4. Results
4.1. Descriptive statistics
Persons living in a building with a lung cancer patient were
much more likely to report in the main survey that lung cancer
was “definitely” more likely if they smoke (Fig. 1). Much of the
increase in the “definite” category comes from a reduction in the
“probable” category. The modal response for lung diseases was
“probable.” But again, the increase in “definites” among residents
of a building with a lung cancer patient comes from a reduction in
“probables.” In control group buildings, 67 percent of respondents
answered “probable” or “definite” (Table 2, Panel B). But 77 percent
of respondents in treatment group buildings did.
The modal responses for stroke are “a little more likely to
increase.” However, the shares of “probables” and “definites” are
higher for residents of buildings with a lung cancer patient. For
heart disease, responses are more evenly distributed among the
five response categories, implying that respondents thought the
link between smoking and onset of heart disease and of stroke is
weaker than between smoking and lung cancer and other respiratory
diseases. In fact, while lower than for lung cancer and other
respiratory diseases, which are so high that the correct response
is clearly “definite,” the objective probabilities of smokers dying
of stroke and heart disease are about double for heart disease and
nearly double for stroke relative to thatfor non-smokers (Doll et al.,
1994). Viewed from another perspective, a substantial proportion
of persons in a large sample of male smokers in our sample appear
to be unaware or at least do not display an awareness of the health
harms of smoking.