Framing the frame: How task goals determine the likelihood and
direction of framing effects
Todd McElroy1
Department of Psychology
Appalachian State University
John J. Seta
Department of Psychology
University of North Carolina at Greensboro
Judgment and Decision Making, vol. 2, no. 3, June 2007, pp. 251-256.
Abstract
We examined how the goal of a decision task influences the perceived
positive, negative valence of the alternatives and thereby the
likelihood and direction of framing effects. In Study 1 we manipulated
the goal to increase, decrease or maintain the commodity in question
and found that when the goal of the task was to increase the commodity,
a framing effect consistent with those typically observed in the
literature was found. When the goal was to decrease, a framing effect
opposite to the typical findings was observed whereas when the goal was
to maintain, no framing effect was found. When we examined the
decisions of the entire population, we did not observe a framing
effect. In Study 2, we provided participants with a similar decision
task except in this situation the goal was ambiguous, allowing us to
observe participants' self-imposed goals and how they
influenced choice preferences. The findings from Study 2 demonstrated
individual variability in imposed goal and provided a conceptual
replication of Study 1.
Keywords: goals, framing, Prospect Theory.
1 Introduction
According to economic accounts, such as the dominant discounted utility
model, the way in which a problem is stated should not influence
individuals' preferences. Rather, when deciding
between options, the individual should choose the option with the
greatest overall utility, regardless of the way in which the problem is
stated (or framed). These accounts (e.g., Edwards, 1954; Von Newmann
& Morgenstern, 1953) consider the value and probability of the
outcome, independent of the context - independent of the way the
decision is framed. Counter to economic accounts, prospect theory
(Kahneman &Tversky, 1979) predicts that the way a decision problem is
framed does influence individuals' preferences.
While this holds great importance for furthering the understanding of
rational choice, prospect theory has failed to consider person and
contextual factors in their editing or encoding rules (e.g., Lopes,
1983; McElroy & Seta, 2003; Rettinger & Hastie, 2003; Reyna &
Braired, 1991; Schneider, 1992). For example, according to Rettinger
and Hastie the strategies that guide decisions are an interactive
product of person and contextual variables. From this view,
information contained in the decision problem is encoded and
represented as a mental model. The content and mental representation
that it generates, in turn, determines the decoding rules that lead to
a decision. Different decoding processes are ordered along a continuum
from most deliberate/analytic to most automatic/intuitive. The
encoding rules described by prospect theory lie in the middle of the
continuum, intuitive but involving some analytical processing, as
framing problems typically involve numbers. Similar research has shown
that framing effects, like those predicted by prospect theory, are more
likely for individuals who are induced or predisposed to process
holistically, using contextual referencing (e.g., McElroy & Seta,
2003; 2004) and when more "gist
like" memory retrieval is utilized (e.g., Reyna &
Brainerd, 1991; Reyna, Lloyd & Brainerd, 2003). This work highlights
the importance of considering both person and contextual variables in
understanding how individuals encode (edit) information in a
decision-problem.
Goals. The consideration of person factors, such as personal
goals is important for understanding the decision-making process
(e.g., Bargh, Gollwitzer, Lee-Chai, Barndollar, & Trötschel, 2001;
Krantz & Kunreuther, 2007; Stapel & Koomen, 2006). It may not only
provide insight into the encoding process and the likelihood of
framing effects, but it also may provide information about the
direction of framing effects. Two studies were designed to test this
possibility. In the typical framing problem used in the literature it
is implicitly assumed that decision-makers view an increase in the
outcome or commodity as desirable and positive. For example, in
risky-choice studies utilizing the classic Asian disease approach,
decision-makers typically make a choice about a situation where
increasing the commodity (e.g., human lives) is the goal. So whether
the problem is framed positively as gains or negatively as losses, the
goal of increasing lives remains constant and desirable to
decision-makers. Studies, such as the Asian disease problem that have
the inherent goal of increasing the commodity (lives), generally
demonstrate findings consistent with prospect theory predictions;
risk-aversion when the problem is framed as a gain and risk-seeking
when it is framed as a loss.
