At The Margin

There are lies, damn lies, and statistics.  One of the things I learned a bunch of years ago was that data doesn’t really mean anything to people.  Instead of looking critically at data and making inferences, they use data to strengthen confirmation biases.

Their confirmation biases become cult-like and fanatical.  We don’t just see this in politics.  It happens in business and it happens a lot of other places too.

I remember speaking with a recent graduate of a college.  She was an economics major.  I asked her how she framed the things she looked at.  Did she understand the differences between classical and Keynesian economics?  She told me she was a “data dog”.

Unfortunately, it made her sound smart to most people but I knew she had no clue what she was talking about.

What’s the most important thing to look at when you are looking at a lot of data?  A lot of times, it’s what is happening at the margin.  This is where an understanding of basic calculus helps interpret things.  It’s also why 99% of the statistical data published today in mainstream media is either interpreted incorrectly or just wrong.  The rate of change matters and how much change happens at a particular point on the curve matters.

Things might be changing, but they might not be statistically meaningful.

It’s also helpful to know the Z score of the statistical distribution you are looking at.  A very high z score means that the statistics have been highly manipulated to force a normal distribution.  In fact, the distribution might be highly skewed or have fat tails.  For example, in venture investing we talk about a power law distribution.  That’s a highly skewed distribution and it is no where near normal.

This is important for startups.  The real change happens at the margin.  What’s it cost to do one more?  Does the marginal revenue increase faster than the marginal cost?  Is it equal to marginal cost?