Auto Statistical Analysis
Paste your data table and the tool detects the data shape, picks the appropriate statistical method, and returns the statistics, P values and a result chart. Three input formats: grouped summary (each line “label n mean±SD…”, multiple outcome columns — e.g. brain-region ADC by gestational-week group), raw values (one group per line), or two-variable data (each line X Y). With ≥3 ordered groups it also runs a linear-trend test. Everything is computed locally; nothing is uploaded.
① Paste your data table
Three formats: (1) Grouped summary — each line 'label n mean±SD …' (multiple outcome columns allowed); (2) Raw values — one group per line, numbers separated by space/comma; (3) Two variables — each line 'X Y'. Paste directly from Excel.
FAQ
How does it choose the method?
By data shape: a grouped mean±SD summary → one-way ANOVA per outcome column (plus a linear-trend test when groups are ordered); raw values one group per line → t-test for 2 groups, ANOVA for ≥3, each with a non-parametric backup (Mann-Whitney / Kruskal-Wallis); two columns X Y → correlation and linear regression. When skew or small samples are detected, the non-parametric result is preferred.
Can it work from mean±SD only, without raw data?
Yes. It reconstructs one-way ANOVA from each group's mean, SD and n (between/within sums of squares) to give F and P. This is approximate and differs slightly from the exact F on raw data, but is ideal for quick checks and reproducing published results. Missing n is assumed 30 (flagged).
What is the linear-trend test?
When groups are ordered numerically (e.g. gestational weeks 18~20, 20~22…), the between-group variance is partitioned into a linear component (1-df polynomial contrast) and a deviation-from-linearity component. A significant linear-trend P means the outcome rises/falls monotonically; a non-significant deviation P means the relationship is approximately straight-line.
Is the output paper-ready?
The results panel gives a formatted write-up (F, df, P, trend direction) plus the chart. A significant ANOVA (P<0.05) only means at least one group differs; to locate which, use the post-hoc multiple-comparisons tool.