Cross-tabulation: test evidence
Every requirement of this module, the tests named for it, and what each test recorded the last time the evidence was rebuilt.
Requirement matrix ↗ — the rows these tests are named for.
- Requirements
- 39
- Tests named for them
- 39 14 unit · 25 browser
- Result
- pass every test passing
- Screenshots
- 10
- Recorded
- 2026-10-09 12:12 UTC
- Environment
- linux 6.17.0-1022-azure · node v22.23.3 · playwright 1.61.1 · chromium 149.0.7827.55
The requirements are the rows of the module’s requirement matrix. A test is named for a requirement by starting its name with the requirement’s ID. The results are read from the committed evidence set; continuous integration reruns every test and fails when the committed set no longer matches.
Requirements and their tests
39 requirements · 39 tests
| Requirement | Tests and evidence |
|---|---|
CT-KIT-001pass The chart is built from safety.viz's kit on the page and draws its bars with the kit's Chart.js, bundling neither; without safety.viz on the page it is refused with a message saying what is missing. | |
CT-CFG-001pass Every setting has a default; | |
CT-DATA-001pass The counts, the row and column totals and the grand total are desktop R's | |
CT-DATA-002pass Each cell's row and column percentages are its count of its row's or its column's total, as desktop R works them out. | |
CT-DATA-003pass A biomarker cut by the shared cut rule makes the rows or the columns, low to high, worked out on the participants the filters keep. | |
CT-DATA-004pass Each percentage is printed to one decimal as R's | |
CT-DRAW-001pass Arm by response on the synthetic study: every count, the totals and each row's percentages on the page are desktop R's, and the stacked bars are the same percentages, drawn to 100. | |
CT-DRAW-002pass With column percentages each cell's percentage is of its column's total and the bars stack each column by the rows; with none there are no percentages. | |
CT-DRAW-003pass Response by CRP at Baseline cut at its median: the columns are the cut's groups, low to high, with desktop R's counts; the footnote states the cut and that every participant the filters keep with a value is cut, whether or not they have a category the other way; the Columns control offers the cut variables the settings name. | |
CT-STAT-001pass R is asked once per table, with one row per participant, the function, the arguments, the identity and the row count that desktop R's recipe writes; the categories are named in the order the table draws them, a cut's low to high and a column's by name with numbers as numbers and A to Z read as a to z (gsm.bio's order), the same in every browser language. | |
CT-STAT-002pass R's chi-square and Fisher results are printed with their method and counts, labelled exploratory and unadjusted, and Fisher's odds ratio with its interval, named by which row is over which and the odds of which column. | |
CT-STAT-003pass The small-expected-count warning is R's: printed with the result as R worded it when R gives it, and not otherwise. | |
CT-STAT-004pass Handed to a connection as stored results, each of desktop R's answers is found by the chart's request for its table. | |
CT-STAT-005pass With R's answers stored in the page, the chi-square and the Fisher results print with their method and counts, Fisher's odds ratio with its interval, and what the chart asked is the key R wrote. | |
CT-STAT-006pass R's small-expected-count warning is printed with the result for a table with an expected count below 5 (response by CRP at Baseline cut at 8), and not for one without. | |
CT-STAT-007pass With no R attached the line says statistics are unavailable, the table and the bars are still drawn, and nothing is fetched; with no test chosen R is asked nothing. | |
CT-STAT-008pass The test is of the participants the filters keep, as desktop R gives it for the women alone, and the line says the filter. | |
CT-STAT-009pass Once the connection is replaced, a late answer from the old one changes neither the line nor what | |
CT-STAT-010pass A table R withholds, a category below its minimum size, prints R's reason with the table's variables named where R names the columns the chart handed it, and no number. |
|
CT-STAT-011pass In the page, a withheld table's line reads R's reason with the variables named, and no p-value. |
|
CT-STAT-012pass After a filter changes, the answer to the table no longer on screen is dropped when it arrives, and the line and |
|
