Risky decision making across three arenas of choice: are younger and older adults differently susceptible to framing effects?
THE WAY IN WHICH CHOICE OPTIONS are phrased may have a profound impact on human decision making. In risky choice framing, describing options as losses (negative framing), as opposed to gains (positive framing), is frequently associated with a difference, or shift, in preference in favor of a risky (probabilistic) option over a certain (deterministic) option (for reviews, see Kuberger, 1998; Rothman & Salovey, 1997). The results of a study by Tversky and Kahneman (1981), pertaining to the Asian disease problem, are clearly illustrative of such a pattern: In that study, in which the participants were faced with a scenario involving the outbreak of a disease that threatened to kill 600 people, Tversky and Kahneman found that a majority of the participants (72%) preferred the certain option that “200 people will be saved” over the risky option that “there is 1/3 probability that 600 people will be saved and 2/3 probability that no people will be saved.” When the formally equivalent options were negatively framed, a majority (78%) instead preferred the risky option (“there is 1/3 probability that no people will die and 2/3 probability that 600 people will die”) over the certain option (“400 people will die”).
This framing effect, which appears to be at odds with the principle of descriptive invariance of expected utility theory (von Neumann & Morgenstern, 1944), has often been interpreted in light of prospect theory (Kahneman & Tversky, 1979). According to prospect theory, gains and losses are evaluated from a subjective reference point. People value a certain gain more than they do a probable gain with an equal or greater expected value; the opposite is true for losses. The function relating the subjective value and the corresponding losses is steeper than that for gains, such that the displeasure associated with a loss is greater than the pleasure associated with the same amount of gains. As a result, people tend to be risk averse when the options are positively framed, and risk seeking when the options are negatively framed.
There is a wealth of studies in which researchers have examined the effect of decision framing on the basis of the original Asian disease problem, similar life-death scenarios, and across other arenas of choice that involve threats to other types of goods (e.g., public property). The mirror-image reversal of choice preferences from risk averse under positive framing to risk seeking under negative framing (a bidirectional framing effect; Wang, 1996c) is robust in life-death scenarios, although on average, it is not as large as it is in the original study by Tversky and Kahneman (1981 ; Kuberger, 1998; Kuberger, Schulte-Mecklenbeck, & Perner, 1999). At the same time, the effect sizes are not homogenous, and in some situations, weak or nonexistent framing effects were observed (e.g., Fagley & Miller, 1987).
Task-related factors are important for understanding the boundary conditions and strength of framing effects (e.g., Kuberger, 1998). The results of a study by Wang (1996c; also see Wang, 1996a; Wang & Johnston, 1995) illustrated the influence of two such factors: (a) the type of goods at stake, and (b) the quantity of the goods at stake. In that study, choice preference patterns in a life-death problem and analogous scenarios that involved a threat to public property (museum paintings) or personal money were examined. In each of the scenarios, the prespecified quantity of the goods at stake (lives, paintings, dollars) varied from 6 through 60, 600, to 6,000. A bidirectional framing effect, which reflected a reversal from risk-averse to risk-seeking choices from the positive to the negative frame was observed for the life-death problem when the scenario involved 6,000 or 600 people, in line with the results by Tversky and Kahneman (1981). However, with a smaller contextual size (expected outcome), a decreasing percentage of participants favored the certain option regardless of framing. Hence, for scenarios that involved 60 or 6 persons, a majority of the participants favored the risky option even under positive framing. Therefore, the shift we observed toward more risk-seeking choices after negative framing occurred within the risk-seeking domain, and we labeled it a unidirectional (risk-seeking) framing effect. In the life-death scenario, the trend was for a decrease in risk-averse choices, with smaller values of expected outcome when public property or personal money was at stake. However, in this case, the observed shift toward more risk-seeking preference occurred within the risk-averse domain, which provided an example of another type of unidirectional (risk-averse) framing effect (cf. Fagley & Miller, 1997). When taken together, task-related factors–including the type of goods involved and the contextual size of the problem–seem to be important predictors of framing effects.
