August 2012


Twitter as a micro-targeted mobilization tool

Higher rates of Twitter usage for blacks and for sub/urban dwellers, but neither category is anywhere near as strongly predictive as age. (Source: Pew Internet & American Life Project)

Yesterday I mentioned the turnout uncertainty of Barack Obama’s young and black supporters, built at least partly on the perception that their votes aren’t needed to get Obama over the top. These aren’t people who need to be persuaded to vote for Obama, but they do need to be persuaded to vote. We know that simply being asked to vote does a surprisingly good job of getting people to the polls, which suggests that the Obama campaign to target these would-be voters with a GOTV message. Traditionally, this is the domain of the campaign’s ground operation, but is that still the best or only thing to rely on?

The demographic make-up of the “unlikely” supporters makes Twitter a potentially attractive venue in which to build a new kind of targeted messaging strategy. Looking at the most recent available data from the Pew Internet & American Life Project (from August 2011, unfortunately), the familiar Twitter usage pattern emerges. Overall usage is about 9.3% of the population — that’s the entire population, not just Internet users, which is the denominator Pew generally uses in their reports. But there’s a clear racial difference, with blacks using Twitter at an 18.4% rate and non-blacks at 8.0%, and an even stronger age relationship. Of those in the 18-24 group (who were mostly not old enough to vote in 2008, keep in mind), 14.4% use Twitter; for 25-34-year-olds, it’s 16.4%, and after that it drops off considerably.

But the big finding for a social messaging campaign is that race and age interact — blacks in those youngest two age groups use Twitter at rates of 33.3% and 22.6%, respectively. Not surprisingly, the division between urban/suburban and rural somewhat mirrors the racial division. The two youngest age groups respectively have 20.3% and 17.2% usage among urban and suburban residents, but just 8.3% and 10.2% among rural residents. Compare this all with Facebook, which sees no race or community type relationships — only age — across its 60% overall usage rate.

The greater prevalence of Twitter in cities suggests something of a geographic bias to its structure, and that’s something that could be leveraged in a GOTV campaign — after all, the “get out” outcome is a pretty key part of such an operation. But its social messaging structure would also be key, allowing the campaign to create starting point messages that could be retweeted and shared throughout a trust-driven social network. Twitter’s differences from Facebook come into play here, as well. Because Twitter is just a message-sharing platform (as opposed to a game/calendar/affinity/etc. platform), it could encourage the spreading of GOTV messages, particularly to individuals who might use it primarily for non-political purposes (e.g., following celebrities or talking with friends).

These messages would be received not by a representative part of the population, but disproportionately by young city-dwellers, meaning they could be targeted with that audience in mind. In fact, once the network is built up, individuals could potentially be targeted with messages tailored either for themselves or to be spread out to their networks. In a way, this approach is the opposite of what they did in 2008 with the My.BarackObama.com site, which facilitated connections within the context of the campaign, and only between those already committed enough to come in and sign up for another social network site. Targeting existing Twitter users and their extended, potentially local networks would bring the campaign to the doorstep of the loosely affiliated voter.

Filed: We R in Control || 8:38, August 21 || No Comments »


Ground and pound

Interesting fact I learned while researching this post: West Virginia is less black than Wisconsin. (Photo by Matt Dell)

I spent the summer busy with travel and research, and let the blog go on hiatus. Well, fall classes start today, so now’s a great time to get back to procrastinating thinking out loud.

USA Today and Suffolk University released an interesting survey of “unlikely” voters a few days ago, which has gotten quite a bit of attention. It’s worth noting that their methodology is unclear in the report — they’ve got a sample of 800, but was that simply culled from a larger, general sample that also included some number of likely voters? At any rate, the big takeaway is that unlikely voters strongly favor Barack Obama over Mitt Romney, though about one-fifth say they would vote for a third-party candidate if they were to vote. This is perhaps the strongest evidence yet for the importance of turnout to Obama’s chances. In 2008, he got much bigger than usual turnout from traditionally low-turnout groups, most notably young adults and African-Americans. Many of these voters are probably still turning up as “unlikely” in typical polling models.

But a deeper look inside the data shows just where the Obama ground campaign is needed the most: convincing overconfident supporters that their vote is needed. Three-quarters of “unlikely” black voters expect Obama to win, but almost all say they would vote if they felt it could swing the election. Now, this probably shouldn’t be taken too literally, since no one really expects the election to come down to one vote, but it is worth thinking about how and where energized black turnout to have the most impact. Of the ten states with the highest proportion of African-Americans, six are solid red, two are solid blue, and North Carolina and Virginia are up for grabs. These were big Obama takeaways in 2008, in part because of black turnout. Florida is number 11 on that list, and carrying those three swing states would basically clinch the election.

The assumption in 2008 was that the Obama campaign had a great turnout operation, but I’m not sure if anyone knows whether that’s true or not — his candidacy certainly energized a lot of black voters on its own, but that doesn’t mean the campaign necessarily did anything extra-special to get them out to vote. If the campaign has data similar to what this poll shows, they would be well served by targeting black voters now, not with persuasive appeals that are largely unnecessary, but with sobering reminders that the election isn’t over until it’s over.

Filed: We R in Control || 10:12, August 20 || No Comments »