Argument by Analogy

An ampliative argument form: from similarities between two cases on known properties, infer they share an additional property.

1. A and B share properties P1, P2, ..., Pn.
2. A also has property Q.
∴ 3. B (probably) also has property Q.

Strength depends on whether the shared properties are relevant to Q.

Two main flavours:

Inductive Analogy: predicts an empirical similarity (e.g., a new drug will work on humans because it worked on mice)
Consistency Analogy: argues for a normative similarity in how we should treat two cases (e.g., if X is wrong here, it’s wrong in this analogous case)

What makes an analogy strong:

Number of relevant similarities: more is better, but only if relevant
Relevance of the similarities: do the shared properties actually bear on Q?
Number and relevance of disanalogies: even one strong disanalogy can defeat the inference
Diversity within the basis: if the analogy holds across many varied cases, it generalizes better

False Analogy is the failure mode: shared properties aren’t relevant to the conclusion, or disanalogies dwarf the similarities.

Example

A country is just a big family; you wouldn’t let strangers into your house, so we shouldn’t let immigrants into the country.

Relevant disanalogies (a country is not a household; citizenship dynamics aren’t family dynamics) sink the inference.