If there is a big lesson from the last presidential election, it is about the value of having good data. Once again, Nate Silver and his FiveThirtyEight blog made a spot-on prediction of the election results. His final map not only predicted that President Obama would win the electoral college, but it was stunningly accurate predicting the margin of victory for each state as well as the overall popular vote.
In the days leading up to the election, most of the news media were focused on how close the election was. There were countless stories about scenarios in which one candidate would win the electoral vote and another would win the popular vote. Pundits talked about momentum, enthusiasm, and reasons why the poll numbers did not tell the full story of the election.
Despite all of the stories, the numbers told the right story all along. The numbers said that the election was not that close, and it was not.
There is an important lesson here for business.
As humans, we love stories. A good anecdote trumps a good graph in a conversation or speech. Our hearts are swayed by managers and CEOs who can spin a good narrative for why a product will succeed or why a business is about to turn around. When we look into our crystal ball to predict the future, though, there is no substitute for good data.
What made Nate Silver and his predictions so accurate was not that there were numbers involved, it was his understanding of the numbers and his willingness to find good data. That is a skill, and one that is under-represented in the business community.
Several years ago, I was working with a group at a large company. By chance, the group had to stop a meeting to hear a presentation about the results of a study that the group had commissioned, and I was invited to tag along. The presenter, who came from a consultancy that had been hired to run the study, began to present the data. It was fairly clear after three slides that something was wrong. Differences between groups that they claimed were statistically reliable could not possibly have been different given the amount of variability in the observations. Something in the analysis was clearly flawed.
Yet, nobody else in the room recognized the problem. There was nobody there who had enough training to look at the graphs that were presented to find the inconsistencies. After asking a series of questions of the presenter, it was clear that there were errors in the way the data were analyzed. The consultants had to go back and re-crunch the numbers.
I worry that if I had not been there, significant decisions would have been made based on flawed analyses. And more generally, I know that there are many people who make decisions based on data that they do not completely understand.
If you are going to succeed in a leadership role in a company, you need to become a good consumer of data. To be a good consumer, you have to break down and really learn some statistics. Run an analysis on your own. Learn to speak statistics like a native. In the end, you can use good stories to convince yourself of anything, but really good data provides a necessary reality check.