Reaching Young Voters
A recent memo from Barack Obama's campaign manager took aim at a familiar part of the Iowa landscape -- a straw man.
To explain away any apparent deficit his candidate faces in Iowa, David Plouffe declares that polls are inaccurate gauges of the status of the race for the Democratic nomination. He argues that polls consistently underestimate the "strength of Barack's support among younger voters."
Despite the set-up, Plouffe's main point is not that polls fail to capture the level of Obama's support among young adults -- he actually posits polls get that right. Instead he lays out three reasons polls misjudge the number of young voters. He is right about the challenge pollsters face, but likely hasty in alleging all polls report biased estimates.
Claim 1: pollsters systematically under-represent young voters because those voters are less likely to have caucused in the past.
Reality: Depends on the poll. Polling in Iowa is notoriously difficult, in part because of how few people participate in the January (or even earlier?) caucuses. And methods always matter: some argue the only way to poll in Iowa is to use the party lists of previous caucus-goers, but our approach in a summertime Post-ABC poll leveled the playing field by using random digit dialing. (ABC's Gary Langer has a detailed explication of this choice. About three in 10 likely Democratic caucus-goers in our poll said they'd be attending their first caucus. On the GOP side, about a quarter of likely voters said the same thing.
Claims 2 and 3: young people are more "mobile" and less likely to be home to answer a phone poll and they're less likely to have a landline anyway so for both reasons, standard phone polls tend to miss them.
Reality: This is true, but probably only in unimportant ways. The proportion of young adults we interview in a typical phone poll is lower than the proportion of that age group in the population. The reasons Obama's top campaign strategist cite are two key factors in explaining this. However, often lost is that young people DO participate in polls, and poll results are biased only if the young adults who answer questions are systematically different from those who don't. If the two groups are similar, the known undercount can be corrected by additional interviews with randomly selected young people, or by statistically adjusting the data to match data from the Census.
There's the rub.
Is there any reason to believe the young people who have gone "cell only" are different from those who are reachable on landlines? Those who have cut the cord likely have different views of technology and perhaps other things, but why would they have different political views?
On the contrary, there's reason to believe the choice of home phone doesn't make a large difference in attitudes. A study by the Pew Research Center shows that national political polls are not likely biased by the exclusion of cell phone numbers from the typical sample. (This is testable because it's possible to poll on cell phones, it's just not standard practice.) And a look at the 2004 NEP exit poll, shows remarkably comparable vote preferences, no matter what kind of phone they had at home.
In 2004, 18 to 24 year-old voters who had only cell phones at home went for John Kerry over George Bush 63 percent to 36 percent (+/- 9 percentage points); those with landlines broke 59 percent to 39 percent (+/- 5). The results were similar, both substantively and statistically.
For more on the Plouffe memo, plus helpful stuff on the cell phone "problem" in survey research, see Mark Blumenthal's take.
Posted by: goldie2 | October 5, 2007 10:46 AM | Report abuse
The comments to this entry are closed.