Playing To Win
A Dangerous Schism
Your Strategy Course is Utterly Inconsistent with your Statistics Course
There is a universally present and dangerous schism between what business students learn in their statistics course versus their strategy courses (and pretty much every other course). I explore the implications in my 23rd Year III Playing to Win Practitioner Insights (PTW/PI) piece called A Dangerous Schism: Your Strategy Course is Utterly Inconsistent with your Statistics Course. You can find the previous 133 PTW/PI here.
The Importance of Business Education
There is a schism between the statistics courses in business education program and most other courses in those programs — though I will focus on the schism with strategy in the context of this PTW/PI series. The schism is just plain not discussed. Because of that, it is very hard for students to identify it, meaning that they graduate largely if not entirely oblivious to the schism. And that is not a small thing.
Business education by far is the biggest higher education enterprise in America. Nothing comes even close. Business degrees comprise over 20% of combined undergraduate and masters graduates conferred each year. That is more than graduate in engineering, mathematics, computer science, biology, chemistry, physics, and geology — combined! These are the numbers.
These nearly 600 thousand newly minted business graduates are injected into the business community each year — over 10 million of them since the turn of the 21st century. Given that the Bureau of Labor Statistics estimates that there are just under 10 million managers working in the US currently, it is fair to guess that the 10 million business graduates take up a large proportion of those 10 million managerial positions. Hence, what their formal business education teaches them is hugely influential in the modern world of business.
Virtually all MBA programs have a first year required statistics course and at some point during undergraduate business programs, students will take a required statistics course. That required statistics course teaches a number of things, but the most centrally important is how to make valid inferences from data. That is, if you analyze a body of data, how can you be confident in making a judgment as to what the data means? For example, how can you tell whether our large-scale plants have better unit costs than our small-scale plants? Do more experienced salespeople sell more than less experienced salespeople? Which do customers value more: product variety or cost-in-use?
To create such a valid inference, students are taught how to structure a scientifically rigorous experiment. On that front, they are taught that the sine qua non is a representative sample of data. For example, if we assembled a data set of only salespeople who sold a certain product in our lineup, our sample wouldn’t be representative of salespeople companywide across product lines, and it would be an error to make an inference from the analysis to all salespeople and products. It is drilled into the heads of all students that in order to be able to make any valid inference from the analysis of data, you must analyze a representative sample. They are warned that if their sample is not representative, the inference they draw from the analysis will be flawed in ways that they can’t predict or understand.
Strategy (and Other) Courses
Students walk out of their statistics class and into their strategy class (and equally their marketing, finance, operations, etc. classes), where they are taught clearly and unambiguously that the only good decision is a decision based on rigorous data analysis. This has been the case for at least the past quarter century. I know because I entered the formal business education space as a dean in 1998 — a quarter century ago — that was a central message to students by that point in time. And that is also what I was taught when I attended business school another 19 years before that. So it has been going on for a long time!
Students are taught an endless array of analytical methods related to strategy, and each of the other disciplines, and encouraged to use them to make business decisions. In fact, they are evaluated and graded on the basis of their ability to utilize the analytical techniques to do rigorous data analysis to recommend courses of action.
Once again, there is a clear and unambiguous message: rigorously analyze in order to decide. If you don’t, you are a bad/sloppy/lazy businessperson.
The Tricky Part
The indoctrination could not be clearer. To be a meritorious businessperson, you must make decisions based on rigorous data analysis and the data analyzed must be a representative sample drawn from the universe to which your decision pertains. Each of the 600 thousand graduates per year learn those two things — the primacy of data analysis and the criticality of a representative sample.
But there is a tricky problem for that prescription. All the data for the prescribed analyses are from one era. That is because 100% of the world’s data across all fields is from the past. There is no data about the future.
But simultaneously, it is also true that all business decisions are about the future. We can’t make decisions about the past — that has already happened. The only decisions are about the future — even decisions that we characterize as being about ‘the present’ are in fact about the near-term future.
