In Search of Massive Data

I hate the answer "because it's too difficult." I like a good challenge, and I loathe admitting that a task is beyond me. In research, however, sometimes one must do just that. On the tenure track, you have only so much time to accumulate a record that demonstrates that you are going to have a good career. Accordingly, longitudinal data are pretty much impossible. There is no way to collect panel data, for example, on teen-agers' media use over time. If you collect data for 5 years, they will not be ready when you go up for tenure.

As part of the strategic communications focus area, I spent a lot of time talking to students about advertising, public relations, and marketing. They want to know what works and why. All too often I don't like the answers that I can provide them. You see, I know that attitudes are only mildly predictive of behavior. The best predictor of behavior is behavior. But I have neither the time nor the money to go around measuring what people do and what they buy. Even more problematic is correlating marketing communications with behavior.

So we bring people into the lab and measure their attitudes. Unlike much research in strategic communications, we do measure behavior. We record psychophysiological responses, and we measure memory, which is behavioral in nature. But we do not know what our experimental participants really buy. And more importantly, we have almost no mechanism for tracking behavior after exposure to an experimental condition.

Every once in a while, I see something that causes the nerd in me to smile. In the Dec. 5, 2005, edition of Advertising Age, there is a story about advertising agency DDB collecting data on "signs" to identify consumer trends (p. 6). In the piece, DDB's Stacey Grier is quoted as saying, "Behavior tells more about motivations than attitudes do." I agree! And the exciting part is that this project is grander in scale than anything most academics could presently undertake.

DDB is turning to its rank-and-file employees for cultural indicators. They are looking for insights into the human condition, which they call "signs." They call the model DDB SignBank, and it was developed by DDB sociologist Eva Steensig. The project involves asking everyone from receptionists to advertising professionals to scan their lives for signs of cultural change (i.e., signs) and record them. So far, they have a database of 30,000 signs. Data on this massive scale allow for insights not possible in smaller, more tenure-friendly studies.

Obviously there are drawbacks to this method. Sadly, the model is highly proprietary, so academic scholars such as myself cannot comment on its rigor. Nonetheless, the idea is yet another piece in the mental puzzle. And, for the sake of circularity, observations on Weblogs are one of the signs.