Results

Before any statistics could be calculated, the data had to be converted into an analyzable format. For each subject’s answer sheet, each sure-option response was coded as “1” and each risky-option response was coded as “0.” These numbers were then entered into a spreadsheet using Microsoft Excel. Each subject responded legibly to all questions, so no correction for error was needed and all data was used. Because “1” represented a sure-option response, averages of the scores represented the percentage of total sure-option responding.

For each hypothesis, a statistical analysis was calculated using a two-tailed, paired t-test. To test the existence of the framing effect, the percentages of total sure-option responding were compared between gain-framed questions and loss-framed questions. Fig. 1 shows that the percentage of sure-option responses for the gain-framed questions (M = .5795) was significantly higher than sure-option responses for the loss-framed questions (M = .4545; t = 2.62, p < .0121). To see if subjects were generally less risky in scenarios involving a larger payoff, differences in sure-option responses were examined between high-payoff and low-payoff scenarios. Fig. 2 shows significant differences between high-payoff (M = .6193) and low-payoff responses (M=.4148; t = 4.02, p < .0002). To examine any differences in the prominence of the framing effect in high-payoff and low-payoff questions (and particularly if this prominence is greater in the high-payoff questions), an interaction was calculated. The difference of sure-option responding (between gain-framed questions and loss-framed questions) was, surprisingly, significantly larger in low payoff scenarios (mean difference = .2386, t = 4.33, p < .0001). and not significant in the high payoff scenarios (mean difference = .0114, t = .15 , p < .881). This did, however, lead to an interaction as shown in Fig. 3 (t = 2.49, p < .017).