SurveyUSA poll: Crist leads by 5% as Scott spends $5 million on TV ads

@MarcACaputo

A new SurveyUSA poll shows Democrat Charlie Crist leading Gov. Rick Scott by about 5 percentage points --and despite $5.2 million in ad buys from the Republican.

The likely voter poll, conducted for Tampa's WFLA station, jibes with about a dozen other recent public surveys except two:

1) An early April poll by Voter Survey Service, which often works for Republicans and was hired by conservative Sunshine State News. It showed Scott with an inside-the-error margin lead of 45-44 percent over Crist, the only survey showing Scott out front.

2) A mid-March poll by University of North Florida (which until the Voter Survey Service was the outlier) that showed Crist with the narrowest of inside-the-error-margin leads, 34-33 percent.  No recent survey has found both candidates pulling less than 40 percent of the vote.

In between those two surveys, St. Leo University released a poll in late March that Crist led Scott 43-39 percent.

And the day after the Voter Survey poll, liberal-leaning Public Policy Polling released its own numbers showing Crist ahead of Scott 49-42 percent.

Considering all those polls, SurveyUSA's numbers of Crist leading Scott 46-41 percent is no shocker.

Bottom line: Crist probably leads by 3-5 percentage points.

But there's a catch: Most of these surveys oversample Democrats. In mid-term elections, they tend to underperform relative to their numbers while Republicans overperform.

For instance, SurveyUSA's poll had 34 percent Republicans; 38 percent Democrats and 27 percent independents. Putting aside a discussion of how voters are identified by party, that's too many independents and too few Republicans. But even if we take the high-watermark voter performance for 2010 (R=44.4%; D=39.6% and I=16%) and apply it to SurveyUSA's poll, it shows Crist with a tenuous .07 point lead (Crist=44.3% and Scott=43.6%).

That's bad news, Republicans.

But the crosstabs of SurveyUSA and St. Leo have bad news for Democrats. They both show Crist losing the independent vote to Scott. They also both show Scott underperforming with Republicans relative to how much Democratic support Crist gets, or getting fewer crossover voters (Democrats who vote for the Republican vs. Republicans who vote for the Democrat) than Crist.

So if Scott keeps serving up the red meat in his Obamacare-centric ads, there's a good chance that more Republicans will come home to him. In that case, he starts to win.

As I've repeatedly stated on the blog, in columns and in news stories: Don't be surprised if this race is so close it requires a recount. It's not a matter of if Scott starts catching up. It's when.

It's still early. And one statistic is beyond a doubt: There's a 100 percent chance of another poll coming out soon.

Oh, and count on Libertarians being miffed. Their candidate, Adrian Wyllie, wasn't polled.

Meantime, here's the analysis from SurveyUSA:

29 weeks till votes are counted in Florida, incumbent Republican Governor Rick Scott is the underdog in a fight against (now) Democratic challenger Charlie Crist, according to a SurveyUSA poll conducted for WFLA-TV in Tampa. Today, it's Christ 46%, Scott 41%. 502 likely voters interviewed.

Scott's support is older. Crist's support is younger. The older the electorate on Election Day, the better Scott will do. The younger the electorate, the better Crist will do.

Scott leads 2:1 in Northeast Florida, and more narrowly leads in Southwest Florida (which includes the Tampa Bay region). Crist leads 3:2 in Southeast Florida and more narrowly in Central Florida and Northwest Florida. A small turnout in Dade and Broward counties helps Scott. A large turnout in Dade and Broward counties helps Crist. Scott holds 71% of the Republican base, compared to Crist, who holds 80% of the Democratic base. Independents split. Conservatives vote 7:1 Republican. Liberals vote 8:1 Democratic. Importantly, moderates break 7:4 for the Democrat Crist.

Scott and Crist are effectively even among college graduates. Crist leads among less educated voters. Scott leads among upper-income voters. Crist leads among lower-income voters. Middle-income voters split. Scott leads narrowly among white voters; Crist leads overwhelmingly among African Americans, and materially among Asian Americans. Both Cubans and non-Cuban Hispanics lean slightly Republican -- but that is based on a small sample and caution should be exercised in projecting those numbers out to Election Day.

