Statistical hypothesis testing is a clearly defined procedure that consists of the following steps:

  • stating the null and alternative hypotheses,
  • choosing a statistical test,
  • choosing a significance level,
  • calculating the test results,
  • comparing the p-value to the significance level, or the test statistic to the critical value, and
  • interpreting the test results.

If you want to test the relationship between two or more variables, you first need to define a hypothesis about their mutual relationship or difference. In this step, you define a null hypothesis that you hold until evidence shows otherwise.

To test a hypothesis, you need to choose the appropriate statistical test according to the type of data collected and what you want to test (relationship or difference). A statistical test is essentially a formula that represents a series of mathematical operations performed on data collected in a study.

In the social sciences, a significance level of 0.05 is most commonly used, and less commonly a significance level of 0.01. The first case means that if the null hypothesis is true, there is a 5% maximum allowable risk of concluding that a relationship exists when it actually does not (a Type I error), and in the second case, 1%.

The calculation of test results is most often done using software, and SPSS, R, Excel, or some other specialized computer program can be used for this purpose.

If the software does not calculate the probability of such a test result, it is necessary to determine the critical value in statistical tables. If the software calculates the probability of such a test result, it needs to be compared with a predetermined significance level.

As a result of comparing the test value and the critical value, a decision is made to retain (fail to reject) or reject the null hypothesis, and a conclusion is drawn about the significance of the decision within the research. If the null hypothesis is rejected, the specific alternative hypothesis chosen before the test is supported.