Advertising Effectiveness in Voting

There’s a phenomenal NYTimes Magazine article today that blends two of my favorite topics: behavioral science and advertising effectiveness, and comes complete with shoutouts to two of my favorite scientists, Robert Cialdini and Richard Thaler.  Yes, I said “my favorite scientists,” and no, I don’t apologize for geekiness.

Cialdini’s Influence was taught in business school, and I found it a truly transformative book.  Thaler uses a “freakonomics” style approach in fascinating ways, including detecting suboptimal behavior in “Deal or No Deal” contestants (a situation that is eminently “solvable” using expected value theory) and finding the pitfalls in the little mental accounting we all do every day.

Some highlights from the article:

  • A sentiment very similar to those we’ve found in clients who are just starting an analytics program:
    "...a worldwide battle for knowledge between two camps he
    thought of as “gurus” versus “data.”"
  • Another parallel in our little corner of the world- firms that have been winning in their history see no need to do anything differently:
    "...everything you did in a winning campaign was a good idea and everything that
    you did in a losing campaign was a bad idea."
  • Happily, they found that those annoying telephone calls from candidates have no impact at all on voter participation.
  • I am not a fan of quantitative attitudinal research, and neither are they:
    "Focus groups offer a rich impression of how certain voters respond...
    but only the instant reaction of someone being paid $100 to have one."

But a big theme at the end brings it all back to observations.   Because one only sees the end result in the final vote, and opinion polls are not robust enough to serve as precise indicators of success, there simply aren’t enough observations to tease out the more subtle drivers like broadcast TV or the impression value of an internet ad.  While our work in business analytics is always challenging, at least we have the benefit of a foundation of extremely rich data upon which to work.