Saturday, July 12, 2008

The Craft of Research: Chapter 9

This chapter is about warrants, which show "why [your] data is not just reliable but relevant" (emph. orig.). Of course, your readers first have to accept your evidence as reliable (but that's why you use reputable sources!). In formal logic, a warrant is similar to both the major premise in a conditional syllogism (p --> q; p, therefore q) and a categorical syllogism (B = C; A = B; therefore A = C).The definition:
A warrant is a general principle that creates a logical bridge between
particular evidence and a particular claim.

Three criteria a warrant must satisfy:
  • Type of evidence in general;
  • Type of claim (again, in general) resultant from evidence; and
  • Stated/implied connection - cause/effect, one-portends-another, or many-times-allows-generalization.

Four undesirable qualities of a warrant that justify refusal:

  1. Falsehood. A too-general claim makes counterexamples easy. A too-narrow claim begs the question. Aim for the middle and don't use absolutes (always, none, etc.) if you can avoid it.
  2. Lack of clarity. Academic-field-specific warrants are usually unstated between researchers, but the outsider probably won't see the connection. State intermediate steps in reasoning.
  3. Inappropriateness. A literature-type warrant (e.g. "When the sound of one word occurs inside another, readers associate the meaning of the inside word with the outside word") would make less than no sense to, say, a historian (who, in the text, is studying the reasons Eisenhower - a man with a comforting voice and a good slogan - won).
  4. Inapplicability. The authors recommend decomposing the argument to test it, thus: (a) State the warrant as evidence-claim. (b) State argument as evidence-claim. (c) Do the major terms in each, match?

Always remember your audience and how they will view your paper. Conclusions that jump out at you will probably not jump out at them.

No comments: