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.

182 HYPOTHESIS TESTING<br />

Table 7.5: Test results <strong>and</strong> incidence of two diseases for 100 patients<br />

Test result Disease A Disease B<br />

Positive 64 8<br />

Negative 16 12<br />

useful test, with a much lower false-positive rate than .4, even if it also has a lower<br />

hit rate than .8. We will not think of looking for such a test, however, if we do not<br />

ask how likely we are to observe the same results if the alternative hypothesis is true.<br />

The same need to consider whether a result is consistent with alternative hypotheses<br />

— before looking for the result — occurs in science. As we can see from<br />

the present example, this heuristic can be justified by a normative theory based on<br />

the idea of expected utility, a theory we shall explore in Chapter 10.<br />

Conclusion<br />

Actively open-minded thinking can avoid one of the main biases in hypothesis testing,<br />

the failure to consider alternative hypotheses (possibilities). In addition, more<br />

careful search for goals can help us avoid confusing of the goal of seeking information<br />

that is useful in decision making with other goals, such as the goal of seeking<br />

information in general, or that of obtaining positive results. These applications of the<br />

prescriptive theory of actively open-minded thinking can bring us closer to the normative<br />

theory of hypothesis testing as specified by probability <strong>and</strong> expected-utility<br />

theory.<br />

There is more to science than hypothesis testing, however. In formulating hypotheses<br />

to test, the scientist must not only consider whether they are testable but<br />

also whether they are likely to be true. To meet the latter requirement, we must think<br />

well about what we already know (<strong>and</strong> the better we know it, the better we can think<br />

about it). We must make sure that our hypotheses are consistent with the evidence<br />

already available before we bother to collect new data.<br />

The rigorous st<strong>and</strong>ards of science are often put in opposition to the idea of “learning<br />

from experience.” In the next chapter, we ask just how good we are at learning<br />

from experience about the relation between one variable <strong>and</strong> another.

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

Saved successfully!

Ooh no, something went wrong!