02.03.2013 Views

Thinking and Deciding

Thinking and Deciding

Thinking and Deciding

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

168 HYPOTHESIS TESTING<br />

given any other hypothesis (∼H). If the critical result is found, the probability of H<br />

increases considerably, because the diagnostic ratio is so high. 5<br />

For Platt, a good experiment is one that seeks a result with a probability of one,<br />

given one set of hypotheses, H1, <strong>and</strong> a probability of zero, given some other set H2;<br />

H2 is the same as ∼ H1 (everything not in H1). If the result is found, we know that<br />

the truth is in H1; if not, we know that the truth is in H2. Moreover, Platt specifies<br />

that the sets should be chosen so that p(H2) <strong>and</strong> p(H1) are about .5. By this choice<br />

of hypotheses to test, we can ask the fewest questions to determine exactly where the<br />

truth lies — that is, in which subhypothesis of H1 or H2.<br />

More generally, however, we should look for results in which the conditional<br />

probabilities for results, given the important hypotheses, differ greatly. Whether we<br />

seek results with probabilities of one or zero will depend to some extent on why we<br />

need to know <strong>and</strong> how sure we need to be. We do not always need absolute certainty<br />

in order to take some practical action or in order to accept some scientific theory as<br />

very likely true, for the purposes of planning our next experiment. A good heuristic<br />

to keep us on the trail of such results is this: “Be sure to consider alternative hypotheses.”<br />

If our planned experiment or observation cannot distinguish the hypothesis of<br />

interest from the alternatives, it is not a good experiment. If no experiment can do<br />

so, our hypothesis is untestable, <strong>and</strong> we ought to re-examine our goal in formulating<br />

it.<br />

Sometimes, results that are predicted by a new hypothesis, but improbable otherwise,<br />

are already known but not noticed before the new hypothesis is stated. For<br />

example, astronomers before Copernicus knew that retrograde motion of planets (opposite<br />

to their usual direction through the stars) occurred only when the planets were<br />

high in the sky at midnight (hence opposite the sun). This fact was implied by Copernicus’s<br />

theory. In Ptolemy’s theory, it was an unlikely coincidence. It was not taken<br />

as evidence against Ptolemy’s theory, though, until a better alternative was found<br />

(Lightman <strong>and</strong> Gingerich, 1991). According to probability theory, this is reasonable:<br />

The fact that evidence is improbable given a hypothesis does not weaken the<br />

hypothesis unless the evidence is more probable given some alternative.<br />

The psychology of hypothesis testing<br />

A traditional theory in psychology (Bruner, Goodnow, <strong>and</strong> Austin, 1956), now largely<br />

discredited, holds that knowing a concept amounts to knowing how to classify some<br />

instance as a member of a certain category. It was argued, moreover, that we classify<br />

instances on the basis of cues. For example, in card games, knowing the concept<br />

“spade” amounts to knowing how to classify cards as spades or nonspades. In this<br />

5 Popper would disagree with this account, because he regards probability as a poor criterion of the<br />

value of a theory. He argues that the “most probable” theory is always the least interesting, citing, as an<br />

example, a noncommittal statement such as this: “The light will be bent some amount, or possibly not at<br />

all.” He does not, however, consider a change in the probability of a hypothesis as a criterion of the value<br />

of a result, which is what is argued here.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!