Wading Through the Data Swamp:
Program Evaluation 201
Hypotheses
Before we can study outcomes we must first have hypotheses in place. Hypotheses are predictions of outcomes and relationships.
In statistics, we always have a null hypothesis and a research hypothesis.
SCIENTIFIC RULE
Your hypotheses should clearly state what you are trying to reject (null) and fail to reject (research).
Take a look at the examples below. The box on the left contains hypotheses written in a statistical manner. The box on the right contains hypotheses written in everyday language. The hypotheses in each box say the same thing and are acceptable as written. Keep in mind that hypotheses can be written in many different ways and yet have the same meaning.
| Statistical Language | Everyday Language |
|---|---|
|
Null Hypothesis H0: The two variables, program participation and marijuana use, are independent at posttest. H0: The two variables, program participation and alcohol use, are independent at posttest. |
Null Hypothesis H0: Program participation does not affect marijuana use. H0: Program participation does not affect alcohol use. |
|
Research Hypothesis Hi: The two variables, program participation and marijuana use, are not independent at posttest. Hi: The two variables, program participation and alcohol use, are not independent at posttest. |
Research Hypothesis Hi: Program participation does affect marijuana use. Hi: Program participation does affect alcohol use. |
Home | Courses | How to Use | Contact Us
For questions or comments, please visit our Contact Us page.
Last Updated: 11/20/03
SAMHSA is an Agency of the U.S. Department of Health & Human Services