Cell-phone and home-phone respondents included in this research: 940 state of Florida adults were interviewed by SurveyUSA 04/10/14 through 04/14/14. Of the adults, 799 were registered to vote. Of the registered, 502 were identified by SurveyUSA as being likely to vote in the 11/04/14 general election. This survey was conducted using blended sample, mixed mode. Respondents reachable on a home telephone (74% of likely voters) were interviewed on their home telephones in the recorded voice of a professional announcer. Respondents not reachable on a home telephone (26% of likely voters) were shown a questionnaire on their smartphone, tablet or other electronic device.

But there's a catch: Most of these surveys oversample Democrats. In mid-term elections, they tend to underperform relative to their numbers while Republicans overperform.

For instance, SurveyUSA's poll had 34 percent Republicans; 38 percent Democrats and 27 percent independents. Putting aside a discussion of how voters are identified by party, that's too many independents and too few Republicans. But even if we take the high-watermark voter performance for 2010 (R=44.4%; D=39.6% and I=16%) and apply it to SurveyUSA's poll, it shows Crist with a tenuous .07 point lead (Crist=44.3% and Scott=43.6%).

That's bad news, Republicans.

But the crosstabs of SurveyUSA and St. Leo have bad news for Democrats. They both show Crist losing the independent vote to Scott. They also both show Scott underperforming with Republicans relative to how much Democratic support Crist gets, or getting fewer crossover voters (Democrats who vote for the Republican vs. Republicans who vote for the Democrat) than Crist.

So if Scott keeps serving up the red meat in his Obamacare-centric ads, there's a good chance that more Republicans will come home to him. In that case, he starts to win.

As I've repeatedly stated on the blog, in columns and in news stories: Don't be surprised if this race is so close it requires a recount. It's not a matter of if Scott starts catching up. It's when.

It's still early. And one statistic is beyond a doubt: There's a 100 percent chance of another poll coming out soon.

Oh, and count on Libertarians being miffed. Their candidate, Adrian Wyllie, wasn't polled.

Meantime, here's the analysis from SurveyUSA:

29 weeks till votes are counted in Florida, incumbent Republican Governor Rick Scott is the underdog in a fight against (now) Democratic challenger Charlie Crist, according to a SurveyUSA poll conducted for WFLA-TV in Tampa. Today, it's Christ 46%, Scott 41%. 502 likely voters interviewed.

Scott's support is older. Crist's support is younger. The older the electorate on Election Day, the better Scott will do. The younger the electorate, the better Crist will do.

Scott leads 2:1 in Northeast Florida, and more narrowly leads in Southwest Florida (which includes the Tampa Bay region). Crist leads 3:2 in Southeast Florida and more narrowly in Central Florida and Northwest Florida. A small turnout in Dade and Broward counties helps Scott. A large turnout in Dade and Broward counties helps Crist. Scott holds 71% of the Republican base, compared to Crist, who holds 80% of the Democratic base. Independents split. Conservatives vote 7:1 Republican. Liberals vote 8:1 Democratic. Importantly, moderates break 7:4 for the Democrat Crist.

Scott and Crist are effectively even among college graduates. Crist leads among less educated voters. Scott leads among upper-income voters. Crist leads among lower-income voters. Middle-income voters split. Scott leads narrowly among white voters; Crist leads overwhelmingly among African Americans, and materially among Asian Americans. Both Cubans and non-Cuban Hispanics lean slightly Republican -- but that is based on a small sample and caution should be exercised in projecting those numbers out to Election Day.

Cell-phone and home-phone respondents included in this research: 940 state of Florida adults were interviewed by SurveyUSA 04/10/14 through 04/14/14. Of the adults, 799 were registered to vote. Of the registered, 502 were identified by SurveyUSA as being likely to vote in the 11/04/14 general election. This survey was conducted using blended sample, mixed mode. Respondents reachable on a home telephone (74% of likely voters) were interviewed on their home telephones in the recorded voice of a professional announcer. Respondents not reachable on a home telephone (26% of likely voters) were shown a questionnaire on their smartphone, tablet or other electronic device.