The Early Adopter Effect
by Patrick Ruffini :: October 26th, 2007 2:28 amTo work in the online political world is to be swimming in a sea of data — but to be without the kind of real, hard data that pollsters and strategists rely on to drive the overall campaign.
What do I mean by this? TechPresident has plenty of data. They have MySpace charts, Hitwise traffic, Evenful demands. Well yeah. But we all know by now that YouTube views doesn’t equal votes. And we really don’t know the demographic that are driving supporters of certain candidates to be more active. Why do Democrats seem to give money online at 2 or 3 times the rate of Republicans? People try to argue stuff like, Democrats have an edge with young voters. But that’s deeply unsatisfying. Young people aren’t 2 to 3 times more Democratic than older people. They’re maybe 10 points moreso. And even so, Ron Paul supporters show how dedicated minorities can leapfrog a far more numerous majority.
As a practitioner of online politics for almost a decade, I can tell you that we are flying blind as far as real demographic and survey data goes. The demographic data that comes from panels like ComScore, Nielsen, and Compete was designed to be usable for large eCommerce sites with millions of unique visitors a month. I wouldn’t trust their demographic breakdowns as far as I could throw them for a Presidential campaign site that gets hundreds of thousands of uniques a month.
Now, the campaigns themselves should have a pretty good handle on who their people are — if they take the step of matching their supporter list to the voter file. But there is no comparative data with rival campaigns. Online polling — the scientific kind — is still in its infancy. You can’t field a poll and ask who’s going to Obama’s site vs. who’s going to Clinton’s site cost effectively. You can’t get the answers to the questions that ComScore or Nielsen won’t give you.
Facebook is on the verge of showing us a better way. Through the pro version of their Flyers self-serve advertising platform, they’ve given us a way to grab precise demographic counts on their membership, including ideology. These aren’t samples, but their entire universe. This was meant to be used for ad targeting, but it is also a powerful data mining tool. I was first turned on to this by Stephen Taylor and iStrategyLabs.
Out of idle curiosity, I started running an ideological breakdown of Facebook users by age, starting at Facebook’s minimum age of 14 and working my way up. The spreadsheet is here so you can follow along.
It was after I started reached the mid-20s that I stumbled upon something that may help quantify the early adopter bias. High school and college users were pretty consistently about 4-8 points more liberal than conservative. That’s sort of where you’d expect them to be given the 18-29 year old vote. And Facebook’s market penetration with this cohort is such that this is likely to be a highly representative sample of Americans that age.
But the older you got, through users in their 20s, the more liberal the user base became. It was inexorable. Each year, liberals picked up a couple of points on conservatives. My fellow 29-year olds on Facebook are +25.3% liberal. The 20-year old bracket is +4.5% liberal.
Given how stable the numbers were for college/high school users, with much higher numbers, this seemed unlikely to suggest an actual demographic shift in Generation Y.
But something else was going on. As liberals were picking up steam, the number of Facebook users were getting progressively smaller with each age cohort.
It makes perfect sense when you think about it. These users represent early adopters who never used Facebook in college. The people who joined Facebook since it opened up, or finagled a way into it before then using a stray alumni .edu address like I did.
This is pretty strong evidence of a liberal/early adopter correlation. Non-college Facebook users in their late twenties are two to one liberal where their college age counterparts are pretty closely matched.
That two-to-one ratio probably correlates with usage of other high-end web services and even traffic to the candidate sites themselves. It also gives quantifiable backing to the idea that Republicans stand to gain as the universe is widened entering the general election, as I’ve long suspected.
This is why it’s idiotic to extrapolate popularity from online metrics. If it’s (still) mainly the early adopters using the tools, then those early adopters can look dramatically different than the voting population without it portending doom.
Most campaign sites are probably getting visitorship in the tens of thousands of visitors per day, if that. That’s still within the early adopter universe. As politics online becomes more mainstream, the Democrats’ potential for growth is considerably constrained. Actual online engagement among people who are fully comfortable with the medium (the Millenials) is no worse than the D/R split in voting. That’s still a problem for Republicans given our challenging numbers with 18-29 voters, but the problem then becomes merged with the electoral one rather than being compounded by online-specific trends. As the popularity of the tools grows and the Millenials go mainstream, the 2-to-1 split Democrats have counted on could be a thing of the past.
There are a couple of other interesting angles to this that I’d like to explore.
First, gender. The gender gap among the (again, highly representative) college age cohort is huge. Men are +2.3% conservative. Women are +13.7% liberal, a 16-point spread. Women are also 54.5% of the self-identifying universe (the guys are probably off playing video games).
