Steve Vancore: A look at polling in the 2014 Florida governor’s race

First of a Series

The question came to us the other day, “Why poll just Rick Scott versus Charlie Crist?”

“Why not also include the other folks on the ballot?”

Good questions.

Why would you poll test just the two, which many high-quality professional pollsters are still doing?  After all, the logic follows, there will be more than just Rick Scott and Charlie Crist on the ballot.  There will also be Libertarian candidate Adrian Wyllie and – many people seem to have forgotten – two other NPA candidates as well. (A few weeks ago there were three other candidates, but one was kicked off the ballot because he failed to name a running mate.)

The primary reasons you would do this are perfectly sound and have to do with external validity and consistency.  Let me explain.

There are times when a campaign or an independent pollster will need to see trend lines.  For good reason, they will want to compare old data with new findings.  To compare the new data with older polls, you need to poll using the exact methodology as the prior poll so that you are comparing apples to apples. This means asking the same question the same way.  This helps people understand things like … How is Rick Scott trending?  How is Charlie holding up with his base?  How are undecided voters breaking?  What movement do we see among likely partisans?  All of these questions need to be answered and can be best answered by keeping the test ballot exactly as it has been for the past six, eight, or 10 months.

For this reason, it often makes sense to measure the top of the ticket test ballot with just Rick Scott vs. Charlie Crist.

But hold the phone.

There are also very good reasons to add in Wyllie and the other candidates.  If you want to see where the race “really” is, it is important to add the other candidates on the ballot as well. If voters are upset with the two major party candidates, the logic goes, they might be more inclined to select a third or fourth option. (As an aside, for some odd reason, almost every publicly reported poll only adds Wyllie into the mix without adding the other NPA candidates.  I cannot explain this.  We believe that if you want an accurate read of the ballot, you must replicate – as close as is feasible – the actual ballot experience.)

Adding these other candidates makes perfect sense if this is what you want to know. But if you had been tracking the race for the past year and had only tested just Scott and Crist all along, to change the wording of the question now would falsely alter your data trend lines.

So is adding Wyllie and the others a better approach?

First, it really depends on the purpose of the poll.  If you want to know where the actual race is right now, probably yes.  But there are other problems in adding, in particular, Adrian Wyllie, whom nobody seems to be talking about.

Adding the words, “Libertarian Adrian Wyllie” or “Adrian Wyllie, Libertarian” as many polls are doing – verbally stating his party as “Libertarian” – can give you a false read – likely inflating his numbers.

Whether we like it or not, the ballot will not include the word “Libertarian.”  It will say “LPF.”  So in fact, when voters SEE his name on the ballot, here is what they will see:

___Adrian Wyllie (LPF)

And let’s be honest, how many voters will know what the acronym “LPF” stands for?

I honestly don’t know any way around this.  Asking “Adrian Wyllie” and then saying, “L-P-F” seems awkward and confusing and it potentially infuses even more error into the equation.

Furthermore, this does not even address the false read you may get from the fact that most pollsters, ourselves included, will attest that most – yes, most – voters do not really know what “Libertarian” actually means.  In fact, we often see a large percentage of self-described liberals claiming they consider themselves to be libertarian.

Finally, we run into the problem of the “No Party Affiliation” candidates Glenn Burkett and Farid Khavari. They will appear on the ballot as:

___Glenn Burkett (NPA)

___Farid Khavari (NPA)

Since about one quarter of voters are registered “NPA” – far more than are registered Libertarian – it stands to reason that more voters will a) know what NPA means and b) be more likely to vote for someone with the same party registration as themselves.

If that were the case, we would be far more likely to see NPA candidates getting out of the single digits.

So what’s one to do?

In a perfect world (a world I surely do not live in), most polls would replicate the visual experience of the voting process.  Pollsters would conduct these polls in person (or on the Internet, but that brings up a whole new slew of problems) and respondents would see the ballot and respond accordingly.  Further, you would now have the opportunity to add the lieutenant governor candidates to get an even more accurate read (a very tough thing to do in a telephone poll).

In-person polls are really expensive and time-consuming.  And, Internet polls have problems that will be addressed in a future column.

So back to our imperfect world. …

We have found that the best option is to ask the question both ways, once with the third-party candidates and once without (and, if you like, rotate the questions to avoid order bias).  In doing so, you can see the overall trending of the race as it pertains to the two likely winners and you can get a decent feel for where the race stands today.  You make your trend-line people happy while also getting a better feel for what impact the third-party candidates are having on the race.

Steve Vancore is president of VancoreJones Communications and Clearview Research and has been conducting polling and voter research in Florida since the mid-1980s.   He can be reached at [email protected]. Column courtesy of Context Florida.