Tables & Figures (hrm-tables-figures)
When to trigger
- The correlation table is missing means, SDs, reliabilities, or has inconsistent decimals
- A significant interaction is reported in text but never plotted
- The regression/HLM tables dump every coefficient with no model build-up logic
- The theoretical model in the intro does not match the hypotheses being tested
- A qualitative paper has rich quotes but no data-structure figure
The exhibits HRM expects (and the conventions referees enforce)
HRM follows management/applied-psychology table norms (APA-aligned house style). The standard set:
| Exhibit |
Must contain |
| Table 1 — descriptives & correlations |
Means, SDs, full correlation matrix, scale reliabilities (α) on the diagonal; significance noted; level-appropriate (within/between if multilevel) |
| Table 2+ — regression / HLM / SEM |
Nested model build-up (controls → main effects → interactions); unstandardized and/or standardized coefficients with SEs; model fit (R², ΔR², pseudo-R², CFI/RMSEA for SEM); df and N at each level |
| Interaction plot |
Simple slopes at ±1 SD, axes labeled in construct units, the moderator legend clear, region of significance where relevant |
| Theoretical-model figure |
Boxes and arrows mapping one-to-one to the numbered hypotheses |
| Mediation figure |
Path coefficients on the diagram; indirect effect + bootstrap CI reported |
| Qualitative data-structure figure |
First-order codes → second-order themes → aggregate dimensions (Gioia-style) |
Make exhibits carry the argument, not just the numbers
-
The correlation table is the credibility table. Reviewers read it first; reliabilities below ~.70, a correlation near 1.0 between "distinct" constructs (discriminant-validity red flag), or a mean at a scale ceiling all undermine the paper before the hypotheses are tested.
-
Build models, don't dump them. A nested progression shows the incremental variance the focal effect explains over controls — that ΔR²/Δ-2LL is the contribution made visible.
-
Always plot a supported interaction. A coefficient is not interpretable as "the effect strengthens"; the plot is. Label axes in real construct units, not z-scores, so an HR reader can see the practical magnitude.
-
The model figure is a contract. Every arrow must be a hypothesis and every hypothesis an arrow; mismatches read as sloppiness or HARKing.
-
Translate magnitude for practice. Where possible, annotate the practically meaningful difference (e.g., the predicted productivity gap between low- and high-HPWS units) so the exhibit serves HRM's practice mandate.
Formatting discipline
- Self-contained titles and notes: a reader should understand each exhibit without the text (N, level, what significance markers mean, abbreviations defined).
- Consistent decimals (typically two) and consistent variable names across all tables and the text.
- Report effect sizes and CIs, not only stars; do not let asterisks substitute for interpretation.
- Place exhibits per Wiley/ScholarOne submission conventions; keep figures legible in greyscale.
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. HRM is empirical HR — multilevel survey data, field experiments, and panels; multilevel 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.
Checklist
Anti-patterns
-
Missing reliabilities: a correlation table with no α on the diagonal
-
Coefficient dump: one mega-table with no model build-up
-
Unplotted interaction: a claimed moderation never shown graphically
-
Figure–hypothesis mismatch: arrows that don't correspond to numbered hypotheses
-
Star-only reporting: asterisks instead of effect sizes and CIs
-
Z-score axes: interaction plots no HR reader can map to practice
-
Orphan exhibits: tables that cannot be read without the surrounding text
Output format
【Journal】Human Resource Management (Wiley "HRM")
【Skill】hrm-tables-figures
【Table 1】M/SD/correlations/reliabilities present? [Y/N]
【Model tables】nested build-up + fit + ΔR²/Δfit? [Y/N]
【Interactions】all supported ones plotted, construct-unit axes? [Y/N]
【Model figure】one-to-one with hypotheses? [Y/N]
【Mediation/qual】path CIs / data-structure figure present? [Y/N]
【Magnitude】practitioner-meaningful annotation added? [Y/N]
【Next skill】hrm-writing-style
