Fix visible copy, study-type labels and number formatting

Merged
LU
9h ago
+402-108

From the second usability review of 21 October 2026. This was an agent-led review using simulated perspectives, not recruited participants. GPT 6.1 Sol implemented the changes; Claude reviewed them. This PR fixes C09 and the objective errors in C13, plus related copy findings. It does not touch the outcome or row-unit vocabulary, the boilerplate caveats, or substantive caveat wording; separate proposal PRs cover those.

Changes

  • C09: the broken sentence in both reproduction views of Years & Disciplines is replaced with complete, study-type-specific sentences. A test now checks the assembled text in every study-type view.
  • FAQ: Markdown escapes, raw column names (doi\_r, alt\_id\_r), grammar and spelling are fixed, and so is the "plugin tags…" sentence. The peer-review answer now matches the dashboard's venue categories.
  • Study-type labels: random examples and Browse headers use each row's actual type, so reproductions are no longer labelled "Replication".
  • Developer residue:
    • reader-facing journal_r and the data/disciplines.json link label are gone, along with "straightforward to extend" and "Uncategorized may dominate";
    • "RRDB" is written out as the Registered Reports database, and "rest" is now "Not found in RR database";
    • the placeholder about workflow secrets is replaced with reader-facing text;
    • a leftover "Loading…" element is removed;
    • the unused fect_results.json request, which returned 404 on every visit, is removed. Its only producer is run_fect.R, which nothing deploys.
  • Numbers:
    • thousands separators throughout;
    • a true minus sign in rendered values;
    • whole percentages in captions, and "<1%" instead of "0%" for non-zero shares;
    • chart CSV values rounded to at most 5 decimals.
  • Labels and hovers: wrapped chart labels are joined with spaces in data tables and CSV. Plotly hovers are rounded and labelled.
  • Grammar: "1 reference pair"; lower-case study-type names inside sentences.
  • Names: the full FLoRA name is used consistently, FORRT is expanded once, "comprehensive" is removed from the meta description, and the About page names the "FLoRA Zotero Plugin".
  • Mislabels: the venue selector says "venues", and the timeline instructions name the Timeline button.

Tests

Four new browser scenarios cover the assembled copy, an FAQ fixture, formatting, exports and hovers, and the low-data and unavailable states. Full suite passes locally (20/20).

For review

  • FAQ fixes are applied when the text is displayed, because the FAQ Markdown comes from an external source that stays live. Fixing the source as well would be better.
  • The robustness view's sentence says "how robust" rather than "how reproducible".

🤖 Generated with Claude Code

Summary by cubic

Fixes visible copy, study-type labels, and number formatting issues from the usability review: reproduction views no longer show broken sentences or mislabelled study types, and counts, percentages, and exports are formatted consistently.

Copy fixes

  • Replaces the broken sentence in both reproduction views with study-type-specific text; tests check the assembled sentence in every study-type view.
  • Fixes FAQ rendering at display time — Markdown escapes, raw column names, grammar and spelling, and a peer-review answer aligned with the dashboard's venue categories — because the FAQ source stays live externally.
  • Removes developer-facing text: journal_r, the data/disciplines.json link label, "RRDB"/"rest" labels, the workflow-secrets placeholder, a "Loading…" element, and the unused fect_results.json request that 404'd on every visit.
  • Uses the full FLoRA name, renames "FLoRA Annotator" to "FLoRA Zotero Plugin", expands FORRT once, and fixes the venue selector and Timeline instructions.

Formatting

  • Adds thousands separators and a true minus sign, shows "<1%" instead of "0%" for non-zero shares, rounds chart CSV values to at most 5 decimals, and labels and rounds Plotly hovers.
  • Joins wrapped chart labels with spaces in data tables and CSV exports.
  • Uses integer tick spacing on small count axes so auto-ticks no longer round fractional steps into repeated labels; count formatting is shared throughout Citation Impact; tests wait for a settled axis before asserting ticks.

The FAQ correction warns when an upstream heading change bypasses the venue fix. Eight new browser tests cover copy, FAQ display and drift, formatting/exports/hovers, small-sample counts and integer ticks, and a non-English locale; the full suite passes locally (24/24). Outcome and caveat wording is unchanged; separate PRs cover that.