Decompose the target attribute into components before judging

Break a hard question into its component parts and answer each one directly, rather than letting a heuristic answer the whole.

Why it works

Attribute substitution thrives on undifferentiated hard questions ("Is this a good investment? Is this person trustworthy?"). Decomposing the question — "What is the expected return? What is the downside risk? What is the time horizon?" — fills each component with a specific target, making substitution harder because each sub-question has fewer candidate heuristic answers that fit naturally.

How to do it

1. Take the hard question and list its component attributes (3–5 dimensions that together determine the answer).
2. Answer each component separately, using evidence or deliberate reasoning rather than overall impression.
3. Combine the component answers explicitly rather than trusting a gestalt impression.

Evidence

Decomposition and structured analysis have been shown to outperform holistic judgment in domains from medical diagnosis to credit risk assessment. The mechanism is that decomposition reduces the scope for single-attribute substitution. Grove et al. (2000), meta-analyzing 136 studies, found mechanical (structured, decomposed) prediction equaled or beat holistic clinical judgment in roughly nine cases out of ten — quantifying how much the undifferentiated overall impression costs. (observational)

Decomposition adds cognitive cost and can be over-applied; it is most useful for novel, high-stakes, and statistically complex judgments, not routine decisions.

Sources

• Grove, W. M., Zald, D. H., Lebow, B. S., Snitz, B. E., & Nelson, C. (2000). Clinical versus mechanical prediction: A meta-analysis. Psychological Assessment, 12(1), 19–30.

Common mistake

Decomposing the question but then using overall impression to weight the components — the decomposition only works if each component is answered on its own merits.