Write to us and a human answers: gheorghe.beschea@overheat.agency. We aim to reply within one business day.
Into a new tab, named after the analysis (for example “Stats — Regression”). Your own sheet is never modified, and the panel tells you the tab name when it finishes. The Go to results button opens it.
All sixteen: descriptive statistics, correlation, covariance, histogram, rank and percentile, t-test (unequal variances / equal variances / paired), z-test, F-test, ANOVA single factor, ANOVA two-factor with and without replication, regression, moving average and exponential smoothing.
Statwise tells you exactly which cells (for example C47 ("n/a")) and offers to skip them
and run anyway. It never guesses silently.
Welch–Satterthwaite, printed in the output so you can check it. This is the formula SciPy and R use. It is normally not a whole number — that is expected.
Regression uses a QR decomposition and reproduces the certified coefficients and standard errors of the NIST Statistical Reference Datasets, including Longley. All other statistics are checked against SciPy in an automated suite that runs before every release.
Up to 500,000 cells in one run. Everything is read once, computed in memory and written once, so a class-sized dataset finishes in about a second; the panel prints how many rows it analysed and how long it took.
Effect sizes with confidence intervals, an APA-style sentence, non-parametric tests (Mann–Whitney, Wilcoxon, Kruskal–Wallis, Spearman), chi-square tests, assumption checks (Levene, VIF, Durbin–Watson) and Holm-adjusted pairwise comparisons after ANOVA. See Pricing.
Statwise needs no test account: install it, open any spreadsheet with numbers, and every feature is reachable. For the Pro features we will issue a reviewer licence key on request — write to gheorghe.beschea@overheat.agency and we answer the same day.
Google Sheets™ is a trademark of Google LLC.