Tuesday, October 6, 2015

The Facebook Barometer

Facebook and USA Today has launched the Facebook Candidate Barometer. This is a measure of how often a candidate is mentioned, liked, or otherwise interacted with on Facebook. Here's a look at Trump vs. Hillary (the two front-runners):

As you can see, Trump is getting more than 2x the Facebook chatter than Hillary is.

What Does This Mean?

Right now, frankly, as an outside observer, it's hard to say. Firstly the numbers don't seem to distinguish between positive and negative interaction (Twitter analysis has had a really hard time with that--and they only have 140 characters to deal with). Secondly, the population of Facebook is not the same as the population of the US. So it's not quite like "a poll."

Do people who "talk politics on social media" get out and vote more? Not really. A meta-analysis of several studies of social media engagement and voting behavior (or participating in campaigns) didn't find much:
The metadata suggest social media has a minimal impact on participation in election campaigns. Popular discourse has focused on the use of social media by the Obama campaigns (Carr, 2008; Lohr, 2012). While these campaigns may have revolutionized aspects of election campaigning online, such as gathering donations, the metadata provide little evidence that the social media aspects of the campaigns were successful in changing people’s levels of participation. In other words, the greater use of social media did not affect people’s likelihood of voting or participating in the campaign.
Of course that's measuring actual GetOutTheVote efforts by campaigns. What if it's your actual friends who are interacting with you? A Facebook study involving 61 million people showed the following:

  • Facebook social messaging turned out 60k voters directly and 280k through "contagion."
  • Strong friends were 4x as powerful as weak relationships.
  • The net-result, however, was very minimal: 60k voters out of millions wasn't that notable. While social media does, in fact, produce a turn-out improvement, it isn't especially significant.
So this whole thing is kind of bunk, right?

Maaaaaaaaybe. While voting may not be a strong effect of social media engagement, there is reason to think protesting might be:
Based on a random sample of the Chilean population, he finds that social media users, measured in terms of frequency of use of four different platforms, were 11 times more likely to engage in a street demonstration or march, compared to non-users
It should be noted that the rest of the meta-analysis was not this conclusive--but that may be due to the sampling technique used.

What if Trump is a Protest Candidate?

What if Trump, drawing more than double what Hillary is, is a protest candidate. In other words, what if social media is good for some types of candidates (non-establishment) and not so much others. How would we tell? Well, the top three protest candidates on the right are Carson, Fiorina, and Trump. On the left, it's Sanders.

The top four people on the social media barometer in the Facebook index are Trump, Sanders, Fiorina, and Carson. In that order. Is it possible that social media activity is fueling their rise moreso than, say, Rubio? The Omnivore thinks it's entirely possible.

No comments:

Post a Comment