PDF

What is a p-value?

A p-value is a number that helps you judge whether you saw the results you did by luck, assuming a standard rule used in statistics called the null hypothesis.

Step-by-step intuition

  1. Set up a null hypothesis: You start with a default idea you try to test. For example, a coin is fair (50% heads).
  2. Collect data: Do an experiment or look at data. Suppose you flip the coin 20 times and get 16 heads.
  3. Ask how surprising the data is: If the coin is truly fair, getting 16 heads out of 20 is unusual but possible. The p-value answers: "How unusual would this be if there were no real effect (under the null hypothesis)?”
  4. Interpret the p-value:
    • A small p-value (commonly ≤ 0.05) suggests your result would be unlikely if the null hypothesis were true. This is some evidence against the null.
    • A large p-value suggests the observed result could easily happen by luck if the null were true. This is not strong evidence against the null.
  5. Important caveats:
    • It does not prove the null is true or false.
    • It depends on how many tests you run and what you consider “unusual.”
    • It’s a tool for decision-making, not a final verdict.

Simple analogy

Imagine you have a deck of 52 cards. If someone tells you a card drawn is a King, a p-value would be a way to measure how surprising that result is if you had not looked at the card yet. A very surprising result would have a small p-value, suggesting something noteworthy happened.

If you’re ever unsure

Ask these questions: What is the null hypothesis? What would be considered a “surprising” result here? Is the p-value small enough to matter given how many tests I did?

In short: a p-value tells you how likely your data would be if there were no real effect. A small p-value hints at an effect, but it’s not a guarantee.


Ask a followup question

Loading...