Wber Tables Figures

Tables and Figures (wber-tables-figures)

When to trigger

• Tables are dense regression dumps a policymaker cannot read
• Significance is shown with asterisks instead of standard errors / confidence intervals
• Effect sizes are reported only in coefficient units, not policy-meaningful magnitudes
• Exhibits are pushing the paper over the 40-page total cap
• An event-study or treatment effect is buried in a table when a figure would carry it

The WBER exhibit standard

WBER exhibits serve two readers at once: the applied economist who checks the design and the practitioner who wants the magnitude and its policy meaning. The best WBER tables and figures make the headline effect legible in policy units — percentage-point change in enrollment, dollars of consumption, cost per outcome — not just a coefficient. And remember the 40-page total cap includes tables, figures, references, and appendices: every exhibit competes for scarce space, so each must earn its place. Lead with the design (balance, first stage, pre-trends) and then the effect; do not bury the main result in column 6 of a kitchen-sink table.

Building exhibits that work

• One question per exhibit. A table that answers "is the design valid?" should not also try to answer "how big is the effect?" Split them.
• Report SEs / CIs, never asterisks. WBER, like serious applied journals, expects standard errors (and ideally confidence intervals) so readers judge magnitude and precision, not a star count. Note the clustering level in the table.
• Translate to policy units in the table or note. Beside the coefficient, give the effect as a share of the control mean, a percentage-point change, or a cost-effectiveness figure. State the units and the baseline.
• Figures for designs and dynamics. Event-study leads/lags, RD scatter with fitted lines and bins, and dose-response curves belong in figures — they show pre-trends, manipulation, and functional form at a glance.
• Self-contained notes. Each note states the sample, the data source (LSMS/DHS/admin), the estimator, the clustering, and the units. A practitioner skimming exhibits should understand the paper without the text.
• Space discipline for the cap. Move secondary specifications to the supplementary appendix; keep the main paper's exhibits to the load-bearing ones.

Exhibit-by-purpose map

Referee pushback mapped to the exhibit fix

• "I can't tell how big this effect is." → Add the control mean and a policy-unit translation (share of mean / pp / cost per outcome) beside the coefficient.
• "Where's the evidence the design is valid?" → Promote balance / pre-trends / density / first-stage from text to a dedicated exhibit.
• "The asterisks tell me nothing about magnitude." → Replace with standard errors and confidence intervals; state the clustering level.
• "This table has too much in it." → Split into a design exhibit and an effect exhibit; move nuisance columns to the appendix.
• "The paper is over length." → Audit every exhibit against the 40-page cap; keep only load-bearing ones in the main text.

Execution bridge (StatsPAI / Stata MCP)

Generate exhibits from the fitted result, not by retyping numbers. Full map: execution-with-mcp. WBER is development economics — RCTs and observational designs in low/middle-income settings; randomization inference + DiD/IV, magnitude in policy units.

• 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.