A statistical method of validating a prediction (i.e., a research hypothesis), by determining that the probability that the null hypothesis (usually stating that the predicted relation between variables does not exist in the population) is very low –meaning that it falls below the previously established low probability threshold (critical value of the test statistic). Rejecting the null hypothesis leads to accepting the alternative (research) hypothesis.

The logic of statistical hypothesis testing (using a sample of scientifically collected, either observed or experimental data) follows these four steps: 1) state the null and alternative (research) hypotheses, 2) collect data and calculate an appropriate test statistic, 3) compare the observed test statistic to a critical value for the test statistic, and 4) state the conclusion based on the difference between the two.