The Beta of Data
The Beta of Data
When making the decision to fly on a trip, most consumers weigh three factors more than any others: cost, flight times, and available seats.
Other data points, such as aircraft type, pilot experience, food or snack choices, and the like take a back seat to the data points consumers give beta weight.
In fact, having more data actually lowers decision quality and makes the decision more complex than it needs to be. This is true more often than not in all decisions. More data is not an answer. The right data matters much more.
In a time where big data reigns supreme and collecting ever more data is in fashion, the idea that less data makes for better decisions can appear counterintuitive. But leaders who reflect on their best past decisions will likely conclude they placed mindful significance on specific data and ignored the rest.
Giving beta weight to certain qualities and data streamlines decision-making and makes the process more manageable. Not surprisingly, except to data wonks, it makes for better decisions.
Take, for example, a hiring decision between two junior candidates. Will a leader who has more than 100 data points, from school selection and grades to personality tests and personal values and everything in between, make the best call?
Or will a leader who believes deeply in two or three qualities, such as self-awareness, grit, and locus of control make a better choice?
Of course, placing a beta weight on particular data presumes an accurate theory. If the factors a leader overweighs are not powerful enough to predict a superior outcome, then focusing on them is foolhardy.
But even when leaders are wrong in their theories as to what indicates or predicts a better result, more data is rarely advantageous. In those cases, leaders need a better theory, not more data.
The idea that ever more data will crystalize a pattern from which such a weighted framework will emerge is wishful thinking. Collecting more ingredients without knowing the recipe rarely makes a dish taste better.
The key is to know what data points matter most before we begin assembling the set. Diving deep on those factors will likely produce a more confident and effective decision.
Less really is more when it comes to data and decisions. This maxim runs contrary to contemporary thinking. Don’t let the big data vogue fool you.
The right data always matters more.
I like this, but I can’t help but think my decision making doesn’t make me paralyzed by more inputs, just more shaded.
I suppose if I came to a problem where I was truly lost on a direction I should head, more data points would make me more lost.
Generally however - I have a confidence in the general direction of a decision and I don’t think of additional data points as course altering as much as it’s peripheral texture.
Even the major league data wonks would use the multiples data inputs to get down to a handful of most helpful statistics. In baseball it was on-base-percentage that is the key performance indicator for offense. More recently, it has become a players WAR that has been the accepted total contribution... a stat that didn't show up on baseball cards until about 2014. WAR uses 7 previous stats with about 11 individual data inputs.
Big data created WAR, but was smart enough to deliver it's results in what feels like a single data point.
To your point above... what does WAR do for us when all we really need is someone who can hit for power 5th in the lineup?