Statistics & Data Analysis
Answer questions about your variables and design to find a suitable statistical test.
What do you want to find out?
What kind of outcome?
Numbers: scores, times, weights. Ranked: Likert items, positions. Categories: yes/no, groups.
How many groups or conditions?
Are the measurements from…
Is the data roughly normal?
Use the Descriptive Statistics Calculator: strong skew or outliers suggest “no”.
Usual choice
Mann–Whitney U test
Two separate groups, when the outcome is ordinal or not normal. Check normality first (a histogram, Q–Q plot or Shapiro–Wilk test).
Check these assumptions
If they don’t hold: Independent-samples t-test if both groups are normal
How to report it: U = value, z = value, p = value, with each group’s median.
It gives the textbook first choice for simple designs. Studies with covariates, several outcomes, nested data or repeated measures over time need more: check with your supervisor or a statistician.
Students who have data, or a study design, and need to know which statistical test fits.
It follows the standard textbook paths for simple designs. Covariates, nested data, several outcomes and longitudinal designs need a fuller analysis plan.