A methodology that relies on approximately true theories would be expected to reliably produce accurate results

The best-known argument of this sort was developed by Richard Boyd in the 1980s (see Boyd 1981, 1984, 1985). It starts by underlining the theory-dependency of scientific methodology, which comprises methods for designing experiments, for assessing data, for choosing between rival hypotheses, and so on. For instance, in considering possible confounding factors from which an experimental setup has to be shielded, scientists draw heavily on already accepted theories. The argument next calls attenti