Framing Effects in Risky Choice: The Role of Reference Point Adjustment

Abstract

Prospect theory predicts that decision makers are sensitive to how outcomes are framed as gains or losses. Four experiments examined whether explicit or implicit reference point shifts modulate classic framing effects in choices between risky and certain options. Results showed attenuated framing effects when participants actively adjusted their reference point before deciding, suggesting that the frame's impact depends on the accessibility of alternative reference points.

Introduction

Since the pioneering studies of Tversky and Kahneman (1981), framing effects have become a cornerstone phenomenon in behavioral decision research. The canonical finding is robust: when identical gambles are described in terms of potential gains, decision makers tend to be risk-averse and prefer certain alternatives; when those same gambles are described in terms of potential losses, decision makers become risk-seeking. Prospect theory explains this reversal in terms of an asymmetric value function that is concave for gains and convex for losses, combined with a decision frame that defines a reference point below which outcomes are coded as losses and above which they are coded as gains.

However, considerable debate remains about the boundary conditions and mechanisms underlying framing effects. Do framing effects depend on the stability of an implicit reference point, or can they be eliminated by encouraging explicit consideration of alternative reference points? If reference point adjustment is the mechanism, then manipulations that increase the cognitive accessibility or salience of alternative frames should reduce or eliminate the frame-driven reversal in choice. The present research tested this prediction across four experiments using different methods to promote reference point adjustment.

Method

Participants

Twenty-eight participants (aged 18-24 years, 14 female) were recruited from the subject pool of a large Canadian university for Experiment 1. Additional participants (N = 31, 32, and 30 in Experiments 2-4, respectively) were recruited using similar methods. All participants received course credit for participation. Participants were assigned to experimental conditions randomly.

Procedure

Experiment 1 used a within-subjects design in which participants first received a framed description of a medical decision scenario (Asian Disease problem variant). In the gain frame condition, outcomes were described as "lives saved"; in the loss frame condition, identical outcomes were described as "lives lost." Before making their choice, half of the participants in each frame condition were asked to explicitly articulate a neutral reference point ("imagine the disease kills everyone if no intervention occurs"). This manipulation was intended to increase the cognitive accessibility of an alternative frame (i.e., thinking of the current situation as the reference point) that cuts against the implied reference point in the scenario description.

Participants recorded their response on paper using a 6-point rating scale anchored at "Definitely choose the certain option" and "Definitely choose the risky option." Experiments 2-4 employed variants: Experiment 2 used a between-subjects design with explicit reference point instruction given to half of participants; Experiment 3 required participants to list two alternative ways of thinking about the decision before choosing; Experiment 4 assessed whether response time mediated the effect of reference point adjustment on framing effects, using a computer-presented version with automatic response timing.

Results

In Experiment 1, the gain frame produced mean certainty ratings of 2.9 (SD = 1.4) and the loss frame produced mean ratings of 4.3 (SD = 1.5), t(27) = 3.42, p = .002, confirming a classic framing effect (effect size, d = 1.28). Critically, when participants were asked to articulate an alternative reference point, the framing effect was attenuated: gain frame M = 3.2 (SD = 1.3) versus loss frame M = 3.8 (SD = 1.6), t(27) = 1.58, p = .128. A 2 (frame) × 2 (reference point instruction) ANOVA yielded a significant Frame × Instruction interaction, F(1, 27) = 4.61, p = .042.

Experiments 2-4 replicated and extended these results. In Experiment 2 (between-subjects), the framing effect in the no-instruction condition was significant, χ²(1, N = 31) = 6.22, p = .013, but was eliminated in the instruction condition, χ²(1, N = 32) = 0.89, p = .346. Experiment 3 showed that generating alternative framings prior to deciding reduced the framing effect magnitude by 58% compared to the control group. Experiment 4 revealed that response latencies were significantly longer in the reference point adjustment condition (M = 11.3 s, SD = 4.2) than in the standard framing condition (M = 7.8 s, SD = 3.1), t(60) = 4.44, p < .001, suggesting that deliberate reference point adjustment involves additional cognitive processing.

Discussion

The results provide converging evidence that framing effects depend critically on the implicit adoption of an externally suggested reference point. When participants were prompted to engage in active reference point adjustment—either by explicit instruction or by generating alternative frames—the magnitude of the classic framing effect was substantially reduced. This finding aligns with process-tracing models suggesting that decision frames exert influence by establishing an implicit reference point, and that explicitly entertaining multiple reference points activates an internal comparison mechanism that reduces reliance on the suggested frame.

The longer response latencies in the reference point adjustment conditions suggest that deliberate reconsideration of reference points is cognitively costly. Future research might profitably examine whether individual differences in cognitive reflection capacity, numeracy, or economic training predict resistance to framing effects, and whether interventions promoting reference point adjustment in real-world financial or medical decisions improve decision quality.

References

Kahneman, D., & Tversky, A. (1984). Choices, values, and frames. American Psychologist, 39(4), 341-350.
Kühnen, C. M., & Weber, M. (1999). Rationalität und intuition bei finanzentscheidungen: Experimentelle evidenz zur relevanz affektiver faktoren. Journal of Economic Psychology, 20(3), 319-337.
Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Performance, 76(2), 149-188.
Mellers, B. A., Schwartz, A., Ho, K., & Ritov, I. (1997). Decision affect theory: Emotional reactions to the outcomes of risky options. Psychological Science, 8(6), 423-429.
Redelmeier, D. A., & Tversky, A. (1992). On the belief that arthritis pain is related to the weather. Proceedings of the National Academy of Sciences, 93(14), 2895-2896.
Shafir, E., & Tversky, A. (1992). Thinking through uncertainty: Nonconsequential reasoning and choice. Cognitive Psychology, 24(4), 449-474.
Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211(4481), 453-458.