So, we are taught to rigorously analyze the past in order to make databased business decisions about the future. But since the data we use must be representative — otherwise we violate that necessary condition, which we learned renders the analysis invalid, and dangerously so — there is one very big and entirely implicit assumption being made:
Yup. The big implicit assumption is that the future is identical to the past. If it is, then a sample drawn from the past is entirely representative of the future, and it is safe to make a decision based on rigorous analysis of such a sample.
But it behooves us to ask: just how safe is that assumption? How many times is the future identical to the past? Occasionally, I guess. Though probably the more accurate answer is that occasionally it is identical but only for short durations of the future.
The overwhelming majority of time in business, the future is different, and often frustratingly so, from the past. Therefore, any inference that you draw from data analysis — which is inevitably about the past — to utilize in making a decision about the future is fundamentally flawed.
[At this point I feel obligated, parenthetically, to address a modern fetish of data analytics enthusiasts. They are so excited about their ability, thanks to modern information technology, to analyze the entire universe, not just a sample thereof, that they argue to me that this sampling issue disappears. Thanks to the wonders of ‘Big Data, the sample is the universe. Nope. There is still zero data about the future so the dataset being analyzed, even if epically-sized, is still fundamentally unrepresentative — and the problem still manifests.]
The Schism and its Downsides
This business school schism between statistics teaching students that they must have a representative sample, while strategy (and most of the rest of courses) teaches students to utilize analytical techniques that implicitly assume the past is representative of the future — while, of course, failing to inform students of that reality (though in defense of most of professors, they are doing so through ignorance, not malice).
There are many downsides of this business educational schism, but I will focus on two of the most damaging ones.
First it teaches students to be wildly overconfident. They feel confident in taking decisive action on the basis of their scientifically rigorous data analysis. Then the world changes — which their analysis convinced them (implicitly) that it wouldn’t, and they get annihilated by a competitor who wasn’t similarly clueless and overconfident — and remember it only takes one more thoughtful and nuanced competitor.
Second, and seemingly contradictory but not in fact, they become massively conservative. Analysis will never provide evidence that something new and different will succeed because that is an impossible conclusion to draw from analysis of past data. It will, in fact, buttress the notion that what is happening now will continue to happen in the future — because that is all data analysis can do.
Overall, it is truly sad. The business educational complex graduates 600 thousand students per year who have been taught to be conservatively overconfident. It is accomplished by teaching them incongruent, contradictory things, and at best being oblivious to the contradictions from one course to the next. At worst, it is accomplished by covering up the schism from the students for whose enlightenment that complex is responsible.
I assume that a key driver is the silo-ization of academic knowledge. Statistics professors don’t sit in on strategy classes (or vice versa) and think: Whoa, what are you teaching? Each wants to stay out of the other’s business. Regardless, it is just plain sad.
On this front, you have to roll your own. You are not going to have anyone from the business educational complex enlighten you on this front.
Every single time that you consider an analysis, you have to ask yourself: Am I sufficiently confident that the future will be identical to the past? If so, go ahead and make the decision based on your rigorous data analysis.
If you aren’t confident that the future will be identical to the past, even deeply suspicious that it most certainly won’t, then be really, really careful about using your analysis (or even performing it in the first place). All you know is that the analysis will lead you astray — but sadly, you will not have a clue as to in which direction you will be wrong. The analysis is incapable of giving you hints about that. The only thing you can know for sure is that the analysis will be invalid in mysterious ways.
If you want to deal productively with the future, you have no choice but — as Peter Drucker admonished — to invent it. And some thoughts on how are sprinkled through previous PTW/PI pieces, which I recommend you reread: balancing reliability and validity, balancing manipulation of quantities and appreciation of qualities, understanding Aristotle’s admonition and advice, thoughts on approaches to creative ideation, the beneficial role of analogies, and freeing new ideas from the presumption of guilt.
The most important thing is to not let your formal education consistently undermine your effectiveness as a business decision maker. It may have hidden the schism from you. But now that you have been made aware, you have no excuse to keep on falling prey to it.