Ask a Bad Question, You’ll Get a Bad Answer

by Stuart Rothenberg July 8, 2011 · 1:58 PM EDT

I avoid back-and-forth public arguments, believing that after I’ve made my case readers can decide for themselves who and what they believe. But I’m making an exception with this follow-up column.

I recently challenged those who argue that unpopular Republican governors seriously damage the prospects of the eventual GOP presidential nominee in key states such as Florida and Ohio. I have yet to see any evidence that past presidential hopefuls lost states that they should have won because of unpopular governors, and the theory doesn’t make sense if you understand how and why voters make decisions in presidential elections.

But a June 16-19 Public Policy Polling survey of 848 Florida voters, accompanied by a June 24 press release, reiterates the assertion that Florida Gov. Rick Scott is an “unambiguous problem” for the Republican presidential nominee next year.

PPP, of course, is a North Carolina-based Democratic automated interview research firm that was among the first to make the argument, in late May, that Scott will hurt his party’s White House nominee.

I’m not taking issue with the PPP numbers themselves, though it is true that the firm has plenty of critics. Let’s assume for a moment that the data are correct. My problem is with the question and the analysis, both of which leave a lot to be desired.

The 12th question in PPP’s survey asked: “Have Rick Scott’s actions as governor made it more or less likely that you’ll vote for a Republican for president next year, or has it not made a difference?”

The results found that 26 percent said that Scott’s actions as governor of Florida made them more likely to support the Republican nominee, while 40 percent said it made them less likely and 34 percent said it made no difference.

“Voters there aren’t in love with [President Barack Obama] but when they look at the Republican alternatives and they look at what Rick Scott’s done while in office as governor, he starts to not look so bad,” PPP President Dean Debnam said in the press release.

The responses to this single question appear to be the crux of the case for those (including PPP and TalkingPointsMemo.com) who argue that these data confirm a causal connection between Scott’s unpopularity and presidential vote choice next year.

In fact, there are lots of problems with the alleged causality, the question and the analysis, and anyone who spends a lot of time with survey research should know it.

First, speculative “more likely” and “less likely” questions are always dangerous because they sometimes measure the underlying attitude rather than the effect that attitude will have on another decision.

In this case, the question about Scott is likely measuring the public’s evaluation of his job performance, which apparently was not asked by PPP in this survey.

Take a wild guess which voters are most likely to agree that Scott’s actions make them less likely to vote for the Republican nominee for president. Of course, it’s self-identified Democrats — who aren’t going to vote Republican in 2012 anyway. (Oddly, 17 percent of self-identified Democrats in the survey said that Scott’s actions made them more likely to vote Republican in the next presidential election.)

Then there is the question of when an attitude becomes so strong and salient that it changes an anticipated behavior.

For example, a voter could say he is less likely to vote for Obama if the president doesn’t come out strongly for gay marriage, yet still vote for him when the election rolls around even if Obama doesn’t meet that test, because for most people vote choices are about more than one issue or one opinion.

But most important, the current political context and the one that will exist when the next election takes place will be dramatically different, making the responses to the Scott poll question virtually meaningless.

In this poll, PPP asked about Scott’s effect immediately after a series of ballot tests and favorable/unfavorable ratings, so that respondents at the other end of the telephone had only Scott and possibly a few other political names on their minds. That was the entire context of their thinking and of their responses.

They weren’t asked about issues, the economy or anything else that might color their thinking about the 2012 presidential race when they actually have to cast a ballot.

By late October 2012, voters will have been inundated with information about Obama and the Republican nominee, as well as articles and ads about a variety of national issues and candidate qualities — whether it’s Medicare and Republican Rep. Paul Ryan’s (Wis.) budget, jobs, the economy, the killing of Osama bin Laden, Big Oil or taxes and spending.

After debates, national conventions, millions of dollars of TV spots and weeks of nonstop news coverage about the candidates, voters will go to the polls to pick the nation’s next leader. When they do so, Scott will constitute only a microscopic part of their information as they consider their vote for president. He certainly won’t be nearly as salient as he was in the June PPP survey, even if Democrats spend millions of dollars in Florida to try to make the presidential election about Scott.

Pollsters always emphasize that surveys are mere snapshots. Respondents can only answer the questions asked, no matter how meaningless they are. This PPP survey question is of little use in predicting voter motivation in November 2012.

I don’t know who will carry the Sunshine State in the 2012 election. But to argue that the Florida governor will cost the Republicans the state in 2012 is to argue that Scott will be more important than the presidential candidates, the issues and all of the media coverage surrounding the contest. If you believe that, you don’t understand campaigns and elections.