Compensation: Correlation is not Causality
Editor's Note: It's always a good time for a Classic cautionary note, brought to us today by Dan Walter, on how we use data to support our decisions - in compensation and beyond!
“My compensation program is working just fine.” said the HR professional who had low employee turnover.
“My pay for performance program is driving our company’s success!” exclaimed the compensation professional after results were announced.
“The more movies I make, the more people who drowned by falling into pools!” mused Nicholas Cage after seeing the chart on this page.
Check out this website on spurious correlations (http://www.tylervigen.com/). It is amazing. I only wish we had a similar set of accessible data for the compensation issues. Who knew that the amounts the US spent on science, space and technology from 1999-2009 correlate at almost 94% to those people who died falling out of their wheelchair? I guess we need to start improving education so people stop falling out of their chairs! Can you imagine if we could look at the rate of retention at our company as compared to the number of times the cafeteria served chicken tacos at lunch?
It would be a useless fact but, who knows, they may correlate very closely. Does this mean we should serve chicken tacos every day and retention will go down to zero? No, absolutely not.
As it turns out, hindsight isn't 20/20. Heck, if you try hard enough, it appears hindsight may be totally blind (I am sure there is data somewhere that supports this).
Compensation people love the data that support their new decisions. Even more, they love the data that proves their prior decisions were correct. Too often people explain how well their stock option plans work by showing how much their stock price grew. Seldom do they explain how their stock option plans failed as the stock price dropped.
Companies crow about the rise in their Total Shareholder Return (TSR) as a direct result of performance linked to executive pay, but complain that the fall in TSR was due to forces outside of their control.
Often these claims are not even as well correlated as the “facts” I showed above. Is there a 95% correlation between what you pay and how your company performs? Maybe. Does this prove that your pay drives performance? Maybe it does or maybe not.
Is your executive retention, high? Sure. Is this because you pay in the 75% percentile? I need to see more than a simple correlation to believe this. We need to move away from the correlation used by cable news pundits and toward the evidence used by more rigorous scientific study. This is the only way we will continue to have a seat at the table in the age of big data.
We are in a world where “big data” will soon be able to show new and dumb correlations as well as important and useful causality. Understanding and getting ahead of this curve is exactly what CEOs and CFOs are demanding from the HR and compensation professionals. Do you have stats, evidence, facts or proof?
P.S. I wrote the original draft of this article two weeks ago. Mike Walsh a keynote speaker at the World at Work conference also mentioned the “spurious correlations” web site. Correlation or Causality?
Contact FutureSense at info@futuresense.com if you need help with your team, management and HR consulting techniques to better retain people, coaching, or even support with enhanced pay structure approaches.
Dan Walter is a CECP, CEP, and Fellow of Global Equity (FGE). He works as Managing Consultant for FutureSense and has three metaphors for every occasion. Dan is also a leading expert on incentive plans and equity compensation issues. He has written several industry resources including the only resource dedicated to Performance-Based Equity Compensation. He has co-authored ”Everything You Do In Compensation is Communication”, , “Equity Alternatives” and other books. Connect with Dan on LinkedIn. Or, follow him on Twitter at @DanFutureSense.