| |
“Framing”
refers to several related phenomena (
Levin, Schneider and Gaeth, 1998). For the purposes of
this chapter, I will simply define it as the influence of the
format or wording of a question on its answer. The classic framing
study by McNeil,
Pauker, Sox, and Tversky (1982) provides a good example. Physicians,
students, and patients with chronic diseases other than lung cancer
were asked to consider two possible treatments for lung cancer:
surgery or radiation. Approximately half of the participants
examined the survival probabilities for the two types of treatment.
The other half of the participants examined mortality data. Of
course, survival data is the converse of mortality data, so the
information provided to each group was essentially identical.
Nevertheless the persons who received survival data chose radiation
therapy less frequently than did subjects who received mortality
data. Because the information shown to each group should have
had no influence on which type of therapy is favored, this result
calls into question exactly what status the patient’s expressed
preference should have. Again, the assumption has generally been
that patients should be given the opportunity to express what
they feel is in their own best interest. If it is known that
seemingly trivial factors can influence what a patient thinks
is in his or her own best interest, then how can a physician be
sure that such interests are truly being heeded?
Some
recent research has suggested that in important medical decisions
in which there is high patient involvement, the framing of a question
may have less influence than in the McNeil
et al., (1982) study, which used persons who did not
have the disease described in the questionnaire. For example,
Donovan
and Jalleh (2000) found that when confronted with an infant
immunization scenario, mothers of infants, pregnant women, and
women who intended to get pregnant in the next 12 months were
not influenced by a framing manipulation. However the framing
of the scenario did influence women who were not in any of these
three categories. Similarly, Christensen
et al., (1995) found that framing had only limited
impact in a large number of medical scenarios. (See also Llewellyn-Thomas,
McGreal, and Thiel, 1995.)
The
frames used by McNeil
et al., (1982), Donovan
and Jalleh (2000), and Christensen
et al., (1995) were all gain/loss manipulations. In
general, participants are more risk-averse when responding to
scenarios framed in terms of gains or survival; participants are
more risk-seeking when responding to scenarios framed in terms
of loss or mortality. To understand this asymmetry it is necessary
to present some key aspects of prospect
theory (Kahneman
and Tversky, 1979).
|