Jhe Tables Figures

Tables and Figures (jhe-tables-figures)

When to trigger

• The main result is settled and must be readable at a glance to a health economist
• Tables are dense, over-decimaled, or bury the headline policy effect
• An event-study / RD / first-stage plot needs to carry the identification visually
• The descriptive picture of the health setting (utilization, spending distribution, coverage) is missing or buried

The JHE exhibit bar

At JHE the main health-policy estimate should be findable in seconds, and the exhibits should let a health economist judge both the magnitude and the institutional plausibility of the result. Elsevier/JHE house style permits significance stars, but standard errors in parentheses are the load-bearing object and clustering must be stated. Two exhibit duties are specific to this journal: (1) a descriptive/institutional exhibit that shows the health setting — the coverage gap, the spending distribution (it is skewed — show it), the policy timeline — so readers trust the variation; and (2) the identification figure (event-study leads, RD continuity, first stage) that makes the design visible before the table.

Exhibit craft

1. One table for the headline. The preferred specification — effect, SE in parentheses, N, dependent-var mean, clustering level — should be readable without flipping pages; demote the controls sweep to the online appendix.
2. Show the distribution, not just the mean. Health spending and utilization are right-skewed and zero-heavy; a distribution figure or quantile effects tell the policy story a mean hides.
3. Make the setting legible. A descriptive exhibit naming the program, eligibility, and timing earns referee trust that the variation is what you say it is.
4. Figures carry identification. Event-study leads, RD continuity, and the first stage are more convincing as clean vector figures than as prose.
5. Right precision and self-contained notes. Two to three significant figures; each note states sample, units, clustering, controls, outcome definition, and (if used) what a star means.
6. Name the outcome in plain terms. "Any inpatient admission (0/1)" beats an opaque variable label; a health-policy reader should know exactly what was measured from the exhibit alone.

Execution bridge (StatsPAI / Stata MCP)

Generate exhibits from the fitted result, not by retyping numbers. Full map: execution-with-mcp. JHE is health economics — insurance/program reforms and selection; foreground DiD/IV/RDD and selection corrections.

• 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

• Headline effect readable in one table: coefficient, SE in parentheses, N, dep-var mean, clustering stated
• A descriptive/institutional exhibit makes the health setting and policy timing concrete
• Skewed/zero-inflated outcomes shown as a distribution, not only a mean
• Identification figure present (event-study with CIs / RD with density / first stage)
• Heterogeneity reported by policy-relevant subgroup with MHT, not subgroup fishing
• Notes self-contained (sample, units, clustering, outcome definition, controls)
• Figures clean vector output; precision 2–3 sig figs; no redundant exhibits
• Outcome variables named in plain terms; magnitudes shown against a base rate

Anti-patterns

• A summary-statistics table with no health-system detail, so the variation looks generic
• A 12-column results table where the policy effect is buried in column 9
• Reporting stars or t-stats but omitting standard errors / clustering level
• Mean-only reporting that hides the skew and zero mass in spending/utilization
• An event-study with no confidence intervals or an unclear reference period
• An RD figure with a high-order global polynomial manufacturing a jump
• A heterogeneity wall of starred subgroups with no MHT and no clinical/policy logic
• An exhibit whose magnitude has no base rate, so the reader cannot judge whether it is large