Fine-print Examples

Of course it’s impossible list every example of engineers’ misuse of statistics but here are a few that come immediately to mind:


  • Not understanding the difference between knowing a parameter’s value and having to rely on an estimate of it.
  • Believing that p-values measure the probability a null hypothesis.
  • Over-fitting: Trying to achieve a better description of the existing data with a more complicated model at the expense predicting future observations.
  • Acting as though “the variance of a sum is the sum of the variances” is always true. (It’s not.)
  • Not understanding that laboratory measurements are only one realization. Another collection of nominally identical measurements would be different.
  • Thinking that repeated inspections improve reliability.
  • Believing in \(R^2\).
  • Not understanding the distinction between a valid statistical statement and a true statistical statement.
  • Beliving that zero correlation means no relationship between two variables.
  • Looking for reasons why your conclusion is correct rather than for reasons it might possibly be wrong. *

* In fairness this is more an engineer’s engineering failing than a statistical one.