About StatLab
Free and open for students and researchers. Built for university students and researchers in Malaysia. StatLab is a free statistics workbench by Jupiter2u: no fee, no advertising, no sign-up needed. It covers the everyday procedures of a statistics course and of thesis research, with SPSS-style tables and APA-style summaries, and it grows wave by wave towards the full feature set of desktop statistics packages.
How does it work?
The whole program runs in your web browser. Data files you open are read and analysed on your device and are not uploaded. Optional accounts (username and password) let you store encrypted copies of projects you choose to save.
How is accuracy checked?
Every validated procedure is run on reference datasets and compared, value by value, with independent results from the open-source scientific libraries scipy and statsmodels, to a relative tolerance of one in a million: 433 reference test cases, 36,554 values checked against scipy/statsmodels/numpy, 0 failures. The report is public: validation.json. The SPSS .sav reader and writer are checked against files read and written by ReadStat (pyreadstat).
Is StatLab related to SPSS?
No. StatLab is an independent, clean-room implementation with its own code and design. It follows widely taught statistical conventions so that students can use their textbooks and lecture notes. SPSS is a trademark of IBM Corp.; StatLab is not affiliated with or endorsed by IBM.
Third-party software and licences
| Component | Used for | Licence |
|---|---|---|
| SheetJS Community Edition 0.20.3 | Reading and writing Excel and CSV files | Apache License 2.0 |
| StatLab code (statistics engine, .sav reader/writer, charts, interface) | Everything else | © Jupiter2u |
Statistical methods follow published algorithms, implemented independently: Royston (1995) for the Shapiro-Wilk test, Copenhaver & Holland (1988) for the studentized range (Tukey HSD), Lentz continued fractions for the incomplete beta and gamma functions. Practice datasets are synthetic and released under CC0 1.0.
Disclaimer
Results are provided as-is, without warranty. StatLab is checked carefully, but you remain responsible for choosing appropriate methods and for verifying results before publication or decisions; consult a statistician for important work.
Feedback
See the feature list and roadmap and the user guide.