All tasks, however, are not oriented in the direction of increasing the
commodity in question; with some tasks, the goal is to decrease it.
One example is when individuals are overweight and seek to lose
undesirable body fat. In this situation, because a gain in body weight
is inconsistent with a decision-maker's goal, each
gained unit of body weight is undesirable; conversely, because a loss
is consistent with the decision-maker's goal, each lost
unit is desirable.
In a situation such as this, prospect theory (e.g., Kahneman &
Tversky, 1979) would not predict individuals to be risk-averse when the
decision problem is framed as a gain and risk-seeking when it is framed
as a loss. Rather, because decreasing the commodity is desirable, a
preference reversal would be expected; individuals should be
risk-seeking (not risk-averse) when the problem is framed in terms of
gains and risk-averse (not risk-seeking) when framed in terms of
losses. One reason why preference reversals are rarely seen in the
literature (see Levin & Chapman, 1990 for an
exception) may be because the vast majority of framing studies have
used problems that clearly involve the goal of increasing the supply of
the commodity. Nevertheless, there are many decisions in life where
individuals have the inherent goal of decreasing a
commodity's supply.
Individuals, however, not only have goals of increasing or decreasing a
commodity but at times the goal also may be to maintain the current
status of a commodity. In this situation, either a gain or a loss in
the commodity is contrary to the individual's goal.
Thus, either type of change is undesirable, leading decision-makers to
make equivalent responses when the problem is framed as a gain or as a
loss
It may not always be the case, however, that the goal of the task is
clear. For example, McCaffery and Baron (2004) examined how attribute
framing influenced opinions about taxation. They found that
contextual cues, such as the attribute frame, evoke internalized
principles that are used for problem analysis. Further, these
different principles could determine the goal, which has substantial
influence on decision processing, even leading to directional shifts
in preference. Although this research did not focus on risky choice,
it nonetheless demonstrated that people can impose different goals
onto a decision which will then affect their processing of the task.
Goal ambiguity. Framing tasks, such as the Asian disease
problem that involve the loss of human lives typically generate uniform
goals. Because of intergroup pressures, most if not all Americans
desire to increase and not decrease the life of another American (i.e.,
ingroup member). Consequently, decision-makers tend to be risk-averse
when the problem is framed as a gain and risk-seeking when it is framed
as a loss. It is not the case, however, that all framing tasks produce
uniform goals. Just as there is variability in the frame that
individuals can impose on ambiguous situations (e.g., Elliot &
Archibald, 1989; McElroy, Seta & Warring, 2007; Wang, 2004) the
question of whether to increase, decrease or maintain the commodity in
question also may be ambiguous and thus open to interpretation. In
this case, there might be considerable variation among decision-makers
in the goals that they impose; some might impose a goal to increase the
commodity (an incremental goal), others to decrease the commodity (a
decremental goal), and still others to maintain it (a maintenance
goal).
Further, if an approximately equal number of individuals impose each of
the three goals then it will appear as though the decision frame is
having little or no effect when we consider the decisions of an entire
population of decision-makers; the choices of individuals who impose a
maintenance goal will not be affected by the frame whereas the choices
of individuals who impose an incremental goal will be counterbalanced
by those imposing a decremental goal. The gain condition for
individuals imposing an incremental goal will be relatively risk-averse
whereas those imposing a decremental goal will be relatively
risk-seeking. Conversely, in the losses condition, individuals
imposing an incremental goal will be relatively risk-seeking whereas
those imposing a decremental goal will be relatively risk-averse.
Thus, it may be the case that although the frame is having a
significant influence on the choices of each individual, the framing
effect for the entire population of decision-makers is masked by
individual differences in goal imposition. And failures to demonstrate
framing effects may in fact be failures to consider the goals of the
decision-makers. When person factors, such as
decision-makers' goals, are not taken into
consideration, it may appear as though the decision-frame is having
very little or no effect on individual choice preference, when in fact
it is having considerable influence.