That holds with Facebookers aged 23 to 29. You see the same 16 point gender gap, but the numbers are more liberal. The actual representation of men and women narrows somewhat, with women at 51.4% of the universe.
With the over-30 crowd, we start to see a nearly 50-50 spread of men and women. But the gender gap narrows dramatically, with a +8% liberal bias in women 30 to 34, a +1.4% bias in women 35 to 39, and women over 40 on Facebook are +7.4% more conservative than their male counterparts (I’ll concede that this is a fairly small universe — just 330,000 Facebookers over 40).
Having been around politics for a little bit, I can tell you this is not reflective of overall trends. The gender gap still exists with 30- and 40-something voters. But the fact that it disappears on Facebook is reflective of something else — the early adopter gap.
Hypothesizing off the stereotype that early adopters tend to be male, interested in technology, and liberal, you start to see how this plays out. (Look at the blogosphere. It skews heavily male.) In the age groups with fewer Facebook users, early adopting men displace general interest men who are likely to be 8 or 10 points more conservative than their female counterparts.
Another oddity in the data is that the pro-liberal trend doesn’t continue forever. In the early thirties, Facebook users start getting more conservative despite some being uber-early adopters. But that doesn’t invalidate the thesis. Remember that most polls show that conservatives enjoy a 3 to 2 edge over liberals in the general electorate, so we have to start picking up somewhere. What we can draw from the Facebook data is that the future electorate is less moderate (28% vs. 49% among actual voters) and that liberal=Democrat and conservative=Republican to a far greater degree than in the current electorate.
Naturally, there is only limited data when it comes to voters outside the 18-29 age group. Only 1.6 million Americans older than me are even on Facebook. But I’ll be enjoying watching the Facebook Flyers data set grow, and will run another set of numbers in a few weeks to see my thesis of later adopters being more conservative is actually playing out.
For Republicans, this gives us quantifiable data for the first time showing where to concentrate our online efforts. Our comparative advantage is not in dominating the existing social networks, but to do things that expand the universe, and to get people on the edges of the process to engage (it’s not all that different than what Hillary has done to transcend the netroots). This sounds like liberal election strategy, but for our purposes, it forms the outlines of a conservative insurgent strategy to retake the Web.
Tagged: facebook 
marketing online politics strategy
Comments (1) 
Trackbacks (7) 
del.icio.us 
digg it 
subscribe
Both comments and pings are currently closed.
I really appreciate the manner in which you have digested the data. As far as partisan politics goes, it seems to be particularly easy to manipulate statistics in a way that presents one’s side to be stronger/weaker. Speaking in terms of quantifiable positions, I feel like this manifests itself in what I generally view as nothing short of propaganda and/or rhetoric. This sort of data-focused breakdown is just what the Republican party (and the Dems for that matter) need to view to coordinate a proper reach-out strategy.
What do you think about a social networking site like Essembly? It seems like this type of network–that is devoted exclusively to politics and activism–would be a great medium to really execute on the numbers you present.
[…] I spotted Patrick Ruffini referencing a post about Facebook demographics in his twitter feed today and was surprised to see the detail they had on a chosen set of demographics. Turns out, they were able to compile those demographics via the interface on the flyer creation page on Facebook. […]
[…] Patrick Ruffini put an intriguing blog post on The Early Adopter Effect.Here’s a quick excerpt:Women are +13.7% liberal, a 16-point spread. Women are also 54.5% of the self-identifying universe (the guys are probably off playing video games). That holds with Facebookers aged 23 to 29. You see the same 16 point gender gap, … […]
[…] Patrick Ruffini gives us one more reason to take a serious look at Facebook, as he explains the Early-Adopter Effect, and demonstrates that Facebook stats show that conservatives may be catching up to liberals in the online social networking scene. This is exactly what FHK is all about - helping conservatives become more effective in the Web 2.0 environment. […]
[…] This is all part of my deep dive into Facebook demographic data provided at by the Flyers Pro advertising engine. I’ve just updated the spreadsheet. (See part I of this series for more.) The data looks like this: […]
[…] Patrick Ruffini :: The Early Adopter Effect (tags: facebook data demographics early adopters granular) […]
[…] And Matt Stoller’s response to a separate piece by Patrick about using Facebook as a microtargeting data tool. […]
[…] At the moment, however, it seems more likely that he will get stuck in Early Adopter-adopter Quarantine. Early Adopter Quarantine (EAQ) is when an idea or product cannot escape its initial cult following and make it into the mainstream. It’s happened to Linux, Apple and a whole slew of brilliant products, companies, artists and ideas you have never heard of. What’s important is why it happens. Why do great ideas do so well initially and then get stuck in the eternal purgatory of always being just on cusp of breaking out? Could it be that the early adopters themselves are to blame? […]