What is a Test Statistic?
A test statistic is the value used in a hypothesis test to decide whether to support or reject a null hypothesis. This statistic compares data from an experiment or sample to the results expected from the null hypothesis.
Each statistic is deployed for a different test, with the most common statistics in machine learning being:
- Z-Score: Used with Z-Test
T-Score: Used with T-Test
- F-statistic: Used with ANOVA test
- Chi-square statistic: Used with Chi-Square Test
Test Statistics Using P-Values
When running a hypothesis test, the test statistic’s graph is often used to define the probability (P-value) range, usually found as a normal distribution or t-distribution. This allows the researcher to calculate the probability value for each outcome, i.e. that the results were caused by random chance or can be attributed to the hypothesis being true.
The larger the test statistic, then the smaller the p-value will be, which means the null hypothesis is more likely to be rejected.