Files
magnus919_agent-skills/research-methodology/references/synthesis-patterns.md
T
Magnus HedemarkandGitHub e863aab851 feat: add research-methodology skill (#11)
* feat: add research-methodology skill

Signed-off-by: Magnus Hedemark <magnus919@pm.me>

* fix: remove nonportable source assumptions

Signed-off-by: Magnus Hedemark <magnus919@pm.me>

---------

Signed-off-by: Magnus Hedemark <magnus919@pm.me>
2026-07-12 14:37:23 -04:00

60 lines
3.1 KiB
Markdown

# Synthesis Patterns
Methods for combining findings from multiple sources into something more valuable than any single source.
## Thematic Synthesis
Best for: Combining qualitative findings, expert opinions, and case studies.
1. **Extract claims.** From each source, pull the specific claims relevant to your research question.
2. **Find patterns.** Group claims that point in the same direction. What themes emerge?
3. **Name each theme.** A good theme name is specific enough to be meaningful, broad enough to contain related claims.
4. **Support each theme.** List which sources support it, which contradict it, and which are neutral.
5. **Identify convergent and divergent themes.** Where do sources agree? Where do they disagree?
### Convergence Signals
| Signal | What it means |
|--------|--------------|
| 3+ independent sources agree | High confidence in the finding |
| Sources from different fields converge | Very high confidence — cross-validation across disciplines |
| Different methodologies produce same result | Method-independent finding |
| Even critics concede the point | Defensible claim |
### Divergence Signals
| Signal | What it means |
|--------|--------------|
| Sources disagree without clear pattern | Territory is unsettled — report both sides |
| Disagreement correlates with methodology | Methods may be driving results |
| Disagreement correlates with funding source | Conflict of interest may be a factor |
| Single source contradicts consensus | Investigate deeper — the outlier may be wrong or may have found something others missed |
## Lines of Argument
Best for: Building a case for or against a position.
1. **State the conclusion.** What are you trying to support?
2. **Build independent lines.** Each line of argument is a separate chain of reasoning that supports the conclusion. They should be as independent as possible.
3. **Test each line.** Does each line stand on its own evidence? Or do they share assumptions?
4. **Weight the lines.** Some lines are stronger than others. Score each: strong anchoring evidence, moderate supporting evidence, or weak circumstantial evidence.
5. **Cross-line consistency.** If multiple independent lines all point to the same conclusion, confidence is much higher than if they share a common assumption.
## Constant Comparison
Best for: Iterative research where findings emerge gradually.
1. **Start with the first source.** Extract initial themes.
2. **Compare each new source.** Does it confirm existing themes? Add new ones? Contradict?
3. **Update themes.** Refine, split, or merge themes as evidence accumulates.
4. **Test saturation.** Are new sources adding new themes or just confirming existing ones? When new sources stop adding new themes, you've reached saturation.
## The Convergence-Divergence Matrix
Map findings along two axes:
| | Convergent (sources agree) | Divergent (sources disagree) |
|---|---|---|
| **Well-evidenced** (many sources) | ⭐ High-confidence finding — report as solid | ⚠️ Active debate — report both sides with evidence weights |
| **Thinly-evidenced** (few sources) | 📌 Tentative finding — flag as provisional | 🔍 Gap — identify as area needing more research |