Prosecutors in the 1995 O.J. Simpson trial maintained that because O.J. had frequently abused his wife Nicole in the past it was likely that he was the one who murdered her. But during the trial the defense easily overcame this argument by telling the jury that of the 4 million women battered by their husbands the previous year, only 1 in 2500 was actually murdered.
The prosecutors never refuted the defense’s argument, and eventually O.J. was acquitted.
Despite the fact that this use of statistics was seriously flawed--on both sides--none of the hundreds of reporters and others following every little twist and turn in the case raised so much as a hint of concern.
In this election season almost 20 years later, the polls are alive with the sound of confusion. Obama by six, no it’s Romney by three, it’s a dead heat in the battleground states, it’s well within the margin of error. Computers are churning out reams of numbers, boatloads of figures that march toward us in neat, impenetrable rows. And in the big-data business future of numbers, algorithms and evidence-based management, no business person will be able to succeed for long without having a much better grasp of statistical reasoning than either O.J.’s defense attorneys or the prosecutors themselves had. It will be quite a change from business as usual.
Used to be that only direct marketing executives ever showed much interest in statistics. They relied on detailed analyses of response rates to evaluate various lists of potential direct mail recipients, and they regularly calculated profitability in terms of likelihoods and confidence intervals. Most business managers outside the direct marketing industry, on the other hand, could work their whole careers without ever encountering a standard deviation or a false correlation. Business decisions used numbers and facts, certainly, but most decisions about the future were still…guesses about the future. Educated guesses, yes, but guesses nonetheless. And you don’t need statistical reasoning to decide whether a business goal is achievable, or a new market is ripe for entering.
Today, by contrast, everyone is a direct marketer. Every business with a website or an email marketing capability must be manned by managers capable of thinking clearly in statistical terms and making decisions based on statistical reasoning. Doing an A/B test (or several) to evaluate an offer to a particular subset of website visitors or prospective customers, or understanding the assumptions underlying the rate of attrition predicted by a customer analytics program--these are the kinds of issues business executives must deal with on a daily basis now.
But take heart, because as business managers become better and better educated with respect to statistics and statistical reasoning, our society will probably make better and better decisions about everything, not just about the likelihood of completing a project under budget or the most efficient way to enter a market, but also about the genuine risk involved in a child’s vaccination or the likely range of costs from global warming. Perhaps, also, people will become better lawyers and jurors.
And as for the O.J. Simpson trial? Regardless of your own opinion about the actual verdict, the outcome would likely have been different had the prosecutors been statistically educated enough to point out a different and more relevant figure: of all the women who were battered by their husbands in the previous year and also murdered, 90% were in fact murdered by their abusive husbands.
[Image: Flickr user Domenico]