One-Tailed Test

Think of a nice, normal distribution...like a bell curve. When you’re working with hypothesis testing, you try to determine the probability that the hypothesis is correct, or that the null hypothesis (no relation) is correct. One of the ways to test a hypothesis is via a one-tailed test.

While a two-tailed test uses both “tails” of a normal distribution, a one-tailed test only uses one of the tails, either the right or left. A one-tailed test is done to see if a value falls into the “tail” area, or under the main bell curve area. What’s the line that separates these two areas? It depends on the test, but often it's a measure of standard deviation away from the mean, which is in the middle of the normal distribution, or more specifically, related to a p-value.

When hypothesis tests are conducted, a researcher can determine if a test fails to reject the null (which means the null hypothesis may be correct, we can’t rule it out…which isn’t looking good for the hypothesis), or if it rejects the null (which means there’s a good chance the hypothesis is onto something). Since a one-tailed test is only testing one side, it’s also known as a directional test for a hypothesis, since we’re only testing one direction of the relationship.