Overview of studies. Our experiments were designed to
determine whether decision-makers' goals interacted
with the way in which the problem was framed. In the first experiment
we made the goal of the risky-choice decision problem explicit by
informing our participants that the goal was to either increase,
decrease or maintain the commodity in question. We capitalized on an
everyday observation in which some individuals are underweight and
their goal is to gain weight, some are overweight and their goal is to
lose weight, and some are "just
right" and their goal is to maintain weight. Thus,
we were able to use the same commodity (weight) but shift
decision-makers' goals.
In Experiment 2 we did not manipulate the direction of the goal that
participants imposed onto the task; rather, we allowed them to
self-impose a goal. To accomplish this, we used the same commodity as
in Experiment 1 but did not make explicit the goal of the decision
problem; rather, we purposefully made the goal of the decision-maker
ambiguous so that individuals would impose their own idiosyncratic
goals on the decision problem; some individuals imposing an
incremental goal, others a decremental goal and still others a
maintenance goal. Framing effects should not be observed for those
imposing a maintenance goal but should be observed for individuals who
impose either an incremental or decremental goal. However, because we
expected the pattern of these framing effects to be in opposite
directions, if an approximately equal number of participants chose each
goal, then we should not find framing effects (or find especially weak
ones) when we examine the decisions of our entire population of
participants.
2 Experiment 1
In this study we made explicit the goal of increasing, decreasing or
maintaining weight. We expected a typical risky-choice framing effect
when the goal was to gain weight; a reversal of the typical effect when
the goal was to lose weight and no framing effect when the goal was to
maintain weight.
2.1 Method
2.1.1 Participants and design
Participants were 150 Appalachian State University undergraduate
students who received class credit for their participation. The design
of our study was a 3 Task Goal (increase, decrease, maintain) X 2 Frame
(gain, loss) between factors design.
Materials and Procedure. After consenting to take part in the
study, participants were presented with our vignette. We created a
decision scenario involving weight control where all three goals as
well as the frame were reasonable. Participants were provided with a
situation involving an athlete who had the goal of weight control.
Each of the weight-goal conditions are presented in italicizes.
Imagine that you are an athlete with the goal of (decreasing,
increasing, maintaining) your weight as much as possible. Because of
your sport, at this juncture in the season, (the lower your
weight the better you can perform, the higher your weight the better
you can perform, your current weight is where you can perform best).
You have to begin a specialized training program and you must choose
between the following two programs. Assume that the following
alternatives represent the exact estimates for each training program.
Participants were then presented with the following alternatives framed
as either gains or losses:
If program A is adopted, 20 pounds will be gained.
If Program B is adopted, there is a one-third probability that 60 pounds
will be gained and a two-thirds probability that no pounds will be
gained.
Or:
If program A is adopted, 40 pounds will be lost.
If Program B is adopted, there is a one-third probability that no pounds
will be lost and a two-thirds probability that 60 pounds will be lost.
Afterward, all participants were asked to rate their opinion of the two
options on a 7-point scale ranging from 1 (Definitely would recommend
A) to 7 (Definitely would recommend B).
Table 1: Average choice preference as a function of problem goal and
positive/negative frame.
| Positive frame
| Negative frame |
Problem goal: | N | Mean | N | Mean |
(lr)2-3(lr)4-5
Increase | 25 | 2.6 | 25 | 3.9 |
Decrease | 25 | 3.9 | 25 | 2.4 |
Maintain | 25 | 3.9 | 25 | 3.6 |
|
3 Results
To determine whether goals influenced participants choice preferences
for the different frames, we performed an analysis of variance on the
data; the goal (incremental, decremental, maintain) and decision frame
(gain, loss) acted as our independent variables and preferences as our
dependent variables. As expected, this analysis did not reveal a
decision frame main effect, F (1, 144) = .3,
p.5. It did, however, reveal our predicted
decision frame X goal interaction, F (2, 144) = 7.5, p < .01.
To explore the interaction we performed
contrasts for gain/loss framing within each of our three goal
conditions (See Table 1).
