Statistical analyses
Run statistical analyses in EnaResearch on an approved plan: descriptive statistics, t-tests, rank tests, chi-square, Fisher, correlation and regression.
- Admin panel
EnaResearch computes statistics only from a plan you approve first. The plan (test, variables, significance level, hypothesis) is saved before any result exists; approval freezes the dataset version and the data cutoff, and the plan cannot change afterwards. Results are computed with established open-source statistics libraries and stored with an input checksum and the library versions, so the same plan and cutoff always give the same result.
For research use only
Analysis results are for research purposes only and are not for clinical decisions.
Required role
- A member of the study team with the export permission (owners and researchers).
- The study needs a variable dataset. See Datasets and versions.
Available tests
| Test | Use |
|---|---|
| Descriptive statistics | n, mean, SD, median, quartiles and the 95% confidence interval of the mean, optionally by group |
| Independent t-test (Welch) | Means of two groups (equal variances not assumed). Group: sex or a 0/1 variable |
| Mann–Whitney U | Distributions of two groups (rank test) |
| Paired t-test | Mean difference between two measures of the same participant |
| Wilcoxon signed-rank | Difference between two measures of the same participant (rank test) |
| Pearson correlation | Linear relationship between two variables |
| Spearman correlation | Rank relationship between two variables |
| Chi-square independence | Two categorical fields (sex or a variable with at most 10 integer levels) |
| Fisher exact test (2×2) | Independence in a 2×2 table (small samples) |
| Linear regression | A continuous outcome and predictors; at least 5 observations per coefficient |
| Logistic regression | A 0/1 outcome and predictors (odds ratios); at least 5 observations per coefficient |
In predictors, sex enters as male = 1, female = 0.
Plan, approve, run
Create the plan
Open the study → Analyses → New analysis. Enter a Title and the Hypothesis (it cannot change after approval; no personal data). Choose the Variable dataset, the Test, its variables (for example Outcome and Group, or First variable and Second variable, or Predictors) and the Significance level (α), for example 0.05.
Approve the plan
The analysis is saved as a Draft. You can Edit draft or Delete draft. When the plan is final, click Approve plan. Approval freezes the dataset version and cutoff; to change anything later, create a new analysis.
Run
Click Run. The result shows the statistics, the observations used and those missing or out of scope, and any warnings. Run again recomputes; the same plan and cutoff give the same result.
Privacy and warnings in results
- Small groups: no statistics are shown for groups with fewer than 5 observations.
- Contingency tables: cells below 5 are not shown; the test still uses the full table. Another cell may be hidden too, so a small group cannot be worked out from the total.
- Multiple comparisons: when a dataset has several approved analyses, the result shows the Bonferroni-adjusted significance level.
- Chi-square: a warning appears when expected counts are below 5; consider Fisher's test for 2×2 tables.
- Errors explain why a test could not run, for example "The group needs exactly two levels." or "The model did not converge."
Approving and running are written to the audit log.
Related articles
Can't find what you need?
Our support team can help, from setting up a study to exporting a dataset. Please include your organization code.
Exports
Export EnaResearch datasets as pseudonymized files - CSV, Excel, SPSS and R - with a data cutoff, manifest and checksum so every file can be reproduced.
AI assistant
The optional EnaResearch AI assistant suggests analyses, interprets approved results and drafts report text. It is off by default and receives only aggregates.