Romney's Post-Debate Bounce: What’s clear and what’s uncertain

Printer-friendly versionSend to friend

By most every account, Republican nominee Mitt Romney shook up the presidential race last week with a strong showing in the first presidential debate. By early last week, polls from Gallup to Pew seemed to indicate the race had changed. 

The question is how much?

To help answer that question Patchwork Nation broke down the data inside the most recent Pew Research Center poll using its 12 types of counties to get a sense of where the Romney surge is most visible. Repeating a method of analysis we have used on previous Pew polls.

The breakdown, released here in conjunction with American University, shows a big surge in the exurbs for Romney (counties called Boom Towns) and some growth of support in the big cities (counties called the Industrial Metropolis). President Obama, meanwhile, still appears to hold the lead in the crucial suburban counties (counties called the Monied Burbs) and to hold great strength in the counties that are a heavy with college students (counties called Campus and Careers).

The map below shows the geographic distribution of the12 Patchwork Nation county types.

But a few important caveats should be noted. That Pew poll like most post-debate data snapshots came from a smaller sample. Pew surveyed 1,110 people for these results. That’s a good number, but smaller than its mid-September poll when it interviewed 2,200. And while there are clearly reasons to believe parts of the electorate moved with the debate, when the poll numbers are examined through the Patchwork Nation typology it appears maybe not as much as some believe.

Solid Gains

The most recent Pew survey broken down with Patchwork Nation shows big Romney gains in many of the 12 county types, but a few increases are particularly worth noting because they follow larger trends. Romney’s jump in the exurban Boom Towns is a good example.

Community Type

%Romney Sept.

% Romney Oct

Change

Population

Monied 'Burbs

42%

46%

4

69 Million

Minority Central

37%

51%

14

14 Million

Evangelical Epicenters

58%

67%

9

14 Million

Tractor Country

N/A

N/A

N/A

2 Million

Campus and Careers

36%

20%

-16

13 Million

Immigration Nation

45%

58%

13

20 Million

Industrial Metropolis

25%

37%

12

54 Million

Boom Towns

 

44%

53%

9

60 Million

Service Worker Centers

58%

53%

-5

31 Million

Emptying Nests

52%

35%

-17

12 Million

Military Bastions

45%

79%

34

9 Million

Mormon Outposts

N/A

N/A

N/A

2 Million

 

 

 

 

 

 

(N/A indicates the survey sample was too small to be used)

Romney’s support in the Boom Towns went from 44% before the debate to 53% after, a gain of nine points. That’s significant on it’s own, but it’s more important because it follows the larger historical trend. In 2008, Sen. John McCain captured 52% of the vote from the Boom Towns. In 2004, President George W. Bush captured 58% of the vote there.

Romney’s increased support in the socially conservative Evangelical Epicenters – up to 67% – also falls close to historical figures. McCain captured 66% of the vote coming from those counties in 2008.

In other words, Romney’s support in the new survey in those types of county is in a range that makes sense. There’s good reason to believe that his gains in those places are very real.

His gains in the Industrial Metropolis counties might also be “real.” His 37% is high, but not ridiculously so. George W. Bush captured 36% of the vote from those counties in 2004.

And Romney’s support in the Monied Burbs also is largely in-line with past Republican presidential performances. McCain captured 45% of the vote from the Burbs 2008. Bush won 49% in 2004.

And Real Questions

But looking at the numbers in a broader context, there are questions about what the Pew poll was capturing with President Obama’s vote. Some of the samples seem off particularly when viewed in a historical context.

Community Type

% Obama Sept

% Obama Oct

Change

Population

Monied 'Burbs

52%

51%

-1

69 Million

Minority Central

55%

42%

-13

14 Million

Evangelical Epicenters

36%

28%

-8

14 Million

Tractor Country

N/A

N/A

N/A

2 Million

Campus and Careers

60%

75%

15

13 Million

Immigration Nation

46%

33%

-13

20 Million

Industrial Metropolis

71%

57%

-14

54 Million

Boom Towns

51%

41%

-10

60 Million

Service Worker Centers

32%

39%

7

31 Million

Emptying Nests

41%

54%

13

12 Million

Military Bastions

N/A

N/A

N/A

9 Million

Mormon Outposts

N/A

N/A

N/A

2 Million

           
           

 

Look for instance at the number in Minority Central, those are communities that have large African American populations. Obama’s number, 42%, comes from a small sample and would extremely low historically. The vote is always close in those counties because of the divided nature of the populations. For instance, Obama won 50% of the vote coming out of those counties in 2008. In 2004, Sen. John Kerry won 48% of the vote.

The same is true with Obama’s numbers in Immigration Nation, which have large numbers of Hispanics. The 33% number in those counties also comes from a small sample and is most likely off. Obama won 51% of the vote in those counties in 2008 and Kerry won 45%. Romney’s 58% support in those places is, similarly, extremely unlikely, particularly considering the troubles the GOP is having wooing Hispanic voters in 2012.

And Obama’s numbers in the Industrial Metros are likely low, as well. In the last three presidential elections no Democrat has gotten less than 63% of the vote from those urban counties.

Meanwhile, Obama’s numbers in the aging Emptying Nest counties in this survey, may be inflated. He won 49% of the vote in those places in 2008 and 54% support would be high above the historical norm for a Democrat.

One Obama number that looks right is the Monied Burbs. His 52% is close to the recent historical norm for a Democrat. Democratic presidential candidates have won between 50% and 55% of the vote coming out of the Burbs in the last three presidential elections.

And now…

Why do all those historical norms matter? Because in the end, despite the common chatter in Washington media circles of groundswells and game-changing elections, Patchwork Nation has found that, historically speaking, the numbers really don’t change that much.

If Obama actually did win the vote coming out of Boom Town counties (as it appeared in the last poll) or if Romney really held 25-percentage point lead in the vote in the Immigration Nation counties (as this poll indicates) those would be startling figures.

Since 2000 the percentage of support for Democratic or Republican presidential candidates has moved by between 1 and 4 points in most of the Patchwork Nation county types. And the vast majority of them do not switch between Democratic and Republican support. The changes are about margins – how much candidates win by.

To be clear, there is nothing “wrong” with the latest Pew poll. All polls have slightly different samples.

The latest Pew numbers seem to capture two things: It caught the peak of the post-debate Romney debate bounce and did that with a sample was a little different than usual. Those things together should serve as a reminder that reading too much into any poll can be a tricky business.

Patchwork Nation will continue to analyze polling data with American University through the election.

This analysis was posted in conjuction with American University's Center for Congressional and Presidential Studies.

METHODOLOGY: The Pew Research Center conducted polls with registered voters on October 4- 7. For each respondent in the polls, zip code data was collected. These data were assigned one of the twelve community types described in Patchwork Nation. Respondents were asked about their support for President Obama or Governor Romney. Responses to these questions were compared for each community type. For each poll and community type, the margin between Obama and Romney was calculated by subtracting the percentage supporting the Republican/Romney from the percentage supporting Democrat/Obama. The margin of error in each community type and poll is dependent on the size of that particular sample.