In the increasing-goal condition we found a significant main effect for
problem framing F (1, 48) = 6.6, p < .01. As may be
seen in Table 1, this effect is consistent with typical findings in
risky-choice framing tasks with participants demonstrating a relatively
stronger risk-averse tendency in the gains condition than in the losses
condition. In the decreasing-goal condition we also found a
significant framing effect F (1, 48) = 8.33, p < .01.
However, and consistent with our predictions, the typical framing
effect was reversed; participants demonstrated a relatively stronger
risk-seeking tendency in the gains condition than in the losses
condition. Finally, in the maintaining-goal condition, we found no
effect for the frame F (1, 48) = .36, p > .5.
This finding fits with our proposition that when the goal is to
maintain the current status, both increases (gains) and decreases
(losses) are perceived as a loss. In fact, the preferences of
participants in this condition did not differ from those in the
increase-loss or decrease-gain conditions F's < 1.
The results of Study 1 demonstrate that the goal of the decision maker
has profound effects on how individuals respond to the framing of
alternatives. These findings further extend our knowledge of framing
effects, providing a fuller understanding of how goals influence the
likelihood and direction of framing effects.
4 Experiment 2
The purpose of Experiment 2 was twofold. First, we sought to determine
whether participants would impose different goals on an ambiguous
decision problem. Specifically, would there be individual differences
in the goal (increase, decrease, maintain) that participants set for
the task? Second, did participants' "imposed goal" influence their
decision in the same way as it did in Experiment 1? Although we
expected a framing effect for individuals who imposed an incremental or
decremental goal, we expected these effects to be in opposite
directions. Further, we did not expect to observe framing effects for
individuals who imposed a maintenance goal. Thus, if roughly
equal numbers of individuals imposed each goal we should either not find
a framing effect or find a weak one for our entire population of
participants.
We provided participants with a weight control situation similar to
Study 1. Different from Study 1 however, we did not include an
explicit "weight-control" goal for
the hypothetical decision task. Rather, we constructed the task so
that the goal of the actor was purposefully ambiguous; allowing
participants to impose their own weight-control goal for the task.
After assessing the goal that individuals imposed, we next observed how
the self-imposed goal influenced the framing effect by measuring
participants' risky-choice preferences.
4.1 Method
4.1.1 Participants and design
Two hundred twenty-eight2 undergraduates participated in this study. The design of
our study included the between factors of
participants' self-imposed goal for the decision task
(increase, decrease, maintain) and the problem frame (gain, loss).
4.1.2 Materials and Procedure
Participants were run in groups of approximately 10 individuals. After
providing informed consent, they were provided with a weight management
task similar to Study 1 except the task did not contain a defined
goal.3 The
situation read as follows:
Imagine that you are an athlete and you have to begin a
specialized weight training program and you must choose between the
following two programs. Assume that the following alternatives
represent the exact estimates for each training program.
Directly afterward, participants were asked to indicate what they
believed the goal of the athlete in the task was (increase, decrease or
maintain weight). After determining the goal that they had imposed
onto the task, we then provided participants with the risk-seeking and
risk-averse alternatives framed either positively or negatively (the
same as in Study 1). Finally, participants were asked to rate their
preference for the two alternatives on a 7-point scale ranging from 1,
definitely would recommend A to 7, definitely would recommend B.
4.2 Results and discussion
As expected, participants imposed different goals for the athlete in
the decision problem. To determine whether the imposed goal influenced
participants choice preference for the different decision frames, we
performed an analysis of variance on the data with imposed goal
(incremental, decremental, maintenance) and decision frame (gain, loss)
acting as our independent variables and risky choice preferences as our
dependent variable. The analysis revealed a main effect for imposed
goal F (2, 220) = 3.71, p < .03, as well as the
expected overall interaction between frame and imposed goal F (2, 220) = 12.09, p < .0001. As may be seen in Table 2,
contrasts revealed that when participants imposed an incremental goal,
they demonstrated framing effects consistent with those typically found
in risky-choice type framing tasks F (1, 70) = 8.3,
p < .005. However, when participants imposed a
decremental goal, the results revealed framing effects that were
opposite to those of participants who imposed an incremental goal and
opposite to those typically found in the literature F (1, 85) = 17.1,
p < .00006. Finally, when participants imposed a
maintenance goal, a framing effect was not obtained F (1, 65) = .84,
p > .35.
