jms-tables-figures

Tables and Figures (jms-tables-figures)

When to trigger

• A regression table is a wall of coefficients with no story or a model figure is missing
• A qualitative paper has rich quotes but no data-structure figure tying them to constructs
• A process model is described in prose but never drawn
• Tables count against the 10,000–13,000-word budget and need pruning
• A reviewer says "I can't follow how the data became the theory" or "the table doesn't answer the question"

The JMS exhibit bar

JMS exhibits must make the argument legible, in whichever idiom. Two house facts shape them: the word count is inclusive of tables, figures, and references, so every exhibit must earn its space; and tables are numbered with Roman numerals, figures with Arabic numerals (verify against current author guidance — 检索于 2026-06；以官网为准). Quantitative and qualitative papers need different exhibit sets — match the set to the design.

Quantitative exhibit set

• Descriptives + correlations (one table): means, SDs, correlations, and reliabilities on the diagonal; flag any near-collinear pairs.
• Measurement model: CFA loadings / AVE / CR where SEM is used — reviewers check discriminant validity here.
• Regression / SEM results: build models hierarchically (controls → main → interactions); report unstandardised coefficients with SEs (and standardised where helpful), N, and fit. Make the focal coefficient visually findable.
• Interaction plot: any moderation hypothesis needs a plotted interaction with simple-slope annotation — a table alone does not convey form.
• Theoretical model figure: boxes-and-arrows mapping one-to-one to hypotheses.

Qualitative exhibit set (first-class at JMS)

• Data-structure figure (Gioia convention): first-order codes → second-order themes → aggregate dimensions, shown as a single visual so a reader sees the abstraction ladder at a glance.
• Representative-quotes table: each second-order theme illustrated with verbatim quotes (attributed to anonymised informants), demonstrating evidentiary depth without dumping transcripts.
• Process / theoretical model figure: the dynamic relationships among aggregate dimensions, with arrows that carry mechanism (and time, for process work) — not a static box diagram.
• Case / informant table: cases, roles, data sources, and counts (interviews, hours, documents) so the evidentiary base is auditable.

Self-sufficiency and house style

• Every exhibit reads on its own: a title that states what it shows, defined variables/constructs, units, N, and notes.
• Place exhibits to serve the argument; avoid duplicating the same numbers in text and table.
• Prefer one well-designed figure to three crowded tables — space is scarce under the inclusive word count.

Execution bridge (StatsPAI / Stata MCP)

Generate exhibits from the fitted result, not by retyping numbers (the usual source of body-vs-appendix drift). Full map: execution-with-mcp. JMS mixes qualitative and quantitative management research; the chain below is for the quantitative-empirical lane.

• Tables: etable (multi-model columns) 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 effect size in interpretable units.

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

Checklist

• Exhibit set matches the design (quant set vs. qual set)
• Quant: descriptives+correlations with reliabilities; hierarchical models with focal coefficient findable; every moderation plotted
• Qual: data-structure figure present; representative-quotes table; process/theoretical model figure; case/informant table with counts
• Theoretical/process model figure maps one-to-one to hypotheses/propositions
• Each exhibit is self-sufficient (title, defined terms, N, notes)
• Numbering follows house style (tables Roman, figures Arabic — verify)
• Exhibits earn their space under the inclusive word count; no redundant tables

Anti-patterns

• Coefficient wall: a regression table with no hierarchy and no visually findable focal effect
• Unplotted interaction: a moderation hypothesis shown only as a product-term coefficient
• Quotes without structure: rich quotes with no data-structure figure linking them to constructs
• Prose-only model: a process/theoretical model described but never drawn
• Transcript dump: pages of raw quotes instead of a curated representative-quotes table
• Redundancy: the same statistics in both text and table, wasting the inclusive word budget

Output format

【Idiom】quantitative / qualitative
【Quant exhibits】descriptives+correlations · measurement model · hierarchical results · interaction plot · model figure
【Qual exhibits】data-structure figure · representative-quotes table · process model figure · case/informant table
【Model figure ↔ hypotheses/propositions】one-to-one? yes/no
【Space】fits inclusive word count? redundancies removed?
【Next step】jms-writing-style