Character Traits as Reference Classes

In United States v. Williams the Third Circuit Court of Appeals spoke of judgments about the probative value of character as "sweeping and non-individualized judgment[s]." In Commonwealth v. Adjutant the Supreme Judicial Court of Massachusetts saw a big difference between evidence such as the reputation of a person for violence and evidence that a person had attacked other people with boiling water, a butcher knife, and a crowbar.

I have some questions about "reputedly violent people" and "people who attack other people with crowbars etc."

Although reference classes are thought to be an ingredient or basis for statistical inference, can we nevertheless legitimately refer to "people who are reputedly violent" and "people who attack other people with crowbars" as "reference classes"?

What if you and I have never actually observed the behavior of a person who is reputedly violent or what if you and I have never observed the other behavior of a person who has attacked another person with a crowbar? Perhaps these are "reference classes" in the sense that they are culturally-inherited generalizations based on societal experience? Or are they instead or just as much "reference classes," or "generalizations," that people such as you and I extract from the other beliefs that you and I have about human behavior, human motivations, human derangement, etc.? If we extract such reference classes from our other beliefs, are we learning from or making use of experience? What if the beliefs from which we extract reference classes rest in part on experience?

What is the relationship between the putative reference classes (i) "people who are reputedly violent" and (ii) "people who attack another person with a crowbar"?

A. Is #(ii) a subset of #(i)?

B. Does #(ii) displace #(i)?

C. Even though #(ii) seems more specific, or more narrow, than #(i), does #(i) carry some information that #(ii) does not harbor?

Of course, there is the preliminary question whether #(i) harbors any information whatever about members of that reference class. I answer peremptorily: "yes." But by what epistemic right do I do so?

Questions ## A-C above were prompted by my recent reading of some material by James Franklin on reference classes. He thinks it's time to think further about reference classes. I agree.

the dynamic evidence page

coming soon: the law of evidence on Spindle Law