Tables & Figures (jedpsych-tables-figures)

In the Journal of Educational Psychology, exhibits must carry the quantitative argument for nested, model-based results: multilevel/SEM estimates, effect sizes with confidence intervals, mediation paths, and growth trajectories. They follow APA 7th-edition conventions and — because review is masked — must not reveal author identity (school names, project sites, identifying acknowledgments). A good JEP figure makes the learning effect, its uncertainty, and its mechanism legible at a glance.

When to trigger

• Designing the main results table/figure (model results, mediation, growth)
• Deciding what goes in the article vs. online supplemental material
• A reviewer found an exhibit unclear, non-APA, or identity-revealing
• Visualizing trajectories, variance components, and uncertainty (not just means)

Principles

1. Show model results, effect sizes, and uncertainty. Tables report estimates with standard errors and confidence intervals, variance components/ICC for multilevel models, and fit indices for SEM — not just stars. Figures display trajectories or effects with CIs, not bare bar-of-means.
2. Self-contained + APA 7th. Titles, notes, variable definitions, Ns at each level, and units make each exhibit intelligible alone; follow APA 7th table/figure formatting (including a clear note row).
3. Make the effect interpretable. Where possible, annotate the educational meaning (months of progress, percentile shift, percent variance explained) so the magnitude is legible to readers and policy audiences.
4. Earn the space. Push secondary exhibits (full covariance matrices, every robustness model, measurement details) to online supplemental material; keep the article focused on the contribution.
5. Anonymized + reproducible + accessible. No identifying site/school names in exhibits or notes (masked review); values generated by the shared analysis script; colorblind-safe and grayscale-legible.

Worked micro-example — the main results exhibits (illustrative)

For the cluster-randomized reading trial, two exhibits carry the argument the prose summarizes.

Table 1. Two-level model of transfer comprehension.
  Rows:    intercept, treatment (classroom level), pretest covariate,
           variance components (student, classroom), ICC.
  Columns: estimate, SE, 95% CI, standardized effect (g).
  Note:    defines levels and Ns (48 classrooms, 1,089 students), the
           outcome metric, and that intervals are 95% CIs; no site names.

Figure 1. Adjusted transfer-comprehension by condition, with mediation.
  Geometry: classroom means + 95% CI (dot/interval), NOT a bar of means;
            inset path diagram for the monitoring mediator (a, b, indirect).
  Annotation: g = 0.23, 95% CI [0.06, 0.40]; ~2.0 months of progress.
  Source:   rendered by the deposited R script so values match Table 1.

Exhibit triage — article vs. online supplemental material

Exhibit-stage reviewer pushback and the venue fix

• "Table reports only stars" → add SE, CI, and a standardized effect column; this is the post-reform expectation.
• "Bar chart hides the spread" → switch to dot/interval with 95% CIs; show cluster means where N allows.
• "No ICC / variance components shown" → report them; reviewers check that nesting was modeled.
• "Figure names the school district" → strip identifying labels for masked review.
• "Figure values don't match Table 1" → regenerate both from the single deposited script.

Exhibit calibration anchors

• Because JEP results are model-based, the table is where the nesting (ICC, variance components) and the effect size with its CI actually live; design it to stand alone if an editor reads only the exhibits.
• A growth figure should show trajectories with uncertainty bands, not just endpoint means; a mediation figure should make the indirect path and its CI visible.
• Masked review is easy to break in exhibits — site names, IRB identifiers, or a recognizable program logo in a figure can de-anonymize the paper; scrub them.
• Accessibility is part of credibility: colorblind-safe palettes and grayscale-legible encodings.

Execution bridge (StatsPAI / Stata MCP)

Generate exhibits from the fitted result, not by retyping numbers (the usual source of body-vs-supplement drift). Full map: execution-with-mcp. JEdPsych mixes field/lab experiments and observational school data; multilevel (student-in-class-in-school) inference and many-outcome corrections matter most.

• 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.

Anti-patterns

• Bar plots of means that hide distribution, uncertainty, and nesting
• Tables reporting only stars/p-values with no effect size, SE, or CI
• Omitting ICC / variance components for a multilevel result
• Identity-revealing labels (school, district, site) during masked review
• Exhibit values that don't match the shared analysis script

Output format

【Main exhibit】what it shows + why a table/figure
【Model detail】effect size + CI + variance components/ICC (or SEM fit)? [Y/N]
【Educational meaning】magnitude annotated (months/percentile/variance)? [Y/N]
【APA 7th + self-contained + anonymized?】[Y/N]
【Article vs supplement】split decided
【Reproducible + accessible?】matches script, grayscale/colorblind-safe? [Y/N]
【Next】jedpsych-writing-style

Supplementary resources

• ../../resources/external_tools.md — papaja, ggplot2, plotting and APA-table tooling
• ../../resources/official-source-map.md — APA 7th style and masked-review requirement