Wednesday, November 18, 2009

The Tipping Point

I know I am late to the party, since this popular book by Michael Gladwell was published almost a decade ago. The book attempts to present a synthesis of many disparate ideas to ponder over the question: "Why do some ideas catch fire?", or as he would probably like it phrased "What makes an idea tip?"

The book itself is enjoyable, as it talks about Paul Revere's midnight ride, the Mavens, the Connectors, and the Salesmen, the fascinating rule of 150, Bernie Goetz and how cleaning up graffiti on subway walls reduced crime in New York, Hush Puppies, the stickiness of Sesame Street and Blue's Clues, Peter Jennings demeanor during Ronald Reagan's candidacy, six degrees of separation etc.

He cites a number of interesting social and psychological research studies, and being the master storyteller that he is, beautifully integrates them into his narrative.

I think he makes a great journalist.

However, I think he would make a bad scientist.

This is pure extrapolation from the one data point I am familiar with. The six degrees of separation reference to Stanley Milgram is bad science. He repeats some of the same stuff in his famous article "Six Degress of Lois Weisberg".

The myth suggests that Milgram gave 160 people in Omaha, Nebraska a package that had to be delivered to a stockbroker who worked in Boston, through the smallest number of intermediaries. He found that "chains varied from two to 10 intermediate acquaintances, with the median at five" in his 1967 paper - which apparently is the basis for the "six degrees" supposition. The big problem for me as a scientist was that only 24 of the original 160 chains was completed - and hence the conclusion probably suffers from a heavy survivorship bias.

Milgram carried out an earlier study where starters were from Wichita, Kansas and were supposed to reach a divinity student on the east coast, and the completion statistics there were more miserable. The measurement error must have been quite large to suggest such a strong conclusion.

Sure, we might indeed be separated by six degrees. But Milgram's study does not definitively prove it.

In fact there are other glaring problems with Milgram's study as this very interesting and more academically rigorous article points out.

PS: I swear I wrote this blog a long time ago, and thought that I would publish it later. In the meantime, I bumped into this article by Steve Pinker (via nanopolitan). It is amazing that he comes to the same conclusion towards the end of his book-review:
Readers have much to learn from Gladwell the journalist and essayist. But when it comes to Gladwell the social scientist, they should watch out for those igon values.

No comments: