Statistician Nate Silver correctly predicted all 50 electoral outcomes in the 2012 U.S. presidential election on his blog FiveThirtyEight. It’s quite simply the type of data success story that left many wondering — how did he do it?
Silver published “The Signal and the Noise: Why So Many Predictions Fail — But Some Don’t” shortly after the election. It contains many useful tips for marketers wrestling with new challenges in the era of big data. How can marketers emulate Silver’s level of conviction in their thinking and ideas? Here are three suggestions.
1. Ignore the Hedgehogs
Nate Silver defines a hedgehog as a “Type A personalities who believe in governing principles about the world that behave as though they were physical laws and undergird virtually every interaction in society.” Hedgehogs know one big thing — and it’s the solution to every problem.”
Silver suggests thinking more like a fox. Foxes “believe in a plethora of little ideas and in taking multitude of approaches toward a problem. They tend to be more tolerant of nuance, uncertainty, complexity and dissenting opinion.” Most innovations and new ideas are found in tiny places where others fail to look. Ignoring the hedgehogs and generally accepted thinking will afford opportunities to see familiar problems in new ways.
2. Change your statistical mindset.
Bayesian statistics supports the continual refinement of an idea. As we gather additional information, we incrementally clarify our position. If marketers plan on implementing against innovations and new ideas, they need to find a better way to express their likelihood of success. And traditional frequentist statistical methods simply fail to meet this criteria.
Silver argues the frequentist approach prevents marketers from considering underlying context or imparting prior experiences because it assumes a study could be tested to a perfect conclusion if only enough data were collected. Bayesian approaches demand a prior probability based on previous data or assumptions, which are then subsequently refined through experimentation. Silver notes this is important because it’s impossible to remove all bias.
3. Experiment with your ideas.
Silver notes “Companies that really get big data, like Google, aren’t spending a lot of time in model land. They’re running thousands of experiments every year and testing their ideas on real customers.”
Silver posits that most of this “new” data is just noise, but he also cautions it should not be ignored. He reminds us that the mistakes we make will not be measured in degrees, but rather result in errors in whole orders of magnitude. Silver says it is this risk that prevents us from acting on new ideas, but experimentation helps us separate the date that matters from the noise that can be set aside.
On the whole, Silver’s book asks a simple question: “How can we apply our judgment to the data —without succumbing to our biases?” As marketers, we have a tendency to think in rigid heuristics — “this” media channel is good for awareness, “this” tactic yields increased purchase intent, etc. — and bring bias to the discussion in the form of previous experiences, both good and bad. But perhaps the challenge of the emergence of “big data” is less about how to cope with increases in data and more about the type of thinking we chose to apply to it.