A statistical hypothesis test offers a set of mutually exclusive hypotheses and states a conclusion in terms of these hypotheses. For statisticians, understanding interpret hypothesis tests is second-nature, but not everyone is a statistician. In many circumstances, it is important for researchers from other fields to interpret the hypothesis test given in a report. However, you will not find such a task to be difficult; you only need locate the small handful of pertinent statistics and results.
Instructions
1. Find the set of hypotheses. For almost all statistical hypothesis tests, there are only two hypotheses. These hypotheses are usually written in the form "H0" for the "null hypothesis" and "H1" or "Ha" for the "alternative hypothesis."
2. Interpret the meaning of these hypotheses. The hypotheses will be mutually exclusive equalities and inequalities. For example, "H0: x = 0" and "Ha: x != 0." In this situation, what the null hypothesis states is that the statistic "x" is equal to zero. On the other hand, the alternative hypothesis states that the statistic "x" is not equal to zero. These are the hypotheses the researchers are testing. It is often appropriate to reinterpret these hypotheses in normal language. For instance, if the x-statistic represents the difference between males and females on a standardized test score, the null hypothesis would be stating that there are no differences between male and female scores.
3. Locate the report's conclusion. The conclusion states which hypothesis the evidence in the data points to. Most statistical reports give their conclusions in the form "we reject the null hypothesis in favor of the alternative hypothesis" or "we do not reject the null hypothesis." You can reinterpret these conclusions in terms of the original hypotheses. For example, if the statistical study on sex differences on a standardized test rejects the null hypothesis in favor of the alternative hypothesis, you can restate this as "the report showed that there are significant differences between male and female test scores."
4. Find the study's p-value. The p-value is arguably the most important piece of information in a statistical hypothesis test, yet many non-statisticians ignore it in their interpretations. The p-value is usually located in the conclusion of a statistical report, often implicitly inside parentheses [e.g. "We reject the null hypothesis (p = 0.03)"]. Although the mathematical definition of a p-value may be complicated for laymen, an easy way to interpret it is the "strength" of the evidence for the conclusion. For rejecting the null hypothesis, p-values closer to zero show stronger evidence, while for accepting (or not rejecting) the null hypothesis, p-values closer to one show stronger evidence.
Tags: null hypothesis, alternative hypothesis, reject null, reject null hypothesis, between male, between male female