Table 2: Average choice preference as a function of self-imposed goal and
positive/negative frame.
| Positive frame
| Negative frame |
Problem goal: | N | Mean | N | Mean |
(lr)2-3(lr)4-5
Increase | 47 | 2.8 | 25 | 3.9 |
Decrease | 29 | 4.4 | 58 | 2.9 |
Maintain | 37 | 4.2 | 30 | 3.9 |
|
An additional point of interest for us was to examine whether a framing
effect would be found across the imposed goal conditions. Because
there was an approximately equal division of the goals that
participants imposed: increase (72), decrease (87) and maintain (67),
we did not observe a decision frame main effect, F (2, 220) = 1.19,
p > .27. The direction of the framing effect
obtained by those imposing an incremental goal was counterbalanced by
the reverse direction of the framing effect obtained by those imposing
a decremental goal. The results of this study are conceptually
consistent with those obtained in Experiment 1 and demonstrate how
goals can influence the likelihood and direction of framing effects.
5 General discussion
In typical risky-choice decision tasks, the goal is to increase the
commodity at stake (e.g., lives, grades, health). Our findings reveal
that, when the goal is different, so too is the perception of gains
and losses. In short, the goal of the decision task can determine
whether a relative gain or loss is perceived as a psychological gain
or loss. We propose that when the task goal is to increase the
commodity at stake, gains are consistent with decision-makers goals
and desirable whereas losses are inconsistent and undesirable. The
opposite is true when the task goal is to decrease the commodity;
gains are inconsistent and undesirable whereas losses are consistent
and desirable. When the goal is to maintain the current status of a
commodity, either a gain or a loss is counter to the individual's goal
and this makes either type of change undesirable and leads to
equivalent responses when the problem is framed as a gain or as a
loss.
Several reasons have been offered for when and why framing effects are
not always obtained. For example, research has shown that processing
style (e.g., Igou & Bless, 2007; McElroy & Seta, 2003; Reyna &
Brainerd, 1991) elaboration (e.g., Sieck & Yates, 1997; Simon, Fagley
& Halleran 2004) and numeric predisposition (e.g., Peters, Vastfjall,
Slovic, Mertz, Mazzocco, & Dickert 2006) can all influence the
strength of framing. The results of the current study provide an
additional reason. When the goal of the decision-maker is not to
increase or decrease supply of the outcome in question, then framing
effects would not be expected from prospect theory. In this situation,
either a gain or a loss in the supply of the outcome would be
undesirable from the perspective of the decision-maker. For example, a
decision-maker who has a personal goal of maintaining the supply of an
outcome such as weight, may project this goal onto a decision problem
involving a gain or loss in weight. If so, from the decision-makers
own perspective, a gain or loss would be equivalently undesirable. In
this case, both would be perceived as losses and framing effects would
not be expected. Situations like this highlight the importance of
decision-makers' goals in determining the valence of an
outcome and thereby the likelihood and direction of framing effects.
We believe that in most situations where risk is involved individuals
are considering taking a chance because they desire to increase some
commodity. Although this goal may be common, it is not inherent in all
situations. A variety of circumstances exist where decreasing or
maintaining a commodity is desirable. Thus, it is important to
consider decision-makers' goals in predicting the
likelihood and direction of framing.
References
Bargh, J. A., Gollwitzer, P. M., Lee-Chai, A., Barndollar, K., &
Trötschel, R. (2001). The
automated will: Nonconscious activation and pursuit of behavioral goals.
Journal of Personality and Social Psychology,
81, 1014-1027.
Elliott, C. S., & Archibald. (1989). Subjective framing and attitudes
towards risk. Journal of Economic Psychology, 10, 321-328.
