is a lot to take in and it might not be enough. The broadcasts try to balance all of the levels of government – from the presidential race to congressional districts – and are flipping back and forth. I appreciated seeing the more simple approach of PBS which went with a lot less data on the screen, bigger images of the talking heads, and simple summary graphics of the winners.

But, if you want the data, the television broadcast does not cut it. Numerous websites offered single pages where one could monitor all of the major races in real-time. Want to keep up on both local and national races? Have two pages open. Want reaction? Add social media in a third window. Use multiple Internet-connected devices including smartphones, tablets, and computers (and maybe Internet-enabled televisions).

Furthermore, web pages give users more control over the data they are seeing. Take the final 2020 election forecast from FiveThirtyEight:

On one page, readers could see multiple presentations of data plus explanations. Want to scroll through in 10 seconds and see the headlines? Fine. Want to spend 5 minutes analyzing the various graphics? That works. Want to click on all the links for the metholodogy and commentary? A reader could do that too.

The one big advantage television offers is that it offers commentary and faces in real-time plus the potential for live coverage from the scene (such as images of gatherings for candidates) and feeling like the viewer is present when major announcements are made. The Internet has approximations of this – lively social media accounts, live blogs – but it is not the same feeling. (Of course, when you have more than ten live election night broadcasts available on your television, the audience will be pretty split there as well.) Elections are not just about data for many; they also include emotions, presence, and the potential for important memories.

Given these differences in media, I did what I am guessing many did last night: I consumed both television and Internet/social media coverage. Neither are perfect for the task. I had to go to sleep eventually. And whoever can figure out how to combine the best elements of both for election nights may do very well for themselves.

Changes in “countrypolitan,” exurban counties more than just political

The analysis of the 2020 election includes many analyses of suburban voters. But, there is more at stake here than just voting patterns as this look at Union County, North Carolina suggests:

Google Maps – Union County, North Carolina

Union County is what one scholar terms a “countrypolitan” place: Under federal government designations, it lies within a metropolitan area, but it also has a strong rural and agricultural history. For the most part, it doesn’t look like a cookie-cutter suburb, nor is it impoverished. In fact, Union is North Carolina’s wealthiest county, according to the Census Bureau. There are places like it around the United States. They are distinct from rural areas, which are mostly Republican, and cities, which are heavily Democratic; many voters in these places are neither die-hard Trump fans nor urban liberals. That makes them pivotal counties, in 2020 and in the future.

Everyone agrees that Union County is changing. The question is how it’s changing, and how fast. There’s no doubt that Republicans will carry the county up and down the ticket this year—Carter, ensconced at East Frank, was the last Democratic presidential candidate to win here, back in 1980—but the GOP’s overall success in the state will hinge largely on how big a margin it is able to run up in exurban counties such as this one. Democrats’ control of inner-ring suburbs continues to strengthen, and the future of the Republican Party nationally depends on keeping firm control over places like Union County…

The reason Union County is changing is simple math. When Helms was born, about 4,000 people lived in Monroe. Today, nearly 36,000 do. Since 1990, the overall county population has almost tripled, from about 84,000 to roughly 240,000. As I traveled around the county, I began to notice something peculiar: Virtually everyone I talked with was a transplant. Some of them had moved only recently, and others had been around for 10 years, or 20…

“Union County when Jesse was in the Senate was a very rural county,” Wrenn says. “Now it’s got a big chunk of suburban in it. If the long-term trends continue, the Republicans are going to have to find a way to compete in the suburbs. It can be done, but you just have to change your whole way of thinking.”

The temptation in such stories about suburban voters is to look at counties and communities and just see political change. And it sounds like Union County has had its share comparing before the Civil Rights Era, after, and today.

The basic explanation for this recent change is new residents. The population has grown and new residents, not as familiar with the ways of Union County, have moved in. What was once a small population with low density is now much larger and in bigger cities and towns.

But, is this all that has changed: new people moved in and they came with some new political views? I suspect there is more going on here that both contributes to the political change and also exists outside of it. Here are a few possible factors at play:

  1. The suburbs and the spread of metropolitan areas are not just about increasing populations and higher population densities. The suburbs come with a particular way of life. People who seek out such locations want single-family homes, middle-class opportunities and peace for their family, responsive local government, and the ability to live in a place they chose with people who look more like them. This is different than a more rural or working-class character in communities. This desire for the American suburbs does not easily line up with either political party on all the issues but it certainly is a different way of life.
  2. The decline of agriculture, particularly family-owned farms and opportunities, could be at play here. As farming becomes more difficult or less desirable for subsequent generations, the land can be sold off and be turned into subdivisions. This is a significant change in land use as well as who lives there: farmers and those connected to agriculture versus middle-class suburbanites.
  3. Connected to #2, the economic landscape has changed tremendously in the last half-century or so, moving away not just from agriculture but also manufacturing and moving more toward retail, services, and a knowledge economy. Union County and many other locations in the United States are still trying to adapt to these large shifts that affect employment, tax bases, and local businesses.
  4. Numerous local institutions have likely had to adjust in light of growing populations. Schools need more space for kids. Local governments need to provide more services (and they might now have larger tax bases to draw on) and local officials are addressing new issues. Established churches now compete with new congregations. In sum, the civic and social institutions that may have existed for decades in roughly the same form now need to adapt. This can present challenges in any community.

In sum, this is not just about politics. A shift toward a suburban lifestyle in Union County has many consequences and politics may just be one of them.

