Login or signup
36
STRATEGY

Nate Silver: 5 Data Rules to Live By

In a recent talk, the statistician highlighted many of the ways in which people go wrong with data. Entrepreneurs, are you listening?

Advertisement

Nate Silver, the former predictor of politics and other things for the New York Times and now sports guy for ESPN, doesn't think of himself as a statistician, he told a recent Montreal gathering of more than 1,000 statisticians. As big data predictive blogger Joseph Rickert reported, Silver talked about the connections between statistics and journalism and offered principles for journalists to better use and understand statistics.

That might not sound particularly interesting if you're not in the media business, but don't tune out. What Silver had to say to journalists is just as important for entrepreneurs to hear. Data can be tremendously useful. You can get insight into markets, analyze trends, or better grasp your own product quality control. But that's only if you approach the data in the right way.

Silver offered 11 points for journalists. Here are five that are particularly useful when looking at numbers in a business context:

1. Correlation is not causation.

You should stencil this onto the wall of your office. People with almost unremitting consistency assume that if two things happen at the same time, one causes the other. You see that sales are up and that you came out a variation on a product, so you've caused the increase. That may be the case. But it might be that a major competitor had production issues and that yours became the default choice. Maybe it was the design, but what drew customers was not what you think. Or, and this is what people really don't like to hear, the whole thing could have been a statistical fluke. Don't assume that you know why two things happen simultaneously unless you can actually prove any connection.

2. Averages can be useful, if you know their limitations.

Silver actually suggested that journalists pay attention to averages because they often pay attention to outliers--extremes can make good stories. But averages can also be wildly misleading. If nine people have $1 each and a 10th person has $21, the average amount each has is $3. But, really, almost all of the people have only $1. Look at the median--the middle amount--for another view of how things stack up.

3. Be wary of intuition.

Going by your gut is sometimes the right thing to do. Often, it will take you in the wrong direction because we have two types of thinking. One is fast, emotional, and often has many biases. The other is slower and more deliberative. Make sure you use the right type when looking at data. Silver suggested the book "Thinking, Fast and Slow" by psychologist Daniel Kahneman who won a Nobel prize in economics for his work.

4. Look for the truth.

Too often, people use statistics in an attempt to justify a decision or to create a particular appearance. That is a waste of time because you've just tossed any value you might have gleaned from the data. Look to understand what is going on and then fit your decisions to reality, rather than trying to bend any reality of data to your preferred view of the world.

5. Make predictions.

When you look at data and predict based on it, you have made a verifiable statement. Maybe you'll be right; maybe not. But as you work with information and see where you come short and where you are on the money, chances are that your thinking and strategic planning will improve.

Do you need to become a statistician? No. But by understanding some basic statistical concepts, you can start making more sense and better use of the information you get.

IMAGE: woraput chawalitphon/Getty
Last updated: Aug 21, 2013

ERIK SHERMAN's work has appeared in such publications as The Wall Street Journal, The New York Times Magazine, and Fortune. He also blogs for CBS MoneyWatch.
@ErikSherman




Register on Inc.com today to get full access to:
All articles  |  Magazine archives | Comment and share features
EMAIL
PASSWORD
EMAIL
FIRST NAME
LAST NAME
EMAIL
PASSWORD

Or sign up using: