What Is Arguing From Analogy? Reasoning From One Case to Another

An argument from analogy notes that two cases share certain features, observes that one of them has a further feature, and concludes that the other probably does too. The conclusion is probable rather than guaranteed, since nothing about shared features entails shared further features, which places this form among the inductive kinds of argument where evidence supports without establishing. That is a limitation rather than a defect, since the same is true of every argument from experience. The form appears constantly in ordinary reasoning, in science where a mechanism known in one system is proposed for another, in law where a decided case governs a new one, and in ethics where an intuition about a clear case is transferred to a contested one. Recognising the structure is the first step in assessing it.

A small set of questions determines how much weight an analogy carries:

• •Relevance, since the shared features must bear on the feature being inferred rather than being merely numerous
• •Number of respects shared, which strengthens the argument when the respects are relevant and does nothing when they are not
• •Number of cases, since an analogy resting on several parallel instances is stronger than one resting on a single comparison
• •Diversity among those cases, since parallels that differ in irrelevant ways rule out coincidental explanations
• •Known disanalogies, which weaken the argument in proportion to their relevance
• •Modesty of the conclusion, since a weaker claim is supported by the same evidence more strongly than a bold one

Legal reasoning in common law systems is built on this form explicitly. A decided case establishes a rule in particular circumstances, and a later court must decide whether the new facts are relevantly similar, which is an argument from analogy conducted openly with both sides arguing about which features matter. Distinguishing a case means identifying a relevant difference. The whole practice assumes that treating like cases alike is a requirement of justice, which is an independent principle that gives the reasoning its force. Ethics uses the form heavily, with thought experiments constructed to produce a clear intuition that is then transferred to a real case, and much of the argument in applied ethics consists of disputing whether the transfer is legitimate. Science uses it in generating hypotheses, where a known mechanism suggests a candidate explanation elsewhere.

Scientific use of the form is distinctive and worth separating from the everyday version. A model organism is an argument from analogy in its entirety, since results in a mouse are taken to bear on people because of relevant similarities, and the assessment questions apply directly, with the relevance of the shared physiology being the point at issue in every case where a treatment works in animals and fails in humans. Physical analogies transfer a mechanism from a well-understood system to a new one, which is how sound was treated as a wave by analogy with water, and such analogies are generative even when they later prove imperfect. The danger is that an analogy used to generate a hypothesis gets treated as evidence for it, which is a category confusion, since a comparison that suggests where to look says nothing about what will be found there.

The characteristic failures are identifiable. A false analogy rests on similarities that are real and irrelevant to the conclusion, which is the commonest fault and is frequently persuasive because the similarities are genuine. Cherry-picked similarities present the parallels and omit the differences, which is a matter of presentation rather than of the underlying structure. Analogies that carry an emotional charge, particularly comparisons to historical atrocities, do their work through the charge rather than through the parallel and are correspondingly hard to assess calmly. Extending an analogy beyond the point it was introduced to make is a recurring error, since a comparison useful for one purpose may fail entirely for another. And an analogy offered as an explanation of how something works is doing a different job from one offered as evidence that something is true, and confusing the two produces arguments that cannot be evaluated.

Shared features in two cases support a probable conclusion about a further feature, never a guaranteed one. Relevance of the shared features matters far more than how many there are, and known differences weaken the argument in proportion to their relevance. Common law reasoning is this form conducted openly, with distinguishing a case meaning finding a relevant difference.