Another class of predictors of framing effects concerns individual difference factors (e.g., Fagley & Miller, 1990; Levin, Gaeth, Schreiber, & Lauriola, 2002; Tanner & Medin, 2004). For example, Stanovich and West (1998) examined participants’ responses to a variety of problems associated with reasoning biases, including the well-known conjunction fallacy (Linda problem) and the framing effect in the Asian disease problem. Their results suggested that the framing effect was driven by a minority of the participants, and that the prevalence of a choice shift from the positive to the negative framing conditions was related to overall cognitive ability (as indexed by SAT scores). In particular, the framing effect seemed to be largely attributable to participants who were cognitively less able.
Researchers have largely neglected studying the effect of development on framing effects, but in 11-year-olds, and not before that age, some researchers have observed a bidirectional framing effect similar to that of older adults (Reyna & Ellis, 1994). In the meta-analytic review by Kuberger (1998), there was no comparison of younger and older adults available for inclusion. This omission is noteworthy both from a theoretical viewpoint and if one considers the day-to-day relevance of decision making under risk for older adults. After midlife, the aging process seems to have a negative effect on a variety of intellectual abilities, including memory (Zacks, Hasher, & Li, 2000), reasoning (Schaie, 1996), and attention (McDowd & Shaw, 2000). If one considers the results of Stanovich and West (1998) and the generalized age-related deficits in cognitive functioning, including self-initiated elaborative processing (cf. Craik, 1986), then one might expect older adults to exhibit larger framing effects when compared with younger adults.
The only study we know of that addresses the issue of age-related differences in risky decision framing is a recent study by Mayhorn, Fisk, and Whittle (2002). Fifty-eight younger participants (M = 20.3 years) and 58 older participants (M = 70.3 years) responded to 16 problems that involved a choice between a certain option and a risky option. The scenarios, which were adopted from Tversky and Kahneman (1988), concerned health and money. With the exception of one monetary problem, in which the elderly favored the risk-averse option more frequently than did the young, no significant age-related differences in choice preferences were observed. On the basis of these results, Tversky and Kahneman concluded that adults of all ages are susceptible to framing effects. However, only six of the conditions for which results were reported allowed for a comparison of the positive and negative framing conditions. In other words, three scenarios allowed for the examination of framing effects. One of these was the original Asian disease problem. In light of the hypothesis that one expects older adults to show magnified framing effects, it is interesting to note that the results showed an exceptionally large framing effect (cf. Kuberger et al., 1999), for which the difference in choice preference between the positively and negatively framed problems exceeded 50%. However, this held true for both young and old participants, which rendered the observation difficult to interpret.
Given the relative lack of studies addressing issues concerning decision making in older adults in general (cf. Sanfey & Hastie, 2000), and the ambiguity of outcome of the only extant study addressing framing effects, our objective in the present study was to investigate further the issue of age-related differences in framing.
We wanted to investigate the issue of age-related differences in framing in the present study first, because there was a notable absence of studies that addressed decision making in older adults (cf. Sanfey & Hastie, 2000) and second, because of the ambiguity of the outcome of the only study (Mayhorn et al., 2002) that had addressed framing effects. In the present study, we included a version of the classic disease problem (Tversky & Kahneman, 1981) and two scenarios (modeled after the study by Wang, 1996c) that involved threats to public property (museum paintings) and personal money in our examination of the dynamics of framing effects and especially the extent to which they are age-dependent across different arenas of choice. On the basis of the study by Wang (1996c), we expected a unidirectional, risk-averse, framing effect rather than the standard bidirectional effects expected for the life-death scenario for the museum paintings scenario and the monetary scenario. (1)
The participants were 192 younger adults (M = 23.9 years, SD = 3.5) and 192 older adults (M = 69.1 years, SD = 7.4). They were recruited at public places (e.g., market places, the local bus station, and, in case of the younger groups, mainly the university campus) in the city of Umea, Sweden. Of the younger participants, 99 were men and 93 were women; of the older participants, 96 were men and 96 were women. The average number of years of formal education was 14.4 for the younger participants (SD = 2.2) and 9.9 for the older participants (SD = 3.9), a difference that was significant, t(382) = 13.8, p < .01. The participants were randomly assigned to the conditions: the positive or negative framing versions in each of the three scenarios. Accordingly, each experimental condition included 32 participants, with an approximately equal number of men and women in each condition.
