Tables, Figures & Qualitative Displays (worlddev-tables-figures)

When to trigger

• Tables report significance with asterisks instead of standard errors / confidence intervals
• Coefficients are shown in model units a non-economist cannot interpret (log points, latent indices)
• A geographic study has no map, or a busy map that communicates nothing
• A qualitative paper has no systematic display — just block quotes scattered through prose
• The main exhibits do not, on their own, let a reader reconstruct the paper's argument

The WD exhibit standard

WD's readership is multidisciplinary and policy-facing, so exhibits must clear a higher legibility bar than in a specialist economics journal: a development scholar from another field, or an informed practitioner, should grasp the headline from the exhibit and its note alone. That changes priorities. Effect sizes in interpretable units beat starred coefficients; a clean figure beats a dense table; a well-built map or qualitative matrix can carry an argument that prose cannot.

Tables

• Report uncertainty, not stars. Standard errors or confidence intervals in parentheses; no ***. WD's audience reads magnitudes, and asterisks invite dichotomous "significant/not" thinking that the journal's evidence culture resists.
• Translate to real units. Convert log points to percent; index effects to a meaningful benchmark (a standard deviation, a mean, a poverty-line equivalent). State the comparison ("relative to the control mean of X").
• One question per table. The title states the question; the note states the sample, the units, the inference, and what is being held fixed.
• Show the magnitude that matters for policy, not only the cleanest specification.

Figures

• Prefer figures for the headline. Coefficient plots, event-study leads/lags, dose-response curves, and distributional plots communicate to a broad audience faster than tables.
• Maps for spatial development data — choropleths for outcomes/treatment, with clear classification, projection, and a note on the unit. A map should answer a question, not decorate.
• Design for grayscale and color-blind safety (Elsevier prints and renders both); never rely on red/green alone.

Qualitative displays

Qualitative WD papers should display evidence, not only narrate it. Use:

• Coding/thematic matrices (themes × cases or × informant groups) to show pattern across the corpus.
• Process-tracing timelines linking events to mechanism steps.
• Illustrative-quote tables that tie each quote to a theme and a source identifier — including, where it matters, a disconfirming quote.
• Typologies that organize cases along the dimensions the argument turns on.

A worked fix (illustrative)

A table reports "0.18***" for a transfer program on consumption. The WD fix: report "0.18 (0.05)" and add a note — "an 18% increase relative to the control-group mean of 1,420 PPP$; clustered at the village level (N=86 clusters)." Better still, move it to a coefficient plot with the control-mean reference line, so a public-health reader sees the magnitude and its precision at a glance.

Execution bridge (StatsPAI / Stata MCP)

Generate exhibits from the fitted result, not by retyping numbers. Full map: execution-with-mcp. World Development is multidisciplinary development studies; the chain serves its quantitative-causal lane, mixed-methods work uses its own standards.

• Tables: etable (multi-model) or did_summary_to_latex straight from the result_id.
• Figures: plot_from_result / enhanced_event_study_plot / event_study_table — axis units and the SE/clustering note baked in.
• Every note names the estimator + clustering and states the magnitude in interpretable units.

See a full fitted-result → exhibit chain in the JF execution walkthrough.

Checklist

• No significance asterisks anywhere; SEs / CIs reported
• Effects in interpretable units, benchmarked to a meaningful quantity
• Each table/figure answers one stated question; notes give sample, units, inference
• Headline result available as a figure, not buried in a wide table
• Spatial data shown on a legible, well-classified map where relevant
• Qualitative evidence displayed systematically (matrix/timeline/quote table), not only quoted in prose
• Grayscale- and color-blind-safe; figures readable at print size

Anti-patterns

• Asterisk-driven tables that train the reader to read "significant vs. not" instead of magnitude
• Coefficients in raw model units no cross-disciplinary reader can interpret
• A decorative map that conveys no comparison, or a missing map for clearly spatial data
• Kitchen-sink regression tables with twelve columns and no signposted headline
• Qualitative evidence reduced to cherry-picked block quotes with no systematic display

Output format

【Journal】World Development (WD)
【Skill】worlddev-tables-figures
【Verdict】reader-ready / revise
【Asterisks】removed; SEs/CIs present? [Y/N]
【Units】effects translated + benchmarked? [Y/N]
【Headline exhibit】figure / map carrying the main result
【Qual display】matrix / timeline / quote table present? [Y/N / n/a]
【Source status】verified URL / 待核实 / not asserted
【Next skill】worlddev-writing-style