Forest Plot

The forest plot displays effect estimates and their 95% CIs for multiple studies or subgroups, with a pooled summary estimate at the bottom. It is the standard display for meta-analyses and subgroup analyses. The pooled estimate diamond, the heterogeneity statistics (Iﾂｲ, Q test p-value), and the model choice (fixed vs. random effects) are all mandatory disclosures.

STEP 1 —When to Use

Use a forest plot when: you are presenting a meta-analysis of multiple studies; you are showing subgroup analyses within a single study (e.g., effect by age group, sex, disease stage); you want to visually display how individual study estimates relate to the pooled estimate; you are comparing effect estimates across multiple predictors in a regression model (coefficients with CIs).

Fixed vs. random effects: use random effects when heterogeneity is present (Iﾂｲ > 25%) or when studies are not exchangeable (different populations, designs). Use fixed effects only when studies are genuinely homogeneous and exchangeable. The choice must be pre-specified and justified.

[Image placeholder: Forest plot with 5 studies + pooled estimate. Left column: study names and year. Center: forest plot with horizontal CI lines and squares (size proportional to weight). Summary diamond at bottom. Right column: effect estimate values and CI as text. Reference line at 1.0 (RR) or 0 (MD). Heterogeneity annotation: Iﾂｲ=34%, Q test p=0.19. Model: random effects. Pooled RR=1.42 (95% CI [1.18, 1.71]).]

Forest plot: 5 studies with CI lines + weight-proportional squares, pooled diamond, right-side numeric table, and heterogeneity statistics.

STEP 2 —Journal Requirement

Effect metric must be labeled: RR, OR, HR, SMD, MD, etc.
Model type (fixed / random effects) and the basis for model selection must be stated.
Heterogeneity statistics: Iﾂｲ with 95% CI, Cochran's Q with p-value —both required.
Study weights (as percentages) should be shown (either as numbers or as proportional square sizes).
Reference line at 1.0 (for ratio measures) or 0.0 (for difference measures) must be present.
PRISMA guidelines (for systematic reviews) require the search strategy and inclusion criteria to be referenced.

STEP 3 —Licklider Implementation

Input

Study/subgroup labels, effect estimates, and 95% CIs (or SEs)
Effect metric type: RR, OR, HR, MD, SMD
Model: random effects (DerSimonian-Laird, default), fixed effects, or user-specified
Optional: subgroup headers for grouped forest plots

Output

Horizontal CI lines with squares proportional to study weight
Pooled estimate diamond (width = 95% CI, height = effect size scale)
Reference vertical line at null effect
Right-side numeric table: estimate and CI for each study
Weight column (percentage) for each study
Heterogeneity statistics annotated below the diamond: Iﾂｲ, Q, p
Model type labeled

[Image placeholder: Licklider forest plot with 7 studies (2 subgroups: clinical trials and observational studies). Subgroup headers. Weight-proportional squares. Pooled diamond per subgroup + overall. Right-side table with RR (95% CI) and weight %. Iﾂｲ and Q displayed below overall diamond. Random effects model labeled. Reference line at RR=1.0.]

Licklider forest plot: subgroup headers, weight-proportional squares, per-subgroup and overall pooled diamonds, heterogeneity statistics, right-side numeric table.

STEP 4 —Draft Output (Draft / Needs review)

Forest plot of [effect measure] for [outcome] across [N] studies.
Effect sizes and 95% confidence intervals are shown for each study.
Square size is proportional to study weight. The summary estimate was
calculated using a random-effects model (DerSimonian-Laird method),
chosen due to expected between-study heterogeneity in populations
and interventions. Pooled [RR/OR/HR/MD] = 1.42 (95% CI [1.18, 1.71]).
Heterogeneity: Iﾂｲ = 34% (95% CI [0%, 68%]), Q(4) = 6.1, p = 0.19.
The vertical reference line is drawn at the null value ([1.0 / 0.0]).