By David Forbes, Ph.D.
In many ways, the ever-expanding world of social media has been a market researcher’s dream—there are now hundreds of millions of willing subjects spending billions of hours telling us what they like and don’t like, and why. They share these thoughts—for free, no less!—in ways that marketers can quickly measure and quantify, whether through Facebook likes, Tweets, retweets, Instagram sharing or video reposting. Dozens of buzz meters can even analyze the content and track the tone of their comments.
No wonder social media has become such an irresistible barometer of brand sentiment.
But market researchers should not rush too quickly to believe everything they find in this goldmine, at least until they understand two flaws inherent in data: One is about the human ego, and the other all about sheep.
The Self-Presentation Bias
Recently, I was lucky enough to hear Valeria Brandini, a Brazilian anthropologist, give a lecture on just how unreliable social media can be. Whether it’s as simple as liking a brand page or as involved as inventing a Facebook identity for the family Labradoodle, Brandini emphasizes that people are pretty consistently mindful of the fact that everything they do in social media says something about who they are. As Brandini expressed, “people say what they want you to think they think, what you think they feel.
Because people are social creatures, they want people to think well of them. So posts aren’t just about self-expression, but are also about self presentation — meant to influence an audience to see the poster in a very specific way. While some of this content may be genuine, it is very difficult to tell the difference between an authentic brand sentiment and what part of a poster’s social window-dressing is.
The baah-baah bias
The second flaw in social media data has to do with what’s known as the herd effect, especially as documented by the intriguing research of network scientist Sinan Aral at the Massachusetts Institute of Technology. His work, involved seeding comment sections on news stories with remarks that are either cynical and pejorative, or accepting and appreciative. His results show that when their remarks are visible to others, people pile on strongly and articulately express sentiment like agreeable sheep. Because of this, the online universe might seem disproportionately in favor of a brand that is boosted by vocal activist enthusiasts – or overwhelmingly negative about a brand that suffers from similarly vocal and articulate detractors.
For marketers, that should mean taking all those “buzz measures” with a giant block of salt.
So how can brands best take the measure of social media, given those inherent distortions?
Given the inherent distortions we’ve outlined above, it’s probably most prudent to cross validate social media with other, more direct measures of brand attitude and brand affect. We have developed one approach to social media cross validation that mimics the “rapid fire” pace of the web experience, while gathering data from the consumers emotional brain in a way that is not susceptible to distortions from self-presentation or “herd like” crowd conformity. I encourage other marketers and market researchers to explore their own options for cross validating social media content. To mistake what consumers want us to think for what consumers actually think can prove a dangerous research course indeed.