Igou, E. R., & Bless, H. (2007). On undesirable consequences of
thinking: framing effects as a function of substantive processing.
Journal of Behavioral Decision Making, 20, 125-142.
Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of
decision under risk. Econometrica, 47, 263-291.
Krantz, D. H. & Kunreuther H. C. (2007). Goals and plans in
decision making. Judgment and Decision Making, 2, 137-168.
Levin, I. P., & Chapman, D. P. (1990). Risk taking, frame of reference,
and characterization of victim groups in AIDS treatment decisions.
Journal of Experimental Social Psychology, 26, 421-434.
Lopes, L. L. (1983). Some thoughts on the psychological concept of
risk. Journal of Experimental Psychology: Human Perception
and Performance, 9, 137-144.
McCaffery, E. J., & Baron, J. (2004). Framing and taxation: Evaluation
of tax policies involving household composition. Journal of Economic
Psychology, 25, 679-705.
McElroy, T., & Seta, J. J. (2003). Framing effects: An
analytic-holistic perspective. Journal of Experimental Social
Psychology, 39, 610-617.
McElroy, T., & Seta, J. J. (2004). On the other hand am I rational?
Hemispheric activation and the framing effect. Brain and
Cognition, 55, 572-580.
McElroy, T., Seta, J., & Warring, D. (2007). Reflections of the Self:
How self-esteem determines decision framing and increases risk
taking. Journal of Behavioral Decision Making, 20, 223-240.
Peters, E., Vastfjall, D., Slovic, P., Mertz, C. K., Mazzocco, K., &
Dickert, S. (2006). Numeracy and decision making.
Psychological Science, 17, 407-413.
Rettinger, D. A. & Hastie, R. (2003). Comprehension and decision
making. In S. L. Schneider & J. Shanteau (Eds.). Emerging
perspectives on judgment and decision research (pp. 165-200). New York,
NY: Cambridge University Press.
Reyna, V. F., & Brainerd, C. J. (1991). Fuzzy-trace theory and framing
effects in choice: Gist extraction, truncation, and conversion.
Journal of Behavioral Decision Making, 4, 249-262.
Reyna, V. F., Lloyd, F. J., & Brainerd, C. J. (2003). Memory,
development, and rationality: An integrative theory of judgment and
decision making. In S. L. Schneider & J. Shanteau (Eds.),
Emerging perspectives on judgment and decision research (pp.
201-245). New York, NY: Cambridge University Press.
Schneider, S. L. (1992). Framing and conflict: aspiration level
contingency, the status quo, and current theories of risky choice.
Journal of Experimental Psychology: Learning, Memory, and
Cognition, 18, 1040-1057.
Sieck, W., & Yates, J. F. (1997). Exposition effects on decision
making: choice and confidence in choice. Organizational
Behavior and Human Decision Processes, 70, 207-219.
Simon, A. F., Fagley, N. S. & Halleran, J. G. (2004). Decision framing:
Moderating effects of individual differences and cognitive processing.
Journal of Behavioral Decision Making, 17, 77-93.
Stapel, D. A. & Koomen, W. (2006). The flexible unconscious:
Investigating the judgmental impact of varieties of unaware
perception. Journal of Experimental Social Psychology,
42, 112-119.
Wang, X. T. (2004). Self-framing of risky choice. Journal of
Behavioral Decision Making, 17, 1-16.
Footnotes:
1Direct correspondence to Todd McElroy,
Department of Psychology, College of Arts and Sciences, P.O. Box 32109,
265 College St., Boone NC. 28608; phone: (828) 262-2720; email:
mcelroygt@appstate.edu. Special thanks to Cathy Seta for her
insightful comments.
2Two participants were not included
in our analysis because they failed to indicate a goal for the decision
task.
3 It is reasonable for individuals to impose different
goals onto this task. For example, many athletes such as football
players need to either gain w
eight or loose weight for their optimal
performance. Further, if they are already at the desirable weight,
maintenance of weight is crucial for best performance.
File translated from
TEX
by
TTH,
version 3.74.
On 25 Aug 2007, 12:35.