The Logic of Scientific Discovery.

Using Epistemic Ratios to Evaluate Hypotheses: An Imprecision Penalty for Imprecise Hypotheses

“…Although sophisticated researchers and philosophers have repeatedly called for psychologists to proffer and test precise and risky hypotheses (Lakatos, 1970(Lakatos, , 1978Meehl, 1978Meehl, , 1990Meehl, , 1997Popper, 1959Popper, , 1962Popper, , 1972Popper, , 1983Roberts & Pashler, 2000;Trafimow, 2003), these calls mostly have gone unheeded. There are at least two reasons for this, which may be related.…”

“…An advantage of this proposal is that it respects Popper's (1959Popper's ( , 1972 corroboration and Platt's (1964) principle of strong inference. A researcher supports a theory by making risky predictions and disconfirming alternative possibilities.…”

“…To understand why the standard procedure has been said to be damaging to the field, I consider the argument that both philosophers (Lakatos, 1970(Lakatos, , 1978Popper, 1959Popper, , 1972Popper, , 1983) and psychologists (Meehl, 1978(Meehl, , 1990(Meehl, , 1997Roberts & Pashler, 2000; for a quantitative demonstration, see Trafimow, 2003) have made: Supporting risky predictions provides much more impressive corroboration (not proof) for a theory than does supporting less risky predictions. That is, researchers should prefer theories that successfully make predictions that are unlikely to be true, without the theory.…”

On Doing the “Impossible”: Inferring that a Putative Causal Relationship does not Exist

“…First, these conditions are not jointly sufficient for warranted causal inference, that is, we can satisfy all three conditions and still be mistaken in believing that A is a contributory cause of B. This feature of the criteria is in accord with the Popperian view of science, according to which all scientific knowledge is necessarily provisional, and hence, can be amended in the light of subsequent evidence [9,10]. Second, satisfaction of condition (3) is desirable but not essential for justifiable causal inference.…”

“…Popper's falsifiability criterion of demarcation between science and metaphysics is of great heuristic value in distinguishing propositions, conjectures and hypotheses that are scientifically testable from those that are not [9,101. It also has a salutary effect on the practice of scientific research by attuning researchers to the greater evidentiary value of falsifications than confirmations.…”