In Latin ‘Post hoc ergo propter hoc’. Free translation: B follows A, so A must be the cause of B. It’s a fallacy. We installed that piece of software; since then, the computer is very slow; that software is causing this performance issue in my PC. We have just come back from a trip abroad; one of the kids now has a serious fever and is sick; she must have got food poisoning from that last dodgy restaurant.
Since everyday life is full of ‘post-hoc fallacies’, there is little point in giving more examples. You have, and will recognise, plenty of them. No surprisingly, ‘post hoc fallacies’ also dominate business life.
All people in the sales force have gone through the new, expensive sales training programme that lasted six months. Our sales figures have markedly improved. That sales training did the trick.
Joe has taken over as the new CEO, after the rather disastrous year of Peter at the helm. The stock price has rocketed. Joe is the right leader, the market always knows.
We have gone through a one year, intensive Employee Engagement programme, with multiple initiatives at all levels, and you can see what happens: the overall company performance this year has been brilliant. And the overall employee turnover halved! Another example of how Employee Engagement pays off.
These are three real stories from my consulting work with organizations, just last year. And ‘stories’ is the right term. Damn good ones, I have to say. But without exercising some critical thinking, these stories may remain at the stage of fallacy.
The sales training may have been excellent, but the markedly improved sales figures could also be explained by a pathetic performance of the main competitor, who screwed up completely their greatly anticipated new product launch.
Joe may, indeed, be what that company needs as a CEO. But the stock price success could also be explained by a cost cutting programme that Peter, the disastrous CEO, had started before he left, and which just now is showing results. No offense, Joe.
The Employee Engagement is a great initiative, but instead of leading to a brilliant company performance, it could be that the brilliant company performance (based upon a series of successful launches) had shaped employee satisfaction and sense of pride. This may be why people scored so high in many parameters in the Satisfaction Questionnaire. A Halo effect.
A fallacy is only a fallacy until one looks critically at it and explores alternative thinking. Left on their own, they may be very good stories of success, but the arguments behind may or may not be true. When, in my Speaking Engagements, I challenge audiences to think of potential fallacies in our arguments, I am conscious that I am pushing dangerous hot buttons. No Training Manager wants to hear that their programmes may or may not have the attributed impact. The same for Investor Relationships, or the Board of Directors, or HR.
Taken to the extremes – people tell me – we would not do anything, since (according to me, they say) we can’t prove much on Management. But this is a narrow view of why we should do things in management. Sales Training programmes need to take place, perhaps CEOs need a replacement, and there is nothing wrong at all with that Employee Engagement programme. We do all these things because we believe in good management and because we are paid to exercise judgement. Don’t stop them!
Exercising critical thinking and practicing good management are not in contradiction! Not all good stories of success contain a fallacy. But spotting management fallacies can only lead to a better management. The key is not to settle for a good story.
Now we are in the age of Big Data, and at least in the hi-tech world, companies are starting to make decisions based on Big Data analysis (otherwise known as “objective empirical analysis”). Who could argue with that? We have all this data about performance, for example correlating some measured characteristics of new hires (degree, university attended, answers to certain interview questions…) with some measure of their later success. Shouldn’t that objective computerized analysis guide our hiring decisions, rather than the gut reaction of a few interviewers? The bigger the company, the more data they have about everything, and the better they can do with this approach.
Well, there’s a problem. It’s called “spurious correlation”: two sets of numbers match up, and you just assume that one is causing the other, and that the correlation will continue on into the future. I think this is the same phenomenon Leandro is talking about, but wrapped up in the cloak of objective science and mathematics.
We also see this when people pick a mutual fund to invest in, based on its past performance. If you look at thousands and thousands of mutual funds, there may be some that perform well because their stock-pickers are better than the others, but there will also be dozens that performed well out of sheer dumb luck.
If you have more data for a single comparison, the odds that the things you are comparing correlate by pure chance go down; but if you have more things to compare, the odds of finding a few correlations go back up. Here’s a good (and sometimes hilarious) set of examples that illustrate this:
So when you see a correlation between X and Y and think it’s causality, you’d better stop and consider whether there’s any plausible story connecting X and Y. It’s easy to dismiss the correlated trends in the web site above because you can’t come up with a story about how X caused Y, or because any story linking the two trends is so far-fetched. The dangerous cases are correlations where there is some possible story linking the two, and where you really want to believe that X causes Y. In such cases, it’s wise to stop and look very closely to see if you can determine what the REAL story is.