Statistical tests
From a one-sample t-test to Cox regression. Not sure where to start? Answer two questions and we will point you to the right test.
Pick what you are trying to do, then one more question.
Tests to look at
Three short sequences, each a handful of tests. Click any test to see when to use it.
Select up to 3 tests in the list above.
The chance of seeing a result this strong if there were really no effect. Small (under about 0.05) suggests the effect is real, not noise. It says nothing about how big the effect is.
The range the true value most likely falls in. A narrow interval is a precise estimate; a wide one is not. If it excludes the no-effect value (0 for a difference, 1 for a ratio), the result is significant.
How big the effect is, on a scale that does not depend on your sample size. This is the number that tells you whether the result matters in practice, not just in statistics.
The assumption that data follow a bell-shaped (normal) distribution. Many tests assume it; nonparametric tests do not.
A condition a test relies on to be valid. Violating a key assumption can make a result untrustworthy.
A test that assumes a specific distribution, usually normality. Powerful when the assumption holds.
A test that does not assume a specific distribution. Often the choice when data are not normal.
Data that fall into groups or labels, such as yes/no or low, medium, high.
Data measured on a scale where any value is possible, such as height or time.
Separate groups with no link between them, for example two different sets of people.
The same people measured twice, or matched pairs. Requires a test for related samples.
The data you actually have. Results from it are used to make inferences about the population.
The whole group your sample is drawn from, and which you ultimately want to understand.
Whether a result is unlikely to be due to chance, usually taken as a p-value under 0.05.
The size of a difference or relationship, on a scale that does not depend on sample size.
Enter two small groups of numbers and see what a t-test reports. This runs the same calculation the app does, right here in your browser.
Every test above is in the app, free to start, no card required.