Nate Silver addresses assembled statisticians at this year's JSM

by Joseph RIckert

Nate Silver of 538 fame gave the President's invited address this year at the Joint Statistical Meeting in Montreal.

Nate began his talk by pointing out that because of his efforts to bring statistical rigor to topics usually covered by journalists he is often portrayed as a kind of statistical journalist. Then, with considerable self-deprecating humor (you had to be there) Nate observed that although he does indeed work in the conceptual space at the intersection of journalism and statistics there are some who would not consider him to be a journalist, and he does not consider himself to be a statistician. It was very nice to see Nate, facing a room of more than 1,000 highly accomplished statisticians, begin with a gesture of humility. I admired him for that. There are many who riding the wave of euphoria in their moment of popularity who don’t have the common sense to make this kind of graceful gesture. Nate went on to structure his talk about the links between journalism and statistics by presenting 11 principles for journalists:

1. Statistics are not just numbers. Here Nate spoke a little about statistical inference.
2. Data requires context. Nate mentioned an article about the population of China that included the fact that China was the world’s second largest economy. Nate pointed out that while true, in the context of the article, it would have been more illuminating to quote a statistic about the per capita economic reality.
3. Correlation is not causation. Nate surmised that journalists are prone to fall into this trap because of their strong desire to tell the story about what they are reporting. It is just human nature to invent causes to connect the dots.
4. The average is still the most useful statistical tool. At first, I thought to disagree with Nate here. Journalists often write about the nonexistent average person, and who has not consulted a physician who seems to describe a treatment designed to cure some average patient. However, Nate was going after a more fundamental point here. Because they are looking for interesting stories, journalists often focus on the outliers.
5. Human intuition is often misleading. Nate recommended that all journalists read Daniel Kahneman’s Thinking Fast and Slow.
6. A probability forecast expresses uncertainty instead of trying to conceal it. Nate pointed out that journalists are very comfortable with the 50/50 chance or the all or nothing story, but find the 75% chance vs. 25% chance problematic. Nate’s advice was to try to tell the truth by communicating the uncertainty.
7. Know thy priors. Nate has embraced the Bayesian paradigm as the model for learning from data. He believes it would be very useful for journalists to recognize their own prior beliefs.
8. The word complex isn’t always a complement. If a journalists or, even worse, an "expert" avoids a clear explanation by saying something is complex Nate sees this as a flag that they just might not know what they are talking about.
9. “Insiderism” is the enemy of scientific objectivity. It is expensive to obtain inside information because it takes journalists a considerable amount of time and effort to cultivate the required relationships. In the end, the information may not, in fact, be useful. In most cases, the benefits of a little scientific objectivity outweigh what could be learned from a little inside info.
10. Making predictions improves accountability. Nate recounted a story in which he was rebuked by a New York Times editor for suggesting that a critic of one of his forecasts wager on the outcome. The editor was of the opinion that gambling in the newsroom is unseemly. Nate doesn’t get that logic. He thinks that if journalists would put their money and their reputations where their mouths are they might get closer to achieving scientific objectivity.
11. Like scientists, journalists ought to be more concerned with the truth rather than just appearances. He suggested that maybe they should abandon the legal paradigm of seeking an adversarial approach and behave more like scientists looking for the truth.

After this Nate was patient, witty and candid in dealing with what seemed like countless questions mostly posed by an audience tweeting to #JSM2013. He hit a home run with the crowd in his reply to the question “What do you think of data science vs. statistics?” Nate replied that he thought that data scientist was just a “sexed up” term for statistician. His advice: “Call yourselves whatever you want.”