Political Animal


June 13, 2012 2:42 PM Likely Voters and GOTV

By Ed Kilgore

In the midst of all the triumphalist talk we’re now hearing from Republicans about the trajectory of the presidential race, you will often hear references to polls showing Obama ahead being tainted because they include registered, not “likely,” voters. Indeed, that’s the standard argument for paying attention to Rasmussen polls that otherwise look pretty plainly like pro-Republican outliers: Raz deploys a “likely voter screen” way out from elections, generally considered a bad idea by public opinion experts.

But other than supplying an excuse for placing a thumb on the scale for every Republican candidate running for offices from president to dogcatcher, what do we really know about likelihood to vote? TNR’s Nate Cohn has a useful primer on the subject, and about what the Obama campaign is doing to boost its own constituencies’ likelihood to vote:

The difference between likely and registered voter polls is largely attributable to changes in the racial/ethnic composition of the electorate, with young and non-white voters constituting a smaller share of likely voters than registered voters. Gallup’s tracking poll shows that 81 percent of white voters say they will “definitely vote,” compared with just 71 percent of non-white voters. Similarly, Gallup finds that 88 percent of Republicans say they will definitely vote, more than the 82 percent of Democrats who respond similarly. Just 58 percent of 18-29 year old voters say they will definitely vote—a number unchanged since April, when 59 percent said they would definitely vote. Armed with data from two Monmouth University polls, Harry Enten has similar findings.
While these numbers might be sobering for Democrats, the heart of the campaign season is still months away and likely voter polls in June might not provide a good indication of the eventual electorate. Even over the last month, Gallup’s weekly tracking poll has shown Hispanic voters becoming more energized, with 66 percent of Hispanics saying they will “definitely” vote, up from 58 percent in April. No other demographic group has shown similar movement….
[T]he likely voter models suggest that the Obama campaign can’t count on repeating their historic performance with non-white voters. In this context, the Obama campaign’s decision to invest enormous sums in a sophisticated ground operation is unsurprising. In addition to potentially prodding Obama supporters who currently seem unlikely to vote, a strong ground operation can help take advantage of demographic changes by adding newly registered voters who can help compensate for diminished enthusiasm and lower turnout rates among Obama ‘08 supporters. While it’s hard to say whether Obama’s vaunted ground operation will replicate ‘08 turnout among young and non-white voters, there’s no question that the Obama campaign is willing to invest the money necessary to find out.

The bottom line is that the whole point to GOTV investments is to boost “likely voter” rates. Different elements of the electorate have different historical rates, but they don’t come down from Mount Sinai on stone tablets. That’s why most pollsters don’t bother deploying a screen right now, and why they are right to wait until much closer to November to begin making assumptions about the shape of the electorate.

Ed Kilgore is a contributing writer to the Washington Monthly. He is managing editor for The Democratic Strategist and a senior fellow at the Progressive Policy Institute. Find him on Twitter: @ed_kilgore.


  • Wallys son on June 13, 2012 3:35 PM:

    Check out gottavote.org, from the Obama team.

  • TCinLA on June 13, 2012 5:12 PM:

    The two smartest guys I ever met regarding politics were Jesse Unruh and Willie Brown. They both had the same rule for winning elections: always run like you're ten points down, regardless of what the polls say.

  • Rip on June 13, 2012 6:42 PM:

    Rasmussen's likely voter model is +1.4% Republican over Democrat in terms of party I.D. According to exit polling the 2008 electorate was +7% Democrat, and in 2010 party I.D. was even. In both years self-identified independents made up about 28% of the electorate.

    Given that the off-year electorate generally skews more Republican than in Presidential years, it's safe to say the 2012 electorate will be somewhere in between, meaning that Rasmussen's likely voter model is at least 3 or 4% more Republican than what the actual electorate will be. In other words: useless.

  • bluestatedon on June 13, 2012 8:00 PM:

    I am continually amazed at how those with the most to gain by voting—historically disadvantaged minorities and those under 25—are the most likely to not care about voting at all. Aside from those illegitimately prevented from voting by GOP machinations, it's really hard for me to have much sympathy for those who can't rouse themselves and get off their asses one day every two years to exercise a fundamental Constitutional right.