There were three scenarios (see Appendix). The first was modeled closely on the original Asian disease problem (Tversky & Kahneman, 1981). The second and the third scenarios were modeled after the paintings problem and the personal money problem in Wang (1996c). In each of the scenarios, the options had the same probability structure, with 1/3 and 2/3 probability of the positive and negative outcomes, respectively. In the life-death and the paintings problems, 600 people and paintings were at stake. In the personal money problem, 60,000 Swedish crowns (about $8,000 at the time of the study) were at stake. We counterbalanced the presentation order of the options across the participants in each scenario and age group (as noted by Kuberger , this control was not implemented in most of the previous studies).
The percentage of the participants who preferred a certain and risky option for each scenario, age group, and framing condition, is presented in Table 1. Preliminary analyses revealed no effect of presentation order of the options, so the data were collapsed across this factor.
As can be seen, there was a clear difference in choice preference patterns between the positively and negatively framed versions for each scenario. This was consistent with more participants favoring the certain option after positive framing, which was in keeping with past research. In addition, the direction of the framing effect appeared to vary across problems. In particular, the scenario that involved threats to lives yielded a symmetrical difference in choice preferences depending on framing (i.e., risk aversion in the positive framing condition and risk seeking in the negative framing condition), whereas the paintings and money scenarios yielded an unclear preference of options in the negative framing condition. It is important to note that both the magnitude of Oframing effects and the pattern of variability in their direction seemed very similar, regardless of age group across the three scenarios. This matter is also clear if one considers the overall results (i.e., based on data collapsed across scenarios) shown in Figure 1. As one can see, the magnitude of the framing effect was virtually identical for the younger and the older group.
[FIGURE 1 OMITTED]
We performed binary logistic regression analyses, one for each scenario or problem, to evaluate statistically the relative contribution of framing condition and age on decision making. When the age groups differed on average with regard to educational attainment, we included the number of years of schooling in the analyses. The results of these analyses are summarized in Table 2. Our analyses showed reliable effects of framing condition (p .20, suggesting that power to detect a possible age difference is an unlikely factor underlying the failure to detect an age difference).
The objective of the present study was to examine the effects of gain-loss framing on decision making in situations involving threats to various types of goods that should be of relevance to most people (lives, paintings, and money), and, in particular, to examine whether younger and older adults differ in the extent to which they are susceptible to such framing effects.
We obtained several clear-cut findings. First, our predictions concerning variability of framing effects across the arenas of choice were confirmed. Specifically, a bidirectional framing effect, reflecting a majority preference in favor of the certain option after positive framing and a majority preference in favor of the risky option after negative framing, was observed for the life–death (disease) scenario, which is in line with the bulk of past research (Kuberger et al., 1999). Moreover, unidirectional framing effects in the risk-averse domain were observed for the scenarios involving decisions about public and personal property, also in keeping with past findings (Wang, 1996c).
These unidirectional framing effects occur when existing risk preference is weak following positive framing, and it is possible that the confirmation bias in form of a choice reversal (or difference), as in the classic Asian disease problem, is simply owing to the lower level of preference in favor of the certain option in the positive framing condition (fairly equal magnitudes of framing effects were observed across problems). Further research is probably needed to explain why people, perhaps counter-intuitively, are less cautious (more risk seeking) when the problem involves lives (that presumably have greater utility) than money or public property in these cases (cf. also Fagley & Miller, 1997). As already noted, the results by Wang (1996c) indicated that decreased quantity of goods at stake generally induces risk-seeking choice behavior. However, the aforementioned pattern held even when the problems, unlike the present money and lives problems, involved the same numerical values.
