Which statistical test should I use?

Download a flow chart and table to help choose the right analysis and there is a handout explaining the terminology.

To answer this question, you need to know the parameter you are comparing e.g. mean and whether you need a parametric test or non-parametric test. Parametric tests are more powerful but can only be used for continuous variables that are normally distributed. For more information.

The following table is a guide to help you decide on an appropriate test with links to websites containing further information on each topic. A brief description of the main sites used is given below the table to help you decide which is best for you.

Summary of tests

Comparison of means Parametric (means) Non-parametric (medians)
Differences between the means of two independent groups Independent t-test Mann-Whitney test
Differences between paired (matched) samples e.g. weight before and after a diet for each subject Paired t-test Wilcoxon signed rank test
Differences in the means of 3+ independent groups for one variable One-way ANOVA Kruskal-Wallis test
Differences between 3+ measurements on the same subject Repeated Measures ANOVA

Friedman test

Relationships between variables



Strength of a relationship between 2 continuous variables Pearson's Correlation Co-efficient Spearman's Correlation co-efficient
Predicting the value of one variable given the value of a predictor variable Simple Linear Regression
Assessing the relationship between two categorical variables. Chi-squared test

If there are numerous independent variables, use multivariate techniques.

Quick description of main websites used:

[1] and [SPSS]. Laerd Statistics
This site is very useful as there are fairly detailed but easy to read descriptions of each test and how to do it in SPSS, (version 18). Unfortunately, there are some differences in the processes for non-parametric tests in SPSS 19 but the information is still useful when it comes to understanding the tests and interpreting the output.

[2]. MLSC, Loughborough Uni Stats help
Loughborough University stats support handouts containing calculations by hand with limited SPSS instructions and interpretation. A wide range of topics are covered.

[3]. Stattrek
Clear summary including the maths behind the tests. Links to keywords within the text. The introduction to each test includes the conditions needed for the test. It also contains video lessons for some topics.

[4]. Steps glossary at Glasgow Uni. This website gives useful summaries of each non-parametric test. It is also good for basic stats definitions.

[5]. Microbiologybytes This website has videos with explanations of each test and how to carry out the test in Excel and SPSS. Unfortunately, both are out of date but can still be used by people with old versions of these programs. Excel requires the Analysis Toolpak to be installed or activated through the Home  Excel options for the latest version of Excel.

For all enquiries, feedback or comments you can use our contact form or email: mash@sheffield.ac.uk