Output & export
Every analysis run drops a report into the output tree on the run page. A procedure can be run repeatedly, all results stay for the session, and reports export in three formats.
The output tree
- Each run appends a card, newest first; the previous run is never overwritten;
- cards default to newest expanded, the rest collapsed; click a title or the arrow to expand or collapse — manual state wins, and running again does not pop old cards open;
- the tree lives for the current session only (not saved into the project, not carried across launches) — reports can always be re-run, so stale numbers are never read;
- the header shows a count, and "Clear output" asks for confirmation before emptying the current procedure's cards;
- each card offers copy, share (plain text), export (PDF / HTML / Markdown) and delete.
Export
| Format | Notes |
|---|---|
Keeps heading levels, zebra tables, running header and "page x / y"; numeric columns right-aligned, p-value columns starred and coloured; Chinese uses a bundled subset font (StatsSansSC, OFL-1.1), so no system font is needed | |
| HTML | Single file with inline styles — open it in a browser or mail it as is |
| Markdown | Plain-text tables, easy to paste into a wiki or commit to a repository |
All three write into the application documents directory and then open the system share sheet. "Export all" in the app bar merges every card of the current procedure into one file: each report is preceded by a "title · generated at" divider, in the same order as the output tree (newest first).
Table styling
Screen, HTML and PDF share one set of rules, so a table looks the same wherever it is read:
- Numeric columns right-aligned: recognises
12.5,.023,<.001,≤0.05,1.2e-05,1,204,12.5%, and ignores the trailing stars a correlation coefficient carries (0.62 **); - p-value columns:
p < 0.05earns stars (*/**/***) and a colour by significance level; - Variable and value labels: tables show a variable's label where it has one (falling back to the name), frequencies and crosstabs show value labels (for example
1 → Male), and report notes spell out the "label = variable name" mapping so nothing has to be guessed.