Welcome to the Listening Post!

Under the Influence of Influence

Tweet

 

This week's Time magazine is especially vapid: "The 100 Most Influential People in the World", with many dubious choices. Neither recent fame (Scott Brown, Lady Gaga) nor past celebrity (Elton John, Prince) translates into influence. Much more enjoyable was columnist Joel Stein's parody of the whole effort with his list of the 100 least influential people.

To add insult to injury, one of my favorite statisticians, Nate Silver of FiveThirtyEight.com (a polling aggregation and modeling site), chimed in with a superficial analysis of the Time 100's social network influence, by averaging their Facebook connections and 2x their Twitter followers.
This is shallow, since follower counts don't equal influence, and we live in an age of experimenting about how to calculate influence, at least for social networks:

• Twitalyzer defines influence as "the likelihood that a Twitter user will either A) retweet something the user [being measured] has written or B) reference the user".
• Twinfluence goes into a rich discussion of first- and second-order networks, "social capital", centralization, velocity and efficiency. It's not just who you influence, but who they can influence.
• The PR agency Waggener Edstrom has a proprietary Twitter tool and an influence manifesto: "Influence is evolving - to the point of what you could call a ‘communications cataclysm'. Waggener Edstrom Worldwide believes there are three core tenets that can help drive understanding of influence: 1) Content is King; 2) Understand the audience; 3) Engagement is nuance. Our ability to understand and apply these three tenets for our clients ultimately helps them to solve their business problems, enabling them to address new markets and seize the opportunities that exist."
• Edelman's TweetLevel is calculated by the formula Rg=>(Fo+Fg+Up+@U+Rt+Ta+TaN:S+Ti+Tg+Ii+Vi)/Z x w(i/p/e/t). Think of it as an aggregation of other tools' measure of influence.

From a marketing research standpoint, influence is important for tracking and monitoring word of mouth - and "word of mouse" - activity. It's no coincidence that PR agencies are so interested in this measure, as are brand monitoring firms. They view influencers as potential propagators of their brand messages.

I was rightly criticized for ignoring influence in my recent post "Social Media Research is Not the Same as Verbatim Coding", as it is an important differentiator from traditional survey research analysis. I had talked about the need during data cleaning to factor out "retweets" (rebroadcasts of other people's Twitter messages, either "as is" or edited, with or without commentary), when it fact retweets are a sign of influence. When I presented the top reasons people return iPads, for instance, had I settled on an influence formula I could have calculated the top reasons cited by influence: perceptions of why people return iPads, as it were.

Influence itself can be a seductively influential idea, but often in market research a good idea is a good idea whether one person with no influence tells you about it (see Don't Underestimate the Power of n=1) or Ashton Kutcher does (@aplusk, 4.8 million Twitter followers). And, as even Time magazine points out, someone can have zero measurable social networking influence, as 25 of its Time 100 do, and yet be among the most influential people in the world.