Much Ado About Polls

As summer approaches (thankfully for many of us), it means that more focus will now shift to the elections in November. We can already guess that it is going to be a very close race for the White House, as well as a nail-biter for the Senate and Wisconsin recall elections. Normally, one looks to polls for information regarding how well each party is doing, but this year, it seems that even they don’t know what is going on.

According to left-leaning Daily Kos, which actually has very good election coverage, polls for the presidential race are anywhere from Romney by 1 (Rasmussen) to Obama by 7 (Fox News). On RealClearPolitics, Obama has a 2.5 lead in the RCP polling average, which is a measure of 9 different polling companies over the last three weeks. RCP also shows Obama with a 253-170 electoral vote edge over Romney, with 125 still up in the air. Depending on who you believe, Obama is either winning big or losing close.

So which view is the accurate view? Well, none are really accurate. Other than the fact that there are still 5 months until the election, many pollsters are still using “registered voters” in their samples, rather than “likely voters”, which lean Democrat due to registration advantage. This is not an accurate reading because some registered voters don’t actually vote. In 2004, only around 70% of registered voters turned out to vote, which was at the time the heaviest turnout for an election in US history. That leaves a large chunk of the population that did not show up to vote, which could affect polling data. Rating likely voters would include more people from the 70% who do go to the polls, giving us a more accurate reading. If you want to know which polls to look at, check the sample size for “likely voters”, as these will be the more accurate ones.

The second thing to look for is party affiliation. Polling companies like Magellan Strategies ( R), and Public Policy Polling (D), have partisan affiliations, so any poll they give can be interpreted as having a 2-5 point swing in that candidate’s favor, due to a heavier sample from partisan voters. They are probably good polls if the samples are correct (PPP switches to Likely Voters as the elections draw nearer), but be mindful that if they are tied to a party, they will be favoring one candidate over another.

The third thing is, for the Presidential race, look at individual state polls rather than the national polls. Remember, it’s about the Electoral College, not national popularity. Most states, by now, will have polling data out, with swing states like Ohio and Florida receiving more attention than dead-certain states like Massachusetts or Texas. On RealClearPolitics’ EV map, you can click on an individual state and see the polling results for that state. This is what goes into their Electoral Vote projections, and it becomes easier to see which states are truly competitive, and which are not worth the time or effort from the campaigns.

The fourth, and most important, qualification is openness. Any good polling firm should have their criteria, crosstabs, and questions available for public access, showing the results from every single cross-section from race to income level. This is a sign of a polling company that knows its stuff, and isn’t afraid to share it. For wonks like me, this is invaluable because you can see how results get where they are. If I can’t see the nitty-gritty of a poll, I don’t trust it. Local media and university polls typically do this, as do some polling companies like PPP, which are often considered the most accurate in local races. If you are following the Wisconsin recall election, I highly recommend Marquette Law School’s polls; they are very comprehensive and done with care and efficiency. If you are a New Englander like me, Suffolk University is the gold standard in the Northeast.

Oh, and for those Fox News, MSNBC, and CBS polls, don’t worry about them. They tend to put too much emphasis on party registration, and almost never paint an accurate picture of the electoral landscape. Same goes for interest-group polling and internal campaign polling. At the end of the day, it’s about their interests, not yours. Hopefully, this will help you navigate the minefield of polls in the months to come, and you can see the shifting sands with a little more clarity and focus.

Published in

Post a comment