Researchers often wonder which statistical tests they should use with their qPCR data. Most frequently these questions arise when analysing gene expression data.

There are broad groups of statistical tests, parametric and non-parametric. Parametric tests assume the underlying data have normal distribution, whereas non-parametric tests do not.

Parametric tests are far more powerful and sensitive so it is better to use them whenever possible.

This means that before doing statistical tests, we should analyse the data distribution (e.g. with a histogram). In case of an obvious non-normal data distribution (at least one group of samples you are comparing) or if the data were collected in form of rankings rather than scores (“first”, “second”, “third”,..) we have to use the non-parametric tests.

In case of small sample size (<30) it is difficult to check the normality of distribution. Non-parametric tests are a good solution for small sample sizes.
If you are comparing two independent groups of samples (e.g. healthy and treatment) you can use parametric test like t-test or its non-parametric counterpart Mann-Whitney (for repeated measurements use Wilcoxon test).

If you are comparing more than two groups, you should use Analysis of variance ANOVA (a parametric test) or a Kruskal-Wallis (a non-parametric test).
To sum up a few tips: use t-tests (or ANOVA) unless the data are obviously non-normally distributed, are in the form of ranks or you have small sample groups.

And … talk to your bio-statistician as much as possible about your experimental design.


By Matjaz Hren, PhD, COO, Head of Research and Development BioSistemika LLC