Statistical Power

The power level of one or more data sets describes how good the inferences drawn from your data are likely to be.

For example, you make 119 measurements of some quantity in millimeters.  Then you determine your data is best analyzed using a one sample Student's t‑test, and you find that the mean of your sample measurements is 11.06 mm.

Now the question to ask is:
How likely is it that someone else who performs these measurements will also get a mean of 11.06 mm?

This is the "power" of your inference.

---

Transcript

So what is statistical power? Which is, after all, what we started this discussion to talk about. Well, statistical power tells us how we can make inferences based upon how likely something is in the data. So, for example, in this case, 119 measurements were made up of a quantity in units of millimeters. We use the one sample Student's T-test, and we found the mean of the samples was 11.06 millimeters.  So the question we ask is: "If someone else were to perform those measurements, how likely would it be that they would also find a mean 11.06 millimeters?". And that's the power of the inference.  Right. It's about - can we take the experience that we have for one person making these measurements and is another person who repeats the same experiment the same number of measurements likely to find the same mean value?