In the October 27, 2008 issue of The Guardian, Alexis Petridis makes an interesting observation

“Last night came final and irrevocable proof that the country is entering tough economic times, unseen since the 80s: AC/DC have returned to the top of the album charts for the first time in 28 years. . . Those keen to draw wider inferences from its success might note that the last time AC/DC made No 1 in Britain, the country was on the brink of recession. . . When the economy recovered, AC/DC’s popularity receded. . . But right on cue the album that returned the band to its heyday was The Razors Edge, released in 1990 – just as Britain headed towards its last recession.”

So, fixing the economy (at least in the UK) seems to be as simple as boycotting AC/DC, right? Of course not.


Most people see this as an interesting coincidence. Some, like Petridis even go on to speculate about a  link between the two

“AC/DC’s appeal in unpredictable times is straightforward. People crave something uncomplicated and dependable in a time of uncertainty, and rock music has never produced a band so uncomplicated and dependable as AC/DC.”

Yet, no one would think that AC/DC’s commercial success causes economic turmoil.


Sometimes that distinction isn’t as clear, especially when the two issues are more closely associated. Confusing coincidence, correlation, and causality can lead to bad decisions and costly actions.


Correlation is when two things consistently move in the same (or opposite) directions. For example, your credit rating and insurance risk are correlated. People with better credit ratings generally have lower risk. Better credit doesn’t cause their risk to be lower. It just happens that those two things tend to go together. They are probably influenced (caused) by similar sets of personal attributes. One advantage of correlations is that they can provide insight into how one thing is likely to behave based on the behavior of something else. As with the case of insurance underwriting, this is especially helpful when it is easier to gain information on one of the two issues at hand.  Correlation doesn’t give you insight into how to change either of those behaviors or how they interact. Contacting one of the many agencies that promises to improve your credit rating won’t make you a safer driver.


Confusing correlation for causality could lead to erroneous and ineffective investments. In the book Freakonomics, Steven Levitt and Stephan Dubner describe such an error. A study showed that children whose households had more books performed better academically. This was a correlation. The number of books doesn’t cause kids to be smarter. Both the number of books in a house and academic performance are probably a result (caused) by parent attitudes about learning.


However, in early 2004, Former Illinois Governor Rod Blagojevich developed a proposal. The state would mail one book per month to every child in Illinois from the time they were born until they entered kindergarten. The legislature ultimately turned the proposal down. Had they not, the state would have spent approximately $26 million in taxpayers’ money on a program that would have provided little value. Confusing correlation and causality can be costly. But it’s easy to fall into the trap. Mailing books is much easier than improving student performance and it creates the illusion of action. It’s tempting.


In a prior entry, I referenced a report that argued companies should invest more in engaging their employees. The report cited a study showing that companies with low employee engagement also had lower shareholder value. However, the data, as presented, did not show causality, it only showed correlation. There might be a causal relationship, but you must find evidence for it, and not infer it from the correlation. Perhaps the causality was reversed – companies with lower shareholder value might not have as much to invest in their people, causing decreased engagement. Or, maybe the poor performance and engagement was due to bad leadership, management, systems, tools, or processes. Before taking an action to fix a problem, it is important to know whether that action will actually change anything.


Many leaders feel pressure to demonstrate action and prove “ROI” on that action. That’s when the lure of correlation becomes strong. It’s convenient to assume that positive business results are caused by the most recent initiatives or actions that you’ve put in place. But, don’t fall into the trap. Before declaring victory on your actions (or investing in new one) make sure that you know if the relationships that you are seeing are causal, correlations, or just simple coincidences.


————————
Brad Kolar is the President of Kolar Associates, a leadership consulting and workforce productivity consulting firm. He can be reached at brad.kolar@kolarassociates.com.