Second, and of main importance, the groups of younger and older adults were equally affected by framing of the options, something that held true regardless of goods at stake. Hence, the present results confirm the robustness of framing effects and extend the generality of the finding of age invariance in magnitude of framing by Mayhorn et al. (2002) by showing age-invariant differences (a) across scenarios involving three types of goods and (b) with the use of a between subjects design, rather than a within-subjects design. Together, the results of the present and those by Mayhorn et al. provide strong evidence that the well-replicated finding of negative age differences in basic cognitive functions do not correspond with increased response bias in the form of magnified framing effects. (2)
Turning to potential explanations of these results, the relation between general cognitive ability and framing effects might be less simple than is suggested by the aforementioned results by Stanovich and West (1998). In fact, the bulk of framing studies have observed substantial framing effects within participant groups that are likely to be above average in cognitive ability (e.g., university students). A more intriguing possibility, of course, is that age-related cognitive deficits can increase the susceptibility to framing effects, but that these effects are counteracted by some form of age-related compensatory mechanism, for example, in the form of added experience. Another possibility that is worth examining further is that age-specific choice preference patterns are observed in cases in which the problem descriptions involve age-specific social cues (cf. Wang, 1996b). It is worth noting that a study by Mandel (2001) challenged the basic phenomenon of framing as depicted by past and more recent studies, including the present one, by arguing that the problems used do not orthogonally manipulate descriptor frame (e.g., saved, die) and outcome frame (positive and negative expected outcome). Obviously, further research that addresses this critique and that departs from the methods suggested by Mandel is needed.
Given that framing effects in this and comparable studies are genuine, it would be interesting to determine the extent to which framing effects might be reduced or eliminated by instructions that guide cognitive processing at the time of making the decisions, and whether such guidance will have an equal effect in young and old individuals. The results on this issue pertaining to young adults only are somewhat mixed. Takemura (1992) observed the classic framing effect under standard conditions (i.e., without specific instructions of how to reason about the problem). However, when the instructions prompted the participants to engage in elaborative thinking, the framing effects vanished. In a related vein, Jou, Shanteau, and Jackson Harris (1996) observed no framing effects when the participants were required to provide a rationale of their choice. At the same time, recent results by LeBoeuf and Shafir (2003) failed to detect reduced framing effects in situations presumably associated with greater processing depth, which suggests that younger participants do not necessarily engage in elaborative processing even when cognitive support is provided.
Finally, future studies should examine the extent to which the framing effects as observed in the present types of problems that involve fictive situations generalize to real-life situations facing younger and older individuals and means to overcome these potential effects.
Choice Scenarios Used in the Present Study
Sweden is preparing for the outbreak of an unusual disease, which is expected to kill 600 people. The following alternative programs have been proposed to limit the spreading of the disease:
If program A is adopted, 200 people will be saved.
If program B is adopted, there is 1/3 probability that 600 people
will be saved, and 2/3 probability that no people will be saved.
If program A is adopted, 400 people will die.
If program B is adopted, there is 1/3 probability that no people
will die, and 2/3 probability that 600 people will die.
Museum Paintings Scenario
A large museum is ravaged by fire. 600 of the world’s most famous paintings run the risk of being destroyed. Two alternative programs have been proposed to rescue the paintings:
If program A is adopted, 200 paintings will be saved.
If program B is adopted, there is 1/3 probability that 600
paintings will be saved, and 2/3 probability that no paintings
will be saved.
If program A is adopted, 400 paintings will be destroyed.
If program B is adopted, there is 1/3 probability that none
of the paintings will be destroyed, and 2/3 probability that
600 paintings will be destroyed.
Personal Money Scenario
Suppose that you have invested in stock equivalent to the sum of 60,000 crowns in a company that just filed a claim [or bankruptcy. They offer two alternatives in order to save some of the invested money:
If program A is adopted, 20,000 crowns will be saved.
