Two polls both show Democrat Terry McAuliffe in a narrow lead over Republican Ken Cuccinelli, but one calls it 'a tie'. How so?
When you think of political polls, the first thing that probably comes to mind is one candidate winning and another losing. After all, we tend to judge a pollster's performance on whether they are right or wrong in predicting the winner. Top lines, however, don't always tell the whole story. Never has this been clearer than in two polls released over the last two days for the Virginia gubernatorial election.
The two polls, released by Public Policy Polling (PPP) and Quinnipiac, are statistically similar at first glance. Democrat Terry McAuliffe leads Republican Ken Cuccinelli in both surveys. His advantage is 5pt per PPP and 1pt per Quinnipiac. This spread fits with the two prior polls in the race that had McAuliffe ahead by 4pt and 1pt, given the margin of error.
The correct read on this race given the four polls is one in which McAuliffe seems to be holding a slight lead, with a long way to go before election day. That, however, was not how either PPP or Quinnipiac described the race in their respective write-ups.
PPP said, "McAuliffe opens 2013 with a lead over Cuccinelli." The release quoted president of Public Policy Polling Dean Debnam saying, "Ken Cuccinelli's unpopularity is really the story of the race at this point."
Quinnipiac countered PPP's headline with "Cuccinelli, McAuliffe tied as Virginia governor race begins." Peter A Brown, assistant director of the Quinnipiac poll, went on to note that while the "candidates for governor have run statewide previously, voter memories are short and they are little-known to Virginia voters."
How is anyone supposed to make sense of the differences presented here? You'd think PPP and Quinnipiac were polling two different races, despite relatively similar headline figures. But if we look under the hood, it turns out that things becomes even more complicated.
The reason PPP regards Cuccinelli as unpopular is that they find he is. His net favorable (favorable minus unfavorable) is -16, with 73% of voters registering an opinion of him. PPP often finds politicians with lower favorables than other polls, but even when controlling for pollster, Cuccinelli is seemingly deeply unpopular. Of the 126 current major statewide office holders PPP has polled over the past few years, Cuccinelli would rank seventh lowest, or be in the bottom fifth percentile. Heck, he can't even get to 50% favorability within his own party, with a favorable rating of just 48% among Republicans.
Quinnipiac, on the other hand, sees Cuccinelli as relative unknown because he is in their polling. His net favorable is +8, with only 59% of all voters holding an opinion about him. Cuccinelli also hits 64% favorability among Republicans. Part of that is question ordering, whereby Cuccinelli is identified as a Republican before favorability is asked. Still, you'd expect that ordering to drive up his unfavorables among Democrats and independents. Instead, Cuccinelli's unfavorables are 20pt lower among both of these groups in the Quinnipiac survey.
PPP and Quinnipiac simply don't jibe. What's more bizarre is the rest of each of the polls tends to match up fairly well, or, if anything, would suggest that McAuliffe would do better in the Quinnipiac poll. The Quinnipiac survey, at 69%, is 4pt less "white" than PPP, which should favor the Democrat McAuliffe. The Quinnipiac universe is significantly younger, with 44% of voters under 45, compared with only 36% from PPP, which should also favor the Democrat McAuliffe. And if you're into this sort of thing, Democrats hold a 7pt lead in self-party identification, per Quinnipiac, which is greater than the 3pt edge PPP finds.
What's more, distinctions between the other possible candidates besides Cuccinelli are relatively minor in the two polls. Both have McAuliffe being unknown by 50%-plus of the electorate. His net favorable of +7 is actually higher in Quinnipiac, though that's not surprising given that PPP generally has politicians with lower favorability overall. The possible Republican-turned-Independent candidate Bill Bolling is unknown, by an even greater number of voters. He has a net favorable of about +10 in both polls. Bolling takes about 15% of the vote in a hypothetical three-way in both surveys.
The only real difference is that PPP shows McAuliffe expanding his lead to 8pt when Bolling is included, while Quinnipiac shows it shrinking by 1pt to a tie. The explanation for this divergence is quite simple. Both Democrats and Republicans pass over their party's candidates equally, to go to Bolling in Quinnipiac's poll, while only Republicans leave their man, Cuccinelli, in PPP's survey. This makes sense given that Republicans are a lot more opposed to Cuccinelli in the PPP universe.
So, we're presented with two polls, from Quinnipiac and Public Policy Polling, for the same race that are the same on top, but different inside. I throw up my hands in puzzlement for why this is. All I can say is that Democrat Terry McAuliffe is a relative unknown and has a small lead in the Virginia gubernatorial election. Republican Ken Cuccinelli's high net unfavorability per PPP, and decently high net favorable rating per Quinnipiac, simply can't be explained – beyond saying it's statistical noise.