---

name: write-results description: Draft the results section from figures and their captions. Builds a coherent narrative from visual evidence. Use when the user types /write_results, when figures exist with captions, or after methods section is complete.

Write Results Section

Draft the results section from figures and their captions. Builds a coherent narrative from visual evidence.

When to Use

• Figures exist in manuscript/figures/
• Figure captions are drafted
• Methods section is complete
• Ready to tell the story of your findings

Prerequisites

• At least one figure with caption in manuscript/figures/figN/
• Methods section reflects the analysis that generated figures
• Project aims defined (to structure narrative around)

Execution Steps

1. Gather Context

Read these files:

• .research/project_telos.md - Aims to structure narrative
• manuscript/methods.md - What analyses were done
• manuscript/figures/*/caption.md - All figure captions
• manuscript/figures/*/*.png|.pdf|.svg - Figures themselves
• manuscript/results.md - Existing draft (if any)

2. Figure-First Approach

Results should be organized around figures, not chronologically.

Map figures to aims:

| Figure | Aim Addressed | Key Finding |
|--------|---------------|-------------|
| Fig 1 | Aim 1 | [Main takeaway] |
| Fig 2 | Aim 1 | [Supports/extends Fig 1] |
| Fig 3 | Aim 2 | [Different aspect] |

3. Results Section Structure

# Results

## [Thematic heading tied to Aim 1]
<!-- 1-3 paragraphs covering related figures -->

[Introductory sentence connecting to methods/aims]
[Main finding with quantitative support]
[Reference to figure: (Figure 1A)]
[Additional details or subfindings]
[Brief interpretation only - save discussion for Discussion]

## [Thematic heading tied to Aim 2]
<!-- Continue pattern -->

## [Additional sections as needed]

4. Writing Standards for Results

Figure-to-Text Relationship:

In Figure	In Text
Visual pattern	Quantitative description
Data points	Statistical summary
Comparisons shown	Effect sizes, p-values
Trends	Direction and magnitude

DO:

• Lead with the finding, not the method
• Include key numbers (means, SDs, p-values, effect sizes)
• Reference specific figure panels (Figure 1A, 1B)
• Use past tense (these are your results)
• One major finding per paragraph

DON'T:

• Interpret or speculate (save for Discussion)
• Include methods details (already in Methods)
• Describe every data point visible in figures
• Use phrases like "It can be seen that..."
• Repeat figure captions verbatim

5. Statistical Reporting Standards

Descriptive Statistics by Data Type:

Data Distribution	Report
Normal/Symmetric	Mean ± SD
Skewed	Median (IQR or range)
Categorical	n (%)

Always include:

• Sample sizes (N or n)
• Central tendency (mean, median)
• Variability (SD, SE, IQR, CI)
• Test statistic (t, F, χ², r, etc.)
• Degrees of freedom where applicable
• P-value (exact or threshold)
• Effect size (Cohen's d, η², r², etc.)

Formatting examples:

"Significant increase in expression (mean ± SD: 2.3 ± 0.4 vs 1.1 ± 0.3; t(28) = 5.2, p < 0.001, d = 1.8)"

"Strong positive correlation (r = 0.72, 95% CI [0.58, 0.82], p < 0.001)"

"Model accuracy was 87% (95% CI [83%, 91%]; AUC = 0.92)"

P-Value Formatting:

✅ Correct formats:
- P = 0.042 (exact value)
- P < 0.001 (when very small)
- P = 0.12 (non-significant, still report exact)

❌ Incorrect formats:
- P < 0.05 (vague, unless truly < 0.001)
- P = NS or P = n.s. (always give exact value)
- P = 0.0000001 (excessive precision)

Effect Size Benchmarks:

Measure	Small	Medium	Large
Cohen's d (mean diff)	0.2	0.5	0.8
r (correlation)	0.1	0.3	0.5
η² (ANOVA)	0.01	0.06	0.14
Odds Ratio	1.5	2.5	4.0

6. Generate Results Draft

# Results

<!-- 
Draft generated by Research Assistant on [DATE]
Based on figures in: manuscript/figures/
Aims from: .research/project_telos.md

⚠️ VERIFY:
- All statistics are accurate
- Figure references are correct
- Nothing is interpreted (save for Discussion)
-->

## [Heading for Aim 1]

[Opening sentence connecting to aim]

To address [Aim 1], we [brief method reference]. Analysis of [N] samples 
revealed [main finding] (Figure 1A). Specifically, [quantitative details 
with statistics]. [Additional observation from Figure 1B].

## [Heading for Aim 2]

[Continue pattern for each aim/figure group]

---

## Figure Summary

| Figure | Caption File | Referenced in Section |
|--------|--------------|----------------------|
| Fig 1 | ✓ | Section 1 |
| Fig 2 | ✓ | Section 2 |

## Verification Checklist

- [ ] All figures referenced in text
- [ ] Statistics match analysis outputs
- [ ] No interpretation (saved for Discussion)
- [ ] Effect sizes included with p-values

7. Caption Guidelines

Each figure should have a caption.md:

# Figure 1. [Short declarative title stating main finding]

**Panel descriptions:**
- **(A)** [What panel A shows, including axis labels]
- **(B)** [What panel B shows]

**Details:**
[Sample sizes, statistical annotations, error bar definitions, 
color/symbol meanings]

**Key finding:**
[One sentence summary of what this figure demonstrates]

8. Results vs Discussion

Belongs in Results	Belongs in Discussion
"Expression increased 2-fold"	"This increase suggests..."
"Groups differed (p < 0.01)"	"This difference may reflect..."
"Model accuracy was 87%"	"This performance compares favorably to..."
"We observed correlation (r=0.7)"	"This relationship supports the hypothesis..."

9. Post-Draft Workflow

Results draft created. Next steps:

A) Review and verify statistics
   Compare numbers against analysis outputs

B) Refine figure captions
   Ensure captions and text complement (not duplicate)

C) Proceed to write discussion
   Use these findings as basis for interpretation

D) Run /next for other suggestions

Any figures missing captions?
⚠️ manuscript/figures/fig3/ has no caption.md

Related Skills

• write-methods - Ensure methods match
• review-script - Verify analysis code
• next - Get next suggestions

Notes

• Results tell WHAT you found, Discussion tells WHY it matters
• Every figure should be referenced at least once
• Numbers in text must match numbers in figures exactly
• Lead with findings, not methods