*4.7. Consistency in Choices*

*Result 6: Reflective subjects are more likely to act consistently and in line with rational, self-interested behavior across all social dilemma tasks compared to intuitive subjects.*

A rather clear prediction for rational, self-interested behavior exists for the *dictator game*, *recipient's* acceptance threshold in the *ultimatum game,* and the *third-party punishment task*. However, the decision as *proposer* in the *ultimatum game* is rather difficult to classify as expectations for the decision of the *recipient* matter. Thus, any offer can be considered rational, self-interested if that is the lowest amount the *proposer* expects to be accepted.

Due to the ambiguity in what constitutes rational, self-interested behavior in the *ultimatum game proposer* decision, I will consider offering DKK 0 or DKK 10 and strategic fairness separately.

First I consider whether *reflective* subjects are more likely to transfer DKK 0 in *dictator game*, offer DKK 0 or DKK 10 as *proposer* in the *ultimatum game*, acceptance threshold of DKK 0 or DKK 10 as *recipient* in the *ultimatum game* and not opting for the *punishment* opportunity in the *third-party punishment task*.

13.4% of the *reflective* subjects complied with the above-mentioned as opposed to 1.6% of those not *reflective.* This difference is statistically significant (*<sup>p</sup>* <sup>=</sup> 0.008, *<sup>χ</sup>*<sup>2</sup> <sup>−</sup> *test*). When controlling for gender, *reflective* individuals are predicted to be 10.8%-points more likely than *non-reflective* subjects to choose as described in these tasks. *Reflective* subjects are predicted to choose as described with a probability of 12.7% as opposed to a predicted probability of 1.9% for those *non-reflective* (see Table 1).

A rational, self-interested individual could, as *proposer* in the *ultimatum game*, offer any share to the *recipient* if this is what the *proposer* believes to be the lowest amount to be accepted. However, in the *dictator game* there is no scope for such strategic considerations why a rational, self-interested individual would offer more as *proposer* in the *ultimatum game* relative to the transfer in *dictator game*. Considering whether *reflective* subjects are more likely to transfer DKK 0 in *dictator game*, have an acceptance threshold of DKK 0 or DKK 10 in the *ultimatum game*, exhibit strategic fairness as *proposer* in the *ultimatum game* and not opting for the *punishment* opportunity in the *third-party punishment task*, I find this to be the case. 20.9% of the *reflective* subjects complied with the above-mentioned as opposed to 6.3% of those not *reflective.* This difference is statistically significant (*<sup>p</sup>* <sup>=</sup> 0.008, *<sup>χ</sup>*<sup>2</sup> <sup>−</sup> *test*). When controlling for gender, *reflective* subjects are predicted to be 12.2%-points more likely than *non-reflective* subjects to choose as described in these tasks. *Reflective* subjects are predicted to choose as described with a probability of 19.8% as opposed to a predicted probability of 7.6% for those *non-reflective* (see Table 1).

**Table 1.** Marginal effects from Logistic regressions.


Standard errors in parentheses. \*\*\* *p* < 0.01, \*\* *p* < 0.05, \* *p* < 0.1. A: Dictator Game transfer = 0, B: Ultimatum Game offer = 0 or = 10, C: Ultimatum Game acceptance threshold = 0 or = 10, D: Strategic fairness; Ultimatum Game offer greater than Dictator Game transfer, E: Punishment sacrifice = 0, F: Compliance with A; B; C; E, G: Compliance with