How we use data to validate our stories and draw bad conclusions.
Today, our every move and whim has the potential to be tracked and analyzed through technology we carry with us, or through security cameras. (See blog on the trade-offs of ‘big data’) That means that anyone with something to market or sell has never been in a more powerful position to influence others’ opinions, actions, and decisions.
But as we know, data can be manipulated. The manipulation of data is not inherently a bad thing. At the core, it simply means organizing data in a way that is more meaningful or easier to understand. For example, alphabetizing data is an example of data manipulation. However, data manipulation can often carry a negative connotation, whereby someone knowingly alters the data or results to fit his study, his story, or the conclusion he hypothesized. This might seem relatively harmless when a 6th-grader alters her science fair data to match her hypothesis. But in the science of medical research, data manipulation can have devastating effects. We see this in the recent measles outbreak. A now discredited study linking vaccinations to autism is arguably partially to blame for creating enough fear, uncertainty and doubt among parents with no medical training to decide to skip the vaccinations for their children. And now we are faced with the resurgence of a disease we thought was largely eradicated in 2000.
The Unconscious ‘Natural Selection’ of Data
I offer two different articles I recently came across to both raise your awareness and caution you as to how you accept or reject arguments relying on data. Our biases, combined with the new world of big data, could have us creating all kinds of bad conclusions unless we inject an ounce of skepticism before we eagerly fit the data to our stories.
A little background on our biases and our stories:
- First, we are the stories we tell ourselves we are
- Second, our biases are always at work to validate our thinking / our stories.
The New England Patriots and “Deflate-gate” offers a relatively recent example. In a Harvard Business Review article about the dangers of reverse causation, Kaiser Fung breaks down the argument used in a Slate article that asserts one data element, the Patriots extremely low fumble rate, as the ‘smoking gun’ proof that the Patriots cheated. In the HBR article, Fung says:
“Big data is exposing all kinds of outliers and trends we hadn’t seen before and we’re assigning causes somewhat recklessly, because it makes a good story, or helps confirm our biases.“
“Data often offers hints, not proof.“
In a more personal example, this Wall Street Journal article details how the author and his family felt duped at the hands of a magician. Using the author’s son as a volunteer, the magician had the whole family believing that their son had amazing powers to discern the truth based on the subtleties of human behavior. Once the con was revealed (to much laughter and some embarrassment), the author realized that when faced with a string of successes, it never occurred to him to question what was really happening, which they would have done if faced with a string of failures.
The Lesson for Big Data, and Our Biases?
We would all do well to inject some skepticism into data that too readily fits our stories. We must be aware of our biases and our stories. Otherwise, we are in danger of accepting faulty logic and validating data that could be translated in a hundred different ways.
A heightened awareness will go a long way to guide you on a more reasonable path to your decision-making, and, will help prevent you from being taken advantage of or manipulated.
(c) Joe Caruso and Caruso Leadership. Reprints available with permission.