Hypothesis Testing

Lately I’ve been reading up a bit on what management consultants do.  Here are a group of people who go into Fortune 500 companies to advise senior management on how to deal with strategic or organizational issues — and get paid lots of money to do it.  Judging how much money they make, they must be providing some level of strategic value.  It turns out these consulting firms hire some of the best and brightest (think MBAs from Harvard), which is one reason why they provide value.  But it turns out that just having smart people come in to tell you what to do isn’t worth that much.  In fact, that’s only a small part of the value they provide.  One of the most important things they do is help identify, analyze and recommend solutions for strategic problems primarily using time tested methods such as problem solving frameworks.

This idea of frameworks is incredibly powerful.  One can imagine the diverse set of different problems that these consultants must encounter jumping from one industry (not just company) to another.  Yet somehow they are still able to provide useful recommendations that their clients are happy with.  This means that they must be doing something right.  One important framework that I’ve read about is this idea of hypothesis testing.

Now this idea is not new; it’s been around since the 17th century more commonly known (as least to scientists) as the scientific method.  The concept is extremely simple: you do not want to bias your investigation with your own preconceptions, instead let reality speak for itself.  Of course it’s easier said than done, but here are the main points:

  1. Formulate Question: The question outlines the initial problem.  It’s important to ensure that the question effectively addresses the problem.  Figuring out if you’re asking the right questions is a whole different story though (and a separate post).
  2. Hypothesis: An initial tentative answer based on known facts or experience (not just the first thing that pops into your mind).  Sometimes it can be based on a hunch, but reflects your best understanding of the situation and reality.  An important property of a hypothesis is that one is able to definitively support or reject them.
  3. Predictions (or assertions): Based on your hypothesis a series of logical conclusions will follow.  Here it’s useful to systematically break down the hypothesis and assertions into what consultants refer to as Mutually Exclusive and Completely Exhaustive (MECE) conditions.  In other words, each statement in the hypothesis (and assertions) should not overlap or depend on each other, and they should sufficiently cover all aspects of the initial situation or problem.  This aspect is important for consultants where the hypothesis needs to be credible.
  4. Testing (and analysis): This is where you determine if the hypothesis is true.  Using known facts or experiments, see if any of the statements in the hypothesis or assertions are contradicted.  If a convincing number of supporting statements or experiments match your initial hypothesis, it probably represents reality fairly accurately.  Otherwise, your hypothesis is false, which is equally important because it allows you to refine your model of reality (and go through the entire process again).

Now everything that I’ve written so far is probably not new to you (at least if you remember what you learned in science class).  However, what I find fascinating about consultants is that they use this framework every time they approach a problem (at least with regards to their client).  This is an important because I think everyone knows that they should do hypothesis testing when running a science experiment, but what about when finding how much the market is willing to pay?  Or whether a stock is any good to buy?  Or whether you should even be following this advice?

Simple ideas are powerful and can produce extraordinary results.  And hypothesis testing is definitely simple and extremely powerful.  It’s something that we should all remember when trying to form a model of reality.  To quote Charlie Munger:

Take a simple idea and take it seriously.