A statistical hypothesis test is a method of statistical inference in which two datasets obtained by sampling are compared, or a data set is compared against a synthetic data set based on an idealized model (data distribution) describing a population. A null hypothesis proposes no relationship between these two datasets, with an alternative hypothesis that there is indeed a relationship. In a hypothesis test a test statistic is calculated and compared with a pre-defined critical value, such as the significance level (a). If the statistic falls above the critical value the null hypothesis is deemed to be rejected. In this case it is unlikely that the reported observations would occur if the null hypothesis were true (we cannot ever say that the null hypothesis is certainly ‘false’). By historical convention a critical value of 5% is used for a. A common example is the use of Student’s t test to compare two sample means, summarising the treatment outcome reported for each arm of a clinical trial. The null hypothesis is that there is no significant difference between these values (i.e. no treatment effect), and the distribution of the treatment outcome is assumed to be normal with the same variance in each study arm. If the value of t calculated from the means for the two study arms falls below the reference value (for a given a and sample size) then the null hypothesis cannot be rejected, and it cannot be concluded that there is likely to be a treatment effect.

How to cite: Hypothesis Testing [online]. (2016). York; York Health Economics Consortium; 2016. https://yhec.co.uk/glossary/hypothesis-testing/