Jep Exhibits For General Readers
Designjep-exhibits-for-general-readersJep Exhibits For General Readers is an Agent Skill for design published in the brycewang-stanford/Awesome-Journal-Skills repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Use when designing figures and tables for a Journal of Economic Perspectives (JEP) article aimed at a broad audience — clarity over density, self-explanatory exhibits. Handles exhibit design; defer prose evidence to jep-evidence-without-equations and overall structure to jep-narrative-arc.
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Inspect the Agent Skill at https://github.com/brycewang-stanford/Awesome-Journal-Skills/blob/HEAD/Journal-of-Economic-Perspectives-Skills/skills/jep-exhibits-for-general-readers/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/jep-exhibits-for-general-readers, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
Exhibits for General Readers (jep-exhibits-for-general-readers)
When to trigger
- A figure or table was lifted straight from a research paper and is dense
- A reader needs the caption and the body text to decode an exhibit
- The piece has too many exhibits, or exhibits that don't advance the argument
- A regression table with coefficients and asterisks is doing the explaining
The JEP exhibit bar
A JEP exhibit must be self-explanatory to a non-specialist and earn its place in the argument. Clarity beats density: the reader should grasp the point from the exhibit and its caption alone, in seconds, without parsing the methods. JEP avoids significance asterisks and coefficient-table presentation in favor of exhibits that show the idea.
Design principles
- One exhibit, one point. Each figure/table makes a single claim that maps to a step in the narrative arc. If it makes three points, split it or cut two.
- Self-explanatory title + caption. The title states the finding ("Pass-through has fallen since 2000"), not the variables ("Price on cost, 1990–2020"). The caption gives source, units, and the one thing to notice.
- Figures over tables for trends and comparisons. A clear line/bar chart usually beats a number grid for a general reader. Reserve tables for a few labeled, rounded quantities.
- Round and label. Plain units, rounded numbers, direct labels on lines/bars instead of a legend the reader must cross-reference.
- No significance asterisks or coefficient dumps. If uncertainty matters, show it as a shaded band or a stated range, not stars (mirrors
jep-evidence-without-equations). - Minimal chartjunk. No 3-D, no needless gridlines, no decorative color. Color should encode meaning or be dropped (and survive grayscale, since print/archives may be monochrome).
- Few, load-bearing exhibits. A short synthesis essay carries a handful of exhibits, each indispensable — not a research paper's full battery.
Table discipline (when a table is right)
- Rows/columns labeled in plain words, not variable names.
- A small number of rounded figures, not full estimation output.
- A one-line note: source, units, sample/coverage, and the takeaway.
- No standard-error parentheses-and-stars apparatus; if precision matters, state a range in the note.
Checklist
- Each exhibit makes exactly one point tied to a narrative step
- Title states the finding; caption gives source, units, and what to notice
- Figures used for trends/comparisons; tables only for a few labeled quantities
- Numbers rounded; lines/bars directly labeled (legend minimized)
- No significance asterisks, no raw coefficient tables
- Uncertainty shown as band/range, not stars
- Exhibit is legible in grayscale; no chartjunk
- Total exhibit count is small and every one is load-bearing
Worked vignette (illustrative)
A draft transplants a research-paper Table 3: eight columns of coefficients, standard errors in parentheses, and rows of asterisks, titled "Regression results." The JEP exhibit replacing it is a single line chart titled with its finding — "Pass-through to consumer prices has roughly halved since 2000" — one line, directly labeled, a shaded band for the range across studies, source and units in a one-line note, and nothing else. The reader gets the point in three seconds, in grayscale, without reading the body — which is the entire test.
Anti-patterns
- A regression table with coefficients, SEs, and *** transplanted from the working paper
- A figure whose point you can't state in its title
- Legends and variable names that force the reader back into the text
- Color or 3-D that carries no information (and breaks in grayscale)
- Ten exhibits where three would carry the argument
- A "kitchen-sink" table of every specification (that belongs in research, not JEP)
Output format
【Exhibit】[figure/table] — one-point claim: [...]
【Finding-title】[...]
【Caption】source / units / what to notice
【Form】figure (trend/comparison) / small labeled table — chosen because [...]
【Uncertainty shown as】band / stated range (no asterisks)
【Grayscale-safe + chartjunk-free?】[Y/N]
【Count check】N exhibits, each load-bearing? [Y/N]
【Next step】jep-writing-style