Transcript
This is a short lesson about the P-value. It might be one of the most important lessons in your statistical life. P-values are everywhere in the research literature, yet they are often misinterpreted and misunderstood, so it's important to know what a P-value is and what it is not. P refers to probability, so the P-value is a probability between 0 and one. P-values come from testing theories using sample data. A common example is a comparison between a standard treatment and a new treatment. Using a randomised control trial, we might, for example, want to compare a traditional method for supplying oxygen to pre-term infants with a new method.
Typically, the calculation of the P-value for theory testing is based on a pessimistic view of what's true in the world. The pessimist view is that the efficacy of the treatments we are comparing to not actually differ; in this view, the death rates, for example, in pre-term babies, will be the same for the two methods. Once the data are collected, a P-value is calculated using a statistical model. We can think of the P-value as a potential challenge to the pessimist view of the world. A small P-value indicates that the data we have is somewhat surprising. If the pessimistic view of the world is correct, most often researchers are keen to find small P-values at heart, they are optimists and want to challenge the pessimistic view. The P-value doesn't tell us directly about the plausibility of the pessimist view of the world, we don't know if this view is right or wrong. It gives us an idea of the plausibility of the data we've sampled, given that view more formally, the P-value tells us the probability of the result we've obtained, or results would challenge the pessimist view more strongly if the pessimist view of the world is right.