Morning Jay: Polling Nonsense
6:00 AM, Jun 29, 2011 • By JAY COST
The Des Moines Register poll of Republicans caused quite a stir this week. The congresswoman from Minnesota could not have asked for a better piece of news to correspond with her official announcement: It showed Michele Bachmann down just one point to Mitt Romney in Iowa. Meanwhile, Tim Pawlenty had to suffer through idle questions about whether or not he was a “first tier” candidate.
But just how seriously should we take that poll, and others like it? I say, not very seriously at all.
Let’s start with some recent history. At this point in 2007 the Iowa caucus polls showed Barack Obama trailing Hillary Clinton and John Edwards. When it was all said and done in January 2008, Obama would pull out an eight-point victory. On the GOP side, early Iowa polls had Mitt Romney in first place, Rudy Giuliani in second, and Mike Huckabee pulling an average of less than 5 percent of the vote!
National polling from June 2007 looks just as ridiculous. At that point, Clinton had a 10- to 20-point lead over Obama, which would expand to 30-points (and more) by the fall. By June 2008, when all the primaries and caucuses were finished, the two had basically split the Democratic vote. On the Republican side, Rudy Giuliani had a 10-point or greater lead over John McCain in the national polls, while Mitt Romney and Mike Huckabee were both polling less than 10 percent each. When it was all said and done, McCain won 47 percent of the vote, Romney and Huckabee both won a touch more than 20 percent, and Giuliani…won just 3 percent!
So the final 2008 results did not correspond at all to the numbers from the summer of 2007. More broadly, the nomination process as we know it today has produced surprising nominees time and time again since it was first implemented some 39 years ago – George McGovern in 1972, Jimmy Carter in 1976, Bill Clinton in 1992, John Kerry in 2004, and John McCain in 2008. At this point in each cycle, nobody really saw any of these guys taking the top prize.
There are good reasons why we are so consistently surprised by the nomination outcome. The biggest one is that primary battles neutralize the most valuable piece of information that voters have: the party label. When you tell people that you are a Republican, they can get a good sense of what your opinions are on a whole host of issues. Ditto if you are a Democrat. If you are running for office, your party label will determine how up to 90 percent of the electorate will vote. But in primaries, that information is simply not useful – all your opponents are either Republicans or Democrats – which means that voters have to gather information in other, more time-consuming, less reliable ways. This makes it really difficult to predict how a primary will happen six months out. Voters know precious little about the candidates at this point, and more importantly most of them are not even trying to learn just yet. It’s the summer for goodness sake, and the 90 percent of Americans who are not political junkies just don’t care yet. There will be plenty of time for them to figure out whom to support when the leaves start falling from the trees...in four months! Thus, the choices that polling respondents offer here in June are far too contingent on future learning to have any real value.
Beyond that, it is worth noting that no pre-primary polling mimics the actual nomination process. The Iowa caucuses are unique in that Republican caucus goers might go in planning to support one candidate, but come out supporting another. The reason is that, in the caucus meeting, participants listen to campaign pitches for each of the candidates. So, weakly felt preferences – the very sort that the Des Moines Register poll is finding today – could very easily change. After Iowa, primary results often depend very heavily on momentum, or the phenomenon wherein a candidate’s victory in a previous contest increases his chance of victory in the next contest. Momentum is closely related to the fact that primary voters know very little about the specifics of each candidate, which makes winners seem especially appealing. Again, this is a dynamic that pre-primary polls simply cannot capture.
So, I would recommend extreme caution in interpreting polls such as the Des Moines Register poll, so early in any presidential cycle. But for this cycle, more prudence than normal is warranted – this is the most open Republican nomination battle in the modern era. If ever was there a year that a candidate was to surge from nowhere to capture the nomination – like Jimmy Carter in 1976, for instance – it looks to me like it will be this year. That would make early polling even more dicey.
Most pundits and journalists are not going to heed this advice. Far too dependent on the political polls, they’re going to cover the nomination battle as if they’re surfing a wave – going up, then down, up then down, up then down. In that way they will miss what is really going on. For some reason, that’s what they always seem to do. To be a smart consumer of political information over the next six months, you’re going to have to take the opining of the media “experts” with a grain of salt, and remember that primary nominations are often quite volatile, and this one looks to be even more so.