Instructions, Choices, and Data for Stewart, Reimers, and Harris

README.txt key

Experiment 1A

Experiment 1A Positive Skew Instructions

Experiment 1A Negative Skew Instructions

Experiment 1A Positive Skew Choices

Experiment 1A Negative Skew Choices

Experiment 1A Raw Data

I've also included some example R code for the analysis, as described in the Appendix

Experiment 1A Analysis R Script

Experiment 1A Analysis R Script Output

Experiment 1A amount map needed for analysis

Experiment 1B

Experiment 1B Positive Skew Instructions

Experiment 1B Uniform Instructions

Experiment 1B Positive Skew Choices

Experiment 1B Uniform Choices

Experiment 1B Raw Data

Experiment 1C

Experiment 1C Positive Skew Instructions

Experiment 1C Uniform Instructions

Experiment 1C Positive Skew Choices

Experiment 1C Uniform Choices

Experiment 1C Raw Data

Experiment 2A

Experiment 2A Positive Skew Instructions

Experiment 2A Negative Skew Instructions

Experiment 2A Positive Skew Choices

Experiment 2A Negative Skew Choices

Experiment 2A Raw Data

Experiment 2B

Experiment 2B Positive Skew Instructions

Experiment 2B Negative Skew Instructions

Experiment 2B Positive Skew Choices

Experiment 2B Negative Skew Choices

Experiment 2B Raw Data


Thank you to Despoina Alempaki for pointing out an error in the description of the analysis for Experiment 2B. In the Method section for Experiment 2B it states that "No participant violated dominance in more than 10% of catch trials, so all data were retained." This is wrong. In the analysis in the paper, three participants (ids 12, 17 and 18) were deleted for violating dominance on more than 10% of catch trials. A fourth participant (id 23) also violated dominance but was not deleted because I failed to see his or her rate was above 10%.

Deleting Participant 23 makes very little difference to the results. Differences between conditions remain significant. Here is a revised Section 6.5.2, with differences in red.

6.5.2. Results and Discussion. Figure 5(b) shows that the weighting function in the positive-skew condition is more concave than the negative-skew condition. Modeling using power law weighting functions reveals a significant difference between conditions (χ2(1) = 119.6, p < 0.0001). Modeling with free weightings tested whether the weighting for the common probability 50% differed across conditions. The weighting of 50% is significantly higher in the positive-skew condition (χ2(1) = 22.7, p < 0.0001), consistent with the rank hypothesis.

Here is a revised Figure B.1 where the confidence intervals in the Experiment 2B columns are slightly moved.

Experiment 3A

Experiment 3A Positive Skew Instructions

Experiment 3A Negative Skew Instructions

Experiment 3A Positive Skew Choices

Experiment 3A Negative Skew Choices

Experiment 3A Raw Data

Experiment 3B

Experiment 3B Positive Skew Instructions

Experiment 3B Negative Skew Instructions

Experiment 3B Positive Skew Choices

Experiment 3B Negative Skew Choices

Experiment 3B Raw Data

Experiment 4

Experiment 4 Positive Skew Instructions

Experiment 4 Uniform Instructions

Experiment 4 Positive Skew Choices

Experiment 4 Uniform Choices

Experiment 4 Raw Data