Framing Effects and Risk Preferences in Medical Decision Making: Evidence from Online and Laboratory Samples

Abstract

The framing effect—reversals in preference when logically equivalent choices are presented in gains versus losses—is a foundational phenomenon in decision research. We tested whether framing effects persist in a contemporary online survey context and whether they interact with decision domain and individual differences in numeracy. Participants (N = 142 online, N = 58 laboratory) chose between medical treatment options. Risk preferences reversed as a function of frame, consistent with prospect theory.

Introduction

Since Tversky and Kahneman's seminal work (1981), the framing effect has been established as a robust violation of expected utility theory and rational choice models. When decision options are presented in terms of gains, decision makers tend toward risk-averse choices; the identical options presented in terms of losses typically produce risk-seeking preferences. This phenomenon suggests that choice is path-dependent, influenced not only by objective outcomes but by how those outcomes are mentally represented or "framed."

However, most empirical investigations of framing have employed convenience samples in laboratory settings with artificial gambles over hypothetical monetary gains and losses. As internet-based research methods have become increasingly prevalent in psychological science, questions arise regarding whether classical framing effects replicate in online survey contexts and whether effects are moderated by sample characteristics, domain specificity, or individual differences in analytical reasoning. The present study addresses these gaps by administering a framing task to both laboratory and online participants using a medical decision domain hypothesized to be more representative of real-world choices.

Method

Participants

Two hundred participants completed the study: 142 recruited through an online survey panel (mean age = 41.2 years, SD = 12.8; 64% female; mean education = 14.1 years) and 58 recruited from university subject pools and community groups (mean age = 23.7 years, SD = 5.4; 59% female). Online participants received small monetary incentives; laboratory participants received course credit or $10 cash compensation. All participants provided informed consent.

Procedure

Participants completed a medical decision scenario adapted from Tversky and Kahneman (1981). In the gain frame, participants chose between a certain medical treatment (assured recovery of 50% of function) or a gamble (80% chance of full recovery, 20% chance of no recovery). In the loss frame, participants chose between a certain treatment (assured loss of 50% of function) or a gamble (80% chance of no loss, 20% chance of total loss). Frame (gain vs. loss) and sample (online vs. laboratory) were manipulated between subjects. After the primary choice, participants completed a 10-item numeracy scale (Lipkus, Samsa, & Rimer, 2001) and provided demographics.

Results

Overall, 68% of participants in the gain frame chose the certain option, whereas only 29% chose the certain option in the loss frame, χ²(1, N = 200) = 31.44, p < .001. This classic reversal pattern was observed in both the online sample (gain: 71%, certain; loss: 28%, certain) and laboratory sample (gain: 62%, certain; loss: 31%, certain), interaction χ²(1, N = 200) = 1.18, p = .277. Numeracy was a weak but significant moderator of framing susceptibility: higher numeracy was associated with slightly reduced risk-seeking in the loss frame, r = .24, p = .038.

A binary logistic regression predicting certainty-seeking across both groups revealed that frame (loss vs. gain) was the dominant predictor (B = 2.34, SE = 0.48, p < .001), while sample type and numeracy contributed minimally (p > .15 and p = .052, respectively). Effect sizes (odds ratios) indicated that loss framing increased the odds of choosing the gamble by a factor of 4.5.

Discussion

This investigation demonstrates that the framing effect is robust across decades, methodologies, and participant populations. Internet-based survey administration does not attenuate the classical reversal in risk preferences, suggesting that online research methods are sufficiently valid for replicating established decision phenomena. The medical domain may heighten emotional or self-relevant processing compared to abstract monetary gambles, yet framing effects persisted with similar magnitude, indicating that the phenomenon is domain-general.

The weak moderating role of numeracy contradicts some recent proposals that improved statistical literacy should reduce heuristic biases. Instead, results suggest that framing effects may reflect deeper structural properties of mental representation rather than deficits in calculation or logic. Future research employing neuroimaging (fMRI, EEG) or process-tracing methods (eye tracking, think-aloud protocols) may illuminate whether frames activate different neural systems or alter the sequence of information processing.

References

Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211(4481), 453-458.
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263-291.
Lipkus, I. M., Samsa, G., & Rimer, B. K. (2001). General performance on a numeracy scale among highly educated samples. Medical Decision Making, 21(1), 37-44.
Johnson, E. J., Häubl, G., & Keinan, A. (2007). Aspects of endowment: A query theory of value construction. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(3), 461-474.
Stanovich, K. E., & West, R. F. (2000). Individual differences in reasoning: Implications for the rationality debate? Behavioral and Brain Sciences, 23(5), 645-665.
Levin, I. P., Schneider, S. L., & Gaeth, G. J. (2002). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Performance, 76(2), 149-188.