Missing any conversation about housing in the 2020 presidential campaign

The Democrat presidential primary for 2020 contained occasional talk of housing on the campaign trail or in debates. But, leading up to Election Day 2020, there has been little attention paid to housing. A few notes on this:

Photo by Element5 Digital on Pexels.com
  1. Some broader campaign themes could be related to housing: talking about the middle-class, taxes, providing opportunities for more Americans. But, these typically do not directly address housing even if it may be implied (such as being middle-class includes the ability to own a home, often in the suburbs).
  2. COVID-19 has had a direct effect on a number of areas. While there has been much said about jobs and the economy, the connection to housing is missing. For many, employment status directly affects housing status and there have been moratoriums regarding rent and evictions. Yet, the primary focus is on the direct means of making money (jobs, stock market, etc.) rather than what many hope their earnings translate into (a nice residence to own or rent in a good neighborhood). Additionally, the housing and real estate industry contribute to the broader economy but little has been said about activity here outside of some people are leaving cities.
  3. Even without the 2020 election and COVID-19, housing is related to a number of other key forces in American society: local schools, property values, building wealth, local neighbors and social networks, access to jobs, health care, amenities, and local services. Yet, it continues to merit little mention or go under the radar all together. Even as it is a difficult issue to address at the national level due to local complications and a complicated and controversial history in the United States, ignoring the issue does little.

Patterns in political yard signs

A new book by three political scientists look at how Americans have deployed and reacted to yard signs in recent election cycles:

We just started being puzzled about it. We did things like code the amount of traffic on a given street, and we thought maybe people on a street with high traffic would be more likely to put up signs. But you find out that those people wanted to let other people know where they stand- that it wasn’t just about catching the eye of passing traffic [to try to get out the vote for a candidate]. We found out that there’s a combination of expressive and communicative motives…

One of the things that was really clear from our studies is that signs are really important to people who display them. They’re emotionally invested in these dynamics and are more likely than people who don’t put up signs to say that it’s a good thing, or a reasonable thing, for neighborhoods to be doing.

I also think this is why we hear about these stories of theft and vandalism-people going to extremes around signs. At least seemingly, in news reports, it can accelerate fast, from people putting up signs to some kind of an altercation, a police report, a fight on the street. I think it’s because people view it as a real affront when someone messes with their expression of self…

You really notice, when you’re walking around, those places where signs are battling one another. But when we did spatial analysis to look at the clustering of signs systematically, in a way that would cut through those strong anecdotal impressions, we found that, really, there wasn’t much evidence of the intermingling of signs-the famous Sign Wars, where there’s a Biden sign at one house and a Trump sign next to it. Really, it was more about like-minded clustering: pockets of Biden supporters signaling to one another, pockets of Trump supporters signaling to one another. More solidarity than outright conflict.

I appreciate the systematic approach for a phenomenon that lends itself to anecdotes. This is how social science can be really helpful: many people have experiences with or have seen yard signs but unless researchers approach the issue in a rigorous way, it is hard to know what exactly is going on.

For example, I regularly walk in two different places in my suburb and I have been keeping an eye on yard signs. At least in the areas I walk, the signs are primarily in favor of one party in the national election while local election signs are more varied. Furthermore, the number of people who have signs is still pretty limited even in a heated political climate. But, just based on my walks, I do not know if what I am seeing match my suburb as a whole let alone communities across the United States. And unless I interact in some way with the people with (and without) yard signs, I have little idea of what is motivating them.

I wonder how the behavior of putting out political yard signs relates to other political behavior. If a political yard sign is expressive, how much does this carry over to other parts of life? Are these the people who are most active in local political activity? Are they the most partisan? Are they the ones always bringing up politics at family gatherings or among friends?

I would also be curious to how this relates to social class and particular neighborhoods. Lawns, in some places, are sacred: they should be green, free of weeds and leaves. Property values are important in many places. Political signs might mess up particular aesthetics or introduce the idea of conflict when suburbanites just want to leave each other alone.

Population density as a factor in deciding suburban votes

An AP story highlights how “Democrats march deeper into suburbia” and discusses the role of density:

https://apnews.com/article/election-2020-virus-outbreak-phoenix-suburbs-health-care-reform-76dea954e5f5dc13b3ad65fd29ac678c

For decades, an area’s population per square mile has been a reliable indicator of its political tilt. Denser areas vote Democratic, less dense areas vote Republican. The correlation between density and voting has been getting stronger, as people began to sort themselves by ethnicity, education, personality, income and lifestyle.

The pattern is so reliable it can quantified, averaged and applied to most American cities. At around 800 households per square mile, the blue of Democratic areas starts to bleed into red Republican neighborhoods.

A purple ring — call it the flip zone — emerges through the suburbs…

In Dallas, the purple ring through the suburbs was 18.7 miles in 2016 out from city hall, at an average of 714 households per square mile. The border runs close to AT&T Stadium in Arlington, where the Dallas Cowboys play. Arlington is a so-called boomburb that morphed through new construction from a suburb to a city of 400,000.

A few thoughts on the potential role of density. First, an additional graphic (see below) works with the last paragraph cited above to draw concentric circles a city. This, however, suggests density is linear as one moves further out from the city. In general, this may be true but it would be interesting to see how pockets of higher density suburbs at different distances from the city affect these patterns.

https://apnews.com/article/election-2020-virus-outbreak-phoenix-suburbs-health-care-reform-76dea954e5f5dc13b3ad65fd29ac678c

Second, Is it density that predicts these outcomes or factors related to density? A third chart in the story looks at the population demographics at different densities and shows differences. Does the density come first or the population changes? The analysis here suggests a relationship or correlation but it is not clear whether this analysis accounts for other possible factors.

In the larger picture, what do Americans think about having these “flip zones” or middle suburbs be the current political battleground? For example, one current argument about getting rid of the electoral college suggests certain parts of the country should not a disproportionate sway over other more populous parts of the country. Right now, these middle suburbanites, particularly in swing states, have the influence and both parties want their votes. Are the interests of these suburban voters the interests of the entire country?