Free tool · Data & statistics
Statistical test chooser
Up to five questions about your data, then the test to use, its non-parametric alternative and why.
Answer up to five short questions about your study, and this tool names the statistical test to use, its non-parametric alternative and why it fits. The ten tests it covers are the ones you will see in most Indian PhD theses. That list goes from the t-test and ANOVA till chi-square, correlation and regression.
Mostly scholars open it while planning the analysis chapter. It is useful later also, when a DC member asks why you took this test and you want to check your reasoning.
Method
How this tool works
Three things decide which test you need. First is your research question. Maybe you are comparing groups. If not, you may be checking a relationship or predicting an outcome. Second is the type of outcome. A continuous outcome is something like marks, income or a scale score. Categorical means yes/no or a district name, and ordinal is a single Likert item. Third is the design of the study. How many groups are there, and were the same people measured more than one time? In the guide we call these the three questions.
The tool puts these questions to you one by one. Questions that don’t apply to your study are skipped. The answers it gives are taken from the decision table in our guide, and they match it row for row.
For every answer you get one parametric test and one non-parametric test. Parametric tests such as the t-test and ANOVA expect the data to be roughly normal inside each group. But if the groups are fairly big and nearly equal in size, the t-test and ANOVA can still handle moderate non-normality. Mann-Whitney U and Kruskal-Wallis are non-parametric, they work with the ranks of the data. If your data is clearly skewed or the samples are small, use the non-parametric test. For an ordinal outcome also use it.
Limits
What this tool can’t do
FAQ
Frequently asked questions
t-test or ANOVA?
Two groups means t-test, and three or more groups means ANOVA. If you have exactly two groups, both tests will give you the same result. With two independent groups we suggest Welch’s t-test, because it works even when the variances are not equal.
My data aren’t normal. What now?
Please don’t jump to a non-parametric test immediately. When the groups are reasonably large and of similar size, the t-test and ANOVA still manage moderate non-normality. First see the plots and look for outliers. Think about your sample size also, and write down the justification for your choice in methodology chapter. If the data is still clearly skewed after that, or samples are small, go non-parametric. Same thing for an ordinal outcome.
Is a Likert item continuous?
A single item is not continuous. It is ordinal. Describe it using frequencies and the median, and for comparing groups use Mann-Whitney U or Kruskal-Wallis. The scale score is a different case. It is the sum or mean of several items measuring one construct, and usually it is treated as continuous after its reliability is checked. The guide’s section on Likert data has more detail on this.
Do I need SEM?
You need SEM only when your model has latent constructs, each measured by several items, and the paths between them have to be tested together. Many scholars are putting SEM because it look advanced. Many times plain regression would answer their research question itself.
Which software should I use?
The software does not matter much to examiners. They check whether the test is right and the reporting is clear. In Indian universities SPSS is common, and R, jamovi and JASP are free to use. Mention the software name and its version in your methodology chapter.
Keep reading
Related guides and services
Choosing the right statistical test
Match your question and data type to the right test, including non-parametric options.
Sample size calculator
Cochran’s formula, the small-population correction and the regression rules of thumb, with every step shown so you can defend the number.
Statistics Training and Analysis Review (SPSS, R, AMOS, SmartPLS)
Learn the right tests on your own data, run them yourself, and get your output and interpretation checked.
Talk it through with a specialist for free
Send us where you are and what you need. We’ll reply within one working day with clear advice, a written scope and a fixed quote. No obligation.
Book a free consultation