« Personal Pay Versus Salary Structures | Main | Are Your Best Employees Holding You Hostage? »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Right on Dan! But HR listens to HR consultants more then cable news pundits and HR consultants are very guilty of making up correlations where none exist.

HR again needs to get out of the "lemming" mode and research these wild "facts" themselves ---- but it's easier to just accept what the "experts" say.

You're right --- maybe the accessibility of big data will show these faulty correlations for what they are. Maybe Lazlo Bock could teach us a thing or two.

Nice post Dan. Your example is why I talked a company out of using TSR in their equity plan when I showed them the company’s financial performance had a low correlation to TSR. Many people think TSR is the perfect metric to tie executives to shareholders but the correlation many not always be there.

Big data in HR (Talent Analytics) is here and companies have to understand how to interpret it before chasing every correlation.

The original textbook (hard-copy, long before the electronic internet age) for the first Statistics Used in Personnel certification course done for ACA & ASPA (which also dates it) contained a few examples of correlation misunderstandings. Think one was how the rainfall in Bombay (now Mumbai) correlated with Topeka birth rates or something.

"Spurious correlation," on the other hand, describes ratio relationships and is different from the basic fact that correlation does not imply causation. "Spurious relationship" means there is a hidden variable that confounds mathematical comparisons like regressions and correlations. Correlations confuse everyone!

Thanks Jacque,

It is so hard for HR people to stop following the leader when their outside "expert" resource keep providing "evidence" that shows trends, market data and best practices.

Being different in the face of this information requires not only doing your own research but having the courage to stand up for your results.


I think the reason people love TSR so much is precisely because it results in the same delivery as stock options (with less of the downside risk). The fact that TSR peer groups are not comparable between companies (both in size and make up) combined with a wide range of design elements that can drive payouts differently for exactly the same results means that most investors (and many companies) have no idea exactly what they are agreeing to.

Jim, I was starting to get worried. I expected you to be the first to comment!

Thanks for the additional insight


Years ago to illustrate uses and limits of multivariate analysis, an executive compensation consultant who was also an actuary at Hay ran a bunch of tests on causality linked to CEO compensation for an article. One of the tests included seven factors such as number of employees, years as CEO, and sales levels. In that test a correlation 74% popped for the CEO's state of birth. Certainly a limit like you wrote about.


Thanks for the additional info Andy.

I like this old joke about correlation and causality.

Man moves into a new apartment in a new city.

Man goes to doctor complaining of a long-term headache, blames it on stress or allergies.

Doctor asks a few questions and gives him a mild pain killer.

Man comes back two weeks later still complaining of headaches.

Doctor does a few tests, finds nothing unusual, and gives him a strong pain killer and some allergy meds.

Man comes back week later and says headaches are worse than ever. He may have to quit his job and move back to his home state. "This state is killing me."

Doctor is now very curious and asks if they can video tape the man at home for a few days to see if they can determine the "environmental cause."

Video tape shows the man sleeping on a loft bed. When his alarm goes off in the morning he wakes up abruptly and slams his head into the ceiling, falling back to sleep for a few minutes.

Headache was real. Environment was the problem. But correlation created an assumption of causality, evidence showed the actual causality.

Man comes back in

One major problem is that the audience will see what they want to see; and to a certain extent some providers will deliver "proof" to support a required end result.


The audience will see what they want to see. You are correct. Perhaps compensation professionals need to be more like the magicians who explain exactly how they accomplish their tracks and less like the smoke and mirror shows in Las Vegas.

People often see what they want to see because we don't take enough time, or have enough data, to make it clear how they should be looking at things.

Your comment about providers giving selective proof to support the end result is a real problem in our industry. Everyone needs to be diligent about asking questions. And we all need to point out when we think the Emperor is wearing less than a full set of clothes.

The comments to this entry are closed.