It is Safe to Say that many readers of this column will Not Escape the Effects of this Terror! Guided by a Master Plan for Complete Domination! It Haunts the Earth in a Terrifying revelation of Things to Come!
It walks Among Us and it Must Be Stopped!
It’s true. With apologies to the trailer for Planet 9 From Outer Space, Big Data has become rather a nuisance. It’s the latest tech, science, and business buzzword that is simply an intimidating way to say “lots of data.” Intimidating, that is, for those who don’t deal well with numbers, which is apparently most of us. It’s not that we’re bad at math (though quite a few of us are) but that we make big mistakes when it comes to big analysis. We lack number sense, according to Kaiser Fung, author of a new book with that name. The message of Number Sense is similar to Fung’s previous book, Numbers Rule Your World. He not only shows how prevalent numbers are in everyday life, but also that we misread data and are prone to faulty analysis. Our lack of number sense leads to many bad, and often expensive, decisions.
Reputed smart person, Bill Gates, succumbed to Big Data. Relying on lots of numbers, he poured millions of dollars into the conviction that small schools turn out better pupils. Ten years (and lots of big money) later, the Gates Foundation “made an about turn.” While small schools account for 12 percent of the top 50 schools, they also represent 18 percent of the bottom 50.
It turns out that small schools “were overrepresented at both ends of the distribution.” Other factors, discounted or ignored in the analysis, had a greater influence. “Data analysis,” Fung notes, “is a tricky business.”
It’s tricky when we don’t differentiate between conclusions with strong intuitive appeal and those emerging from properly analyzed data. Strong intuitions usually win. (After all, shouldn’t smaller schools and classes translate into better test scores?) In addition, we often ask the wrong questions or analyze the wrong things.
Fung, for example, asks the urgent question, “Why can’t I fit into those jeans?” not realizing, of course, that some of us have given up jeans altogether. His point, however, is that weight gain solutions are influenced by mistakes in analysis. The commonly used measure, BMI (body-mass index), categorizes many Americans (and, increasingly, people worldwide) as obese. BMI, however, is inaccurate. It doesn’t differentiate between fat and muscle. A more accurate (and expensive) measure isn’t the answer, though – it actually shows a much higher incidence of obesity. But, no one dies of obesity. They die of diseases associated with obesity. That might seem a fine point until you realize that we try to cure obesity with remedies that have a low success rate (dieting and exercise) instead of addressing the real problems.
Big Data also has implications for your business. “Online marketers worship the clickstream as the Holy Grail,” writes Fung. They assert predictions from large aggregations of data. What they measure: such things as age, college degree, and place of residence (city or suburb), has little relevance, however, in the cause-effect conclusions they make. For example, Fung notes that even if most people who buy his book live in cities, living in a city isn’t why they bought the book. Book-buying, instead, has to do with things that aren’t measurable or fall outside of typically collected data. You bought it on impulse, from peer pressure, or because your boss required it (or because you read this review!).
But, doesn’t that still mean that you should target people in cities? The answer to that question, and many other data analysis problems, is a worthwhile read. If you’re trying to maximize the effects of marketing dollars, it pays to know what you should be targeting. Fung’s critique of Groupon is especially revealing on why number sense can really pay off.
Even though Big Data walks among us, we can fight back, as any fan of apocalyptic film and TV can attest. If enough people with the right know-how put their minds to it, even the biggest monsters fall. The numbers are against them. FBN