Free AI-Assisted

Statistical Test Chooser

Select defensible statistical tests with automated normality checks, methods text, and exportable R/Python code. Built-in AI agent assistant support.

Open tool workspace

Need help with bioinfo analysis? See our bioinformatics services.

Key facts

Key facts about Statistical Test Chooser
FactValue
TemplatesqPCR, flow cytometry, western blot, survival, RNA-seq (small n), custom
Assumption checksShapiro-Wilk (n≤50), Levene's test, 1.5× IQR outliers
CSV formatsWide (columns = groups) or long (value + group column); max 500 rows
Tests coveredt-test, Welch, Mann-Whitney, ANOVA, Kruskal-Wallis, chi-square, Fisher, log-rank
Post-hocTukey HSD, Dunnett, Games-Howell, Dunn (BH-FDR)
ExportMethods paragraph, R code, Python code, summary CSV
Session historyLast 10 sessions in browser localStorage
Runs in browserYes — uploaded CSV never sent to a server
Account requiredNo
AI assistantBuilt-in; helps fill questionnaire, validate parameters, and interpret assumption badges

What it does

Wet-lab researchers often reach for the wrong test because normality, pairing, and group count are easy to misjudge. The Statistical Test Chooser evaluates your experimental design and uploaded data against formal statistical assumptions, returning a justified test recommendation, effect size metric, methods paragraph, and copyable R/Python code in seconds.

Select an experiment template (qPCR, flow cytometry, western blot, survival, or small-n RNA-seq) or enter your design through a 5-step questionnaire. Uploading a CSV (Wide format with group columns or Long format with a factor column, up to 500 rows) automatically executes client-side Shapiro-Wilk normality testing (n ≤ 50), Levene's variance homogeneity checks, and 1.5× IQR outlier flags across your groups without sending data to a server.

The result panel identifies the optimal primary hypothesis test (t-test, Welch, Mann-Whitney, ANOVA, Kruskal-Wallis, Chi-square, Fisher's exact, or Log-rank) and appropriate post-hoc comparison (Tukey HSD, Dunnett, Games-Howell, or Dunn's test with Benjamini-Hochberg adjustment) for multi-group studies. It generates manuscript-ready methods text citing the specific test and post-hoc logic, exports clean R (rstatix) and Python (pingouin/scipy) scripts, and provides an interactive visual decision path tracing every rule triggered.

Why researchers use it

  • Avoid wrong test selection by verifying distribution assumptions automatically
  • Eliminate manual decision tree lookup errors during peer review preparation
  • Select valid post-hoc tests and effect sizes for multi-group assays
  • Generate publication-ready methods paragraphs citing exact test rationales
  • Export reproducible R and Python scripts without writing code manually
  • Protect sample privacy with zero-upload local browser processing

Best for

  • Selecting t-test vs Mann-Whitney for two-group expression or intensity data
  • ANOVA vs Kruskal-Wallis with Tukey or Dunn post-hoc for 3+ groups
  • Chi-square vs Fisher's exact for categorical count tables
  • Log-rank test guidance for Kaplan-Meier survival comparisons
  • Documenting test choice rationale for manuscripts and lab notebooks
  • Teaching graduate students how assumption checks drive test selection

When to use this vs alternatives

Choose the Statistical Test Chooser when you need instant, guided test selection, assumption diagnostics, and methods text without per-seat software licenses. Use ΔΔCt Clarity to calculate normalized qPCR fold-changes directly, Dose Curve Fitter for non-linear IC50/EC50 regression, or ELISA Plate Mapper for 96-well layout design. Commercial packages like GraphPad Prism or R fit raw execution of complex multi-factor ANOVA models.

What makes it different

Static decision-tree websites explain flowcharts but never inspect your data. GraphPad Prism runs tests but costs per seat and does not auto-generate multi-language code exports. R packages like rstatix automate checks but require programming skill.

The Statistical Test Chooser combines a life-science template picker, guided stepper UI, live assumption diagnostics on CSV upload, transparent decision-path trace, and one-click methods plus R/Python export—free, browser-based, and private. Researchers switch when static flowcharts fail to inspect real sample distributions, spreadsheet software lacks built-in normality checks, or paid software per-seat costs restrict lab access.

How to get started

  1. Open the workspace on the Navigator tab and select an experiment preset from the Experiment template dropdown, or click Load example.
  2. Step through the 5-step questionnaire to specify your Outcome variable type, Comparison design, Number of groups, and Paired or unpaired structure.
  3. Drag and drop a CSV file into the Upload CSV area to automatically run Shapiro-Wilk normality and Levene variance checks.
  4. Review the Recommended Test, post-hoc recommendation, effect size metric, and interactive Decision Path trace in the result panel.
  5. Click Copy methods text, Copy R code, Copy Python code, or Download summary CSV to export your results.

Frequently asked questions

Does this tool run the statistical test and return a p-value?
No. It recommends the appropriate test based on your design and assumption checks, then exports code you can run in R or Python. This keeps the tool fast, transparent, and aligned with how manuscripts report analyses.
Why does a small sample size (n < 5) trigger non-parametric test recommendations?
When any group has fewer than five observations, normality tests have very low power. The navigator defaults to rank-based tests in that case, following common biostatistics practice for small-n wet-lab replicates.
What CSV formats can I upload for automated assumption testing?
Wide format: each numeric column is a group (typical for qPCR or densitometry exports). Long format: one numeric value column plus one categorical group column. The parser auto-detects format or you can toggle manually.
Is my data uploaded to Pepkio servers?
No. CSV parsing, Shapiro-Wilk, Levene, and outlier detection run entirely in your browser. Session history is stored only in your device localStorage.
Can I use an AI agent with the Statistical Test Chooser?
Yes. The workspace includes an in-browser AI assistant to help fill the questionnaire and interpret assumption badges. External agents can call the tool via MCP using pepkio_biostat-test-navigator.

Client source code & registry

Open tool workspaceBack to tools

Last updated . Pepkio builds free lab calculators alongside bioinformatics CRO services.