What happens if a hypothesis is tested and shown to be false?

Answer 1

We do not make a clear separation between "proven true" and "proven false" in hypothesis testing.

Hypothesis testing in statistical analysis is used to help to make conclusions based on collected data. We always have two hypothesis and must chose between them. The first step is to decide on the null and alternative hypothesis. We also must provide an alpha value, also called a level of significance.

Our null hypothesis, or status quo hypothesis is what we might conclude without any data. For example, we believe that Coke and Pepsi tastes the same. Then we do a survey, and many more people prefer Pepsi. So our alternative hypothesis is people prefer Pepsi over Coke. But our sample size is very small, so we are concerned about being wrong.

From our data and level of significance, we find that we can not reject the null hypothesis, so we must conclude that Coke and Pepsi taste the same.

The options in hypothesis testing are: Null hypothesis rejected, so we accept the alternative or Null hypothesis not rejected, so we accept the null hypothesis.

In the taste test, we could always do a larger survey to see if the results change.

Please see related links.