Social Media Slant: Comparing Polls to Web 2.0 Coverage

Editors Note: Before looking at the numbers, it is important to understand that the uniqueness of Obama’s and Kucinich’s last names, and commonality of Clinton’s and Edwards’s, could skew the results even more. In all likelihood, both Clinton’s and Edward’s coverage differentials are even lower, but appear more reasonable because other individuals with the name Edwards or Clinton make the social media sites.

	Digg	Reddit	Delicious	Propeller	Polling	Verdict
Obama	+4.6%	-3.2%	+10.9%	+6.1%	30.3%	+4.6%
Clinton	-12.8%	-3.6%	+0.2%	+14.9	36.6%	-.3%
Edwards	-9.4%	-6.9%	-1.6%	-8.3%	17.9%	-6.5%
Kucinich	+30.9%	+26.6%	+7%	—	2.6%	+15.9%

Analysis & Implications

The clear loser is John Edwards. Despite the fact that he has polled consistently between 15% and 20%, the same media coverage blackout mentioned on DailyKos appears to have made its way to the social media scene.

The clear winner is Dennis Kucinich, whose mention-rate on web 2.0 sites is 6x his actual polling results.

However, I think the biggest winner is actually Barack Obama. On the most highly visited sites, Obama’s differentials made him the most commonly mentioned and discussed candidate.

One interesting thing to look at are the web 2.0 sites themselves. It appears that Propeller is most representative of traditional media, pushing up the “frontrunners” to the exclusion of the other candidates. Reddit and Digg appear similarly infatuated with Dennis Kucinich and, perhaps because of Reddit’s (warning, shameless stereotype) willingness to espouse strong, unpopular opinions, they under-represent all of the major candidates, while Digg over-represents Obama.

Methods

Admittedly, my methods were crude. This was not meant as any form of comprehensive study, just a quick question I had that I wanted to answer for myself in as satisfactory a manner as practical – meaning under 30 minutes. Essentially, I ran search queries via Google for each of the occurrences of the candidate’s last names on each of the web 2.0 sites in question. Where possible, I tried to control for unique occurrences (for example, trying to count only the article pages, and not the “who-dugg” pages, because popular stories could create hundreds of results for the same article in Google). I then compared these numbers to create a % of stories about each candidate on each site, and determined the differential between that and the average of the latest 5 national polls I found. I only looked at Democrats still in the race with whom I was familiar.

No tags for this post.