Sequential Test Simulation Lab
See why repeated peeking changes false-positive rates, using simulated trials you can reproduce.
Standard-normal IID null with known variance, cumulative two-sided z tests. Compares naive repeated alpha, Bonferroni alpha/looks and fixed-final rejection. Educational null simulation, not an approved stopping design.
Using Sequential Test Simulation Lab
See why repeated peeking changes false-positive rates, using simulated trials you can reproduce.
- Try the ready simulation, then choose its seed, trial count and stopping settings.
- Open Details & settings to choose explicit fields, rules or methods. Select a record to inspect its source evidence and add review notes. Import-repair replacements require individual approval.
- Use Preview for result tables, metrics, charts and method assumptions. Review missingness, warnings and unsupported features before relying on a result.
- Download the workflow ZIP for data, task-specific artifacts, settings and source fingerprints. Save review to reopen your decisions with the exact original files.
Example
Run 500 seeded standard-normal null trials with ten cumulative looks. Compare naive repeated testing, alpha/looks adjustment and a fixed-final test. Change one assumption at a time.
Questions & answers
What does Sequential Test Simulation Lab support?
Standard-normal IID null with known variance, cumulative two-sided z tests. Compares naive repeated alpha, Bonferroni alpha/looks and fixed-final rejection. Educational null simulation, not an approved stopping design.
What is in the downloads?
The primary ZIP includes result CSV/JSON, findings, source fingerprints, settings and task-specific artifacts described by the tool. Analysis workflows also include chart SVGs when applicable. CSV formula escaping is on by default; JSON preserves types and exact tagged binary values. Original sources can be downloaded separately.
Can I reopen my work and are files sent to a server?
Processing runs locally in this browser with local libraries and bounded workers. Save review retains settings, decisions and exact source fingerprints; reopen it with the matching original files. Sources are not automatically uploaded or fetched from URLs. Public pages retain the site’s approved ads and optional analytics.
What are the practical limits and review requirements?
Up to eight files and 80 MiB total, normally 25 MiB per file. The line slicer accepts an 80 MiB source and exports up to 25 MiB. Data workflows support 50,000 rows and 300 columns; binary parsers have depth/object guards and analysis simulations have explicit work budgets and a 30-second timeout. Review source evidence, assumptions and the result in your target application; unsupported features are not silently treated as supported.
Related tools
A useful next step for your task.
Everyday CalculatorsSurvey Response AnalyzerSummarize CSV survey answers with explicit missing-data denominators, multi-select counts, ordered Likert scoring, group filters, charts and…
Developer & DataForecast Backtest LabCompare naive, moving-average and exponential-smoothing forecasts against held-out observations, with errors shown for every prediction.
Developer & DataTime-series Anomaly ReviewInvestigate unusual observations with a trailing baseline, visible candidates and a review log.
Developer & DataDistribution Comparison ExplorerCompare two groups using cumulative distributions, quantiles and directly interpretable differences.
Help improve this tool
Report a problem or suggest an improvement
Describe the issue without pasting private tool input. Feedback goes to our admin inbox.