CT-STAT-013pass Changing what the percentages are of redraws the table and the bars and asks R nothing: the line keeps R's answer for the table. |
|
CT-STAT-014pass A column's categories are drawn in one order in every browser language, numbers in their names as numbers (2 mg before 10 mg), and R is handed them in that order, so Fisher's odds ratio is of the table drawn, on the side of 1 the drawn table is, and names its rows and columns. |
|
CT-STAT-015pass In the page, with the arms named as doses, the table draws 2 mg before 10 mg, R is handed the rows in that order, and Fisher's odds ratio prints as R gave it for the table drawn, named 2 mg over 10 mg. | |
CT-STAT-016pass Fisher's infinite odds ratio, stored by desktop R as | |
CT-STAT-017pass A table with a category below R's minimum group size prints Fisher's exact test as R computed it, the p-value with its method and counts and the odds ratio with its interval, while the chi-square test of the same table still prints R's reason; R's note that the minimum was not applied names the table's variable, not the column the chart handed R. |
|
CT-STAT-018pass In the page, for a table with a column below R's minimum group size, the line prints Fisher's p-value, odds ratio and R's note with the table's variable named, and at 390px the page does not scroll sideways; a p-value R gives as 1 prints |
|
CT-LIST-001pass Clicking a count lists that cell's participants, the ones desktop R puts in that cell, in the kit's listing, and a row of it opens safety.viz's participant profile of that participant. | |
CT-FAIL-001pass When drawing fails the chart says so in its footnote, empties the listing and the rail, keeps its controls and leaves nothing half drawn, and draws again once it can. | |
CT-DROP-001pass With a participant table, participants it does not have and rows with no participant id are counted by reason; a participant table without the id column is refused with a sentence; filters that let nobody through say so; the id column and the table change together. | |
CT-MOBILE-001pass At 390 pixels the controls are folded away, the table fits or scrolls inside its own box, and the page does not scroll sideways. | |
CT-SITE-001pass The live demo holds at a 390-pixel-wide viewport with no horizontal scroll, with the controls open. | |
CT-SITE-002pass The gallery lists the chart with links to its live demo, its evidence page and its API reference; the demo opens on arm by response, draws the table where R cannot be reached and says so, and offers CRP at Baseline cut at its median. | |
CT-SITE-003pass The API reference has a table row for every setting the chart has, with the default the code has; the site build fails when one is missing. | |
CT-SITE-004pass On the demo page, whose stylesheet styles every table, heading cell and cell by its tag, the table of counts keeps the chart's own look: its face, size, weight, spacing, rules, ground and margins are the same, part for part, as on a page with no stylesheet, with no capitals and no letter-spacing, and the table is the size it is with the page's stylesheet switched off. | |
CT-LIVE-001pass Against R started in the browser on the demo: arm by response, by chi-square and by Fisher's exact test, every count, statistic and p-value equals desktop R's within 1 part in 10^8. | |
CT-LIVE-002pass Against R in the browser: response by CRP at Baseline cut at its median, by both tests, and cut at 8, where R warns of small expected counts, equal desktop R's, warning included. | |
CT-LIVE-003pass Against R in the browser: the women alone, by the filter, equals desktop R's. | |
CT-LIVE-004pass Against R in the browser: response by CRP at Baseline cut at 9.5 and at 10, three and two participants above the cut, by Fisher's exact test equals desktop R's, and by chi-square the table cut at 10 is withheld with R's reason. |
Visual evidence
Every screenshot below is a committed baseline: the same picture is what the browser tests compare the page with and what is shown here. Click one for the picture at its full size.

CT-DRAW-001— arm by response
CT-DRAW-001— as the gallery shows it
CT-DRAW-003— response by crp median
CT-STAT-005— fisher with odds ratio
CT-STAT-006— small expected counts
CT-LIST-001— participants of a cell
CT-MOBILE-001— the table on a phone
CT-SITE-001— demo on a phone
CT-LIVE-001— r in the browser
CT-LIVE-004— fisher on a small margin
Checking this page
npm ci
npm run evidence:check # rerun every test and compare with the committed evidence set
npm run evidence # rebuild docs/evidence/cross-tab/evidence.json