If program B is adopted, there is 1/3 probability that 60,000
crowns will be saved, and 2/3 probability that no money will
If program A is adopted, 40,000 crowns will be lost.
If program B is adopted, there is 1/3 probability that no money
will be lost, and 2/3 probability that 60,000 crowns will be saved.
TABLE 1. Preferred Option (Percentage Certain or Risky) Across
Scenarios (Lives, Painting, Money), Age Groups (Younger vs. Older),
and Framing Conditions (Positive vs. Negative)
Lives Paintings Money
Younger Older Younger Older Younger Older
Option P N P N P N P N P N P N
Certain 59 31 72 44 72 50 59 44 81 50 81 53
Risky 41 69 28 56 28 50 41 56 19 50 19 47
Note. P = positive framing; = negative framing.
TABLE 2. Summary of Logistic Regression Analyses of Choice Preferences
Scenario B SE Wald p Exp(B)
Framing condition 1.191 .381 9.778 <.01 .304
Education .008 .058 .017 ns 1.001
Age .009 .010 .825 ns 1.009
Framing condition -.791 .368 4.621 <.05 .453
Education .048 .071 .454 ns 1.049
Age -.003 .012 .074 ns .997
Framing condition .351 .412 1.778 <.01 .259
Education .111 .065 2.973 ns 1.118
Age .013 .010 1.536 ns 1.013
(1.) In fact, Wang (1996c) failed to detect a framing effect for the museum paintings problem using the same value of the expected outcome as that in the original Asian disease problem (2/3 x 600), used at present. However, given the trend of a negative linear relation between size of expected outcome and the percentage risk-averse choices discernible in the results by Wang, such an effect may be predicted.
(2.) Unlike the study by Mayhorn et al. (2002) in which the age-invariant framing effect was paired with substantial age-related deficits in perceptual speed (digit symbol) and reversed digit span, we did not collect data on cognitive measures. However, fine-grained normative data exist on various cognitive measures for the population in Umea (Nilsson et al., 1997). Within the unselect population the latter study demonstrated age differences exceeding 1 SD across the age range targeted at present, on various cognitive measures, including measures of episodic memory and WAIS-R Block Design. In the present study, the older adults’ educational levels are comparable with those expected from the study by Nilsson et al. This matter, paired with the fact that our younger participants were primarily drawn from a student population (expected to exhibit higher cognitive tests performance as compared with the unselect population) leaves little doubt that age-related deficits would have been observed if such measures had been collected.
Craik, F. I. M. (1986). A functional account of age differences in memory. In F. Klix & H. Hagendorf (Eds.), Human memory and cognitive abilities (pp. 409-422). Amsterdam: Elsevier.
Fagley, N. S., & Miller, E M. (1987). The effect of decision framing on choice of risky and certain options. Organizational Behavior and Human Decision Processes, 39, 264-277.
Fagley, N. S., & Miller, P. M. (1990). The effect of framing on choice: Interactions with risk-taking propensity, cognitive style, and sex. Personality and Social Psychology Bulletin, 16, 496-510.
Fagley, N. S., & Miller, P. M. (1997). Framing effects and arenas of choice: Your money or your life? Organizational Behavior and Human Decision Processes, 71, 355-373.
Jou, J., Shanteau, J., & Jackson Harris, R. (1996). An information processing view of framing effects: The role of causal schemas in decision making. Memory and Cognition, 24, 1-15.
Kahneman, D., & Tversky, A. (1979). Prospect theory. Econometrica, 47, 263-292.
Kuberger, A. (1998). The influence of framing on risky decisions: A meta-analysis. Organizational Behavior and Human Decision Processes, 75, 23-55.
Kuberger, A., Schulte-Mecklenbeck, M., & Perner, J. (1999). The effects of framing, reflection, probability, and payoff on risk preference in choice tasks. Organizational Behavior and Human Decision Processes, 78, 204-231.
LeBoeuf, R. L., & Shafir, E. (2003). Deep thoughts and shallow frames: On the susceptibility to framing effects. Journal of Behavioral Decision Making, 16, 77-92.
Levin, I. P., Gaeth, G. J., Schreiber, J., & Lauriola, M. (2002). A new look at framing effects: Distribution of effects sizes, individual differences, and independence of types of effects. Organizational Behavior and Human Decision Processes, 88, 411-429.
Mandel, D. R. (2001). Gain-loss framing and choice: Separating outcome formulations from descriptor formulations. Organizational Behavior and Human Decision Processes, 85, 56-76.
Mayhorn, C. B., Fisk, A. D., & Whittle, J. D. (2002). Decisions, decisions: Analysis of age, cohort, and time of testing on framing of risky decision options. Human Factors, 44, 515-521.
McDowd, J. M., & Shaw, J. S. (2000). Attention and aging: A functional perspective. In F. I. M. Craik & T. A. Salthouse (Eds.), Handbook of aging and cognition (2nd ed., pp. 221-292). Mahwah, NJ: Erlbaum.
Nilsson, L. G., Backman, E., Erngrund, K., Nyberg, L., Adolfsson, R., Bucht, et al. (1997). The Betula prospective cohort study: Memory, health, and aging. Aging, Neuropsychology, and Cognition, 4, 1-32.
Reyna, V. E, & Ellis, S. C. (1994). Fuzzy-trace theory and framing effects in children’s risky decision making. Psychological Science, 5, 275-279.
Rothman, A. J., & Salovey, P. (1997). Shaping perceptions to motivate healthy behavior: The role of message framing. Psychological Bulletin, 121, 3-19.
Sanfey, A. G., & Hastie, R. (2000). Judgement and decision making across the adult life span: A tutorial review of psychological research. In D. C. Park & N. Schwartz (Eds.), Cognitive aging: A primer. Philadelphia: Taylor & Francis.
Schaie, K. W. (1996). Intellectual development in adulthood: The Seattle Longitudinal Study. New York: Cambridge University Press.
Stanovich, K. E., & West, R. E (1998). Individual differences in framing and conjunction effects. Thinking and Reasoning, 4, 289-317.
Takemura, K. (1992). Influence of elaboration on the framing of decision. The Journal of Psychology, 128, 33-39.
Tanner, C., & Medin D. L. (2004). Protected values: No omission bias and no framing effects. Psvchonomic Bulletin & Review, 11, 185-191.
Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211, 453-458.
Tversky, A., & Kahneman, A. (1988). Rational Choice and the framing of decisions. In D. E. Bell, H. Raiffa, & A. Tversky (Eds.), Decision making: Descriptive, normative, and prescriptive intractins (pp. 167-192). Cambridge, England: Cambridge University Press.
von Neumann, J., & Morgenstern, O. (1944). Theory o[games and economic behavior. Princeton, N.J.: Princeton University Press.
Wang, X. T. (1996a). Domain-specific rationality in human choices: Violations of utility axioms and social contexts. Cognition, 60, 31-63.
Wang, X. T. (1996b). Evolutionary hypotheses of risk-sensitive choice: Age differences and perspective change. Ethology and Sociobiology, 17, 1-15.
Wang, X. T. (1996c). Framing effects: Dynamics and task domains. Organizational Behavior and Human Decision Processes, 68, 145-157.
Wang, X. T., & Johnston, V. (1995). Perceived social context and risk preference: A re examination of framing effects in a life-death decision problem. Journal of Behavioral Decision Making, 8, 279-293.
Zacks, R. T., Hasher, L., & Li, Z. H. (2000). Human memory. In F. I. M. Craik & T. A. Salthouse (Eds.), Handbook of aging and cognition (2nd ed., pp. 293-357). Mahwah, N J: Erlbaum.
Manuscript received March 22, 2004
Revision accepted August 31, 2004
Department of Psychology
Umea University, Sweden
Address correspondence to Michael Ronnlund, Department of Psychology, Umea University, S-901 87 Umea, Sweden; email@example.com (e-mail).
COPYRIGHT 2005 Heldref Publications
COPYRIGHT 2005 Gale Group