Human Information Processing - Human Factors and Ergonomics - page 91

In the best of worlds we would like to avoid trade-off decisions. If the sensitivity—
also called d′ (d prime)—can be increased, we will have solved the problems. Sensitivity
or d′ measures the distance between the mean values of the two distributions. Ideally, if
we could separate the distributions totally there would be no false alarms and no misses.
In some cases a medical doctor will know that the diagnosis is 100% accurate, and
therefore there will be no false alarms or misses. Sometimes, however, medical evidence
is very uncertain. Rhea et al. (1979) reported that radiologists missed about 20-40% of
tissue abnormalities, which would lead to a tumor diagnosis. Parasuraman (1985)
reported that radiologists did not change the data when screening patients as compared to
when investigating referrals. Since referral patients have already been screened, one
would think that the criterion β would be set lower resulting in fewer misses.
The sensitivity (d′) or the separation between the two curves is affected by operator
experience. An inexperienced radiologist will have a lower value for d′ than an
experienced radiologist, which means that the two curves will come closer together. An
experienced person has learned what to look for, and as a result d′ is greater, and the rates
of both misses and false alarms are lowered.
Sensitivity can also be used to evaluate equipment used to diagnose patients. Swets et
al. (1979) found that computerized tomography (CT) has greater sensitivity than a radio
nuclide scan apparatus.
5.13 VIGILANCE AND SUSTAINED ATTENTION
As mentioned above, many research studies have been performed with people looking at
displays or listening to rare signals. The job of military radar operator is
FIGURE 5.9