Comparative Analysis

Comparative analysis is a structured side-by-side: two or more theories, methods, studies, or positions, evaluated against the same criteria. The mindmap's sibling-node structure is purpose-built for this — it lets you see the comparison instead of reading paragraphs about it.

When to use it

  • Comparing theoretical frameworks ("which of these explains the data better?").
  • Comparing methods or instruments ("which approach is appropriate here?").
  • Comparing studies that reach different conclusions on the same question.
  • Comparing your work against prior work.

If a chapter or section of your eventual writing is going to be a comparison, building that section as a comparative subtree first usually saves time.

Step 1 — Pick the items and the criteria

Both halves matter. The criteria are what make the comparison meaningful — without them, you have a list, not an analysis.

Examples of criteria sets:

  • For theories: explanatory scope, parsimony, predictive power, fit with primary evidence, openness to falsification.
  • For methods: sample assumptions, robustness to noise, computational cost, interpretability.
  • For studies: sample size and quality, methodology, key findings, limitations, replication status.

Start with 4–6 criteria. Too few and the comparison is shallow; too many and it's exhausting.

Step 2 — Build the scaffold

Create a structure like:

Comparison: Cline vs. Drews on the Late Bronze Age collapse
├─ Cline (Systems collapse, 2014)
│   ├─ Explanatory scope
│   ├─ Primary evidence used
│   ├─ Methodological approach
│   └─ Limitations / critics
├─ Drews (Sea Peoples invasion, 1993)
│   ├─ Explanatory scope
│   ├─ Primary evidence used
│   ├─ Methodological approach
│   └─ Limitations / critics
└─ Cross-comparison
    ├─ Where they agree
    ├─ Where they diverge
    └─ Adjudication

You can build this manually or ask the Editor to create the scaffold:

Create a comparative subtree for Cline (2014) and Drews (1993). Use these criteria as children of each: explanatory scope, primary evidence, methodology, limitations.

Step 3 — Populate each side

For each item being compared, run a focused agent instruction:

Select the Cline (2014) subtree. Fill in each criterion node using only what's in the KB, with citations.

Then do the same for the other side. Keep instructions parallel — the comparison is only as good as the symmetry of effort that went into each side.

Step 4 — Cross-comparison nodes

This is the interpretive layer. Select the "Cross-comparison" parent and ask:

Looking at both subtrees, identify where the two positions overlap, where they diverge sharply, and where one is more strongly supported than the other. Create child nodes for each.

The agent surfaces the structural differences; you decide whether they matter.

Step 5 — Adjudicate (or don't)

Some comparisons need a verdict, some don't. If yours does, write the adjudication node yourself — that's a judgment call, not a synthesis task. The agent can help by stress-testing your reasoning:

Critique the following adjudication: your draft. What are its weakest points given the evidence on this subtree?

Common mistakes to avoid

  • Asymmetric effort. If side A has twice as many citations as side B, your comparison is biased before any conclusion is drawn. Force symmetry.
  • Comparing on criteria that favor one side by construction. Reuse criteria that both sides at least claim to address. If a criterion is only relevant to one approach, it belongs in a "unique strengths" appendix, not the main comparison.
  • Skipping the cross-comparison nodes. Without them, you have two parallel summaries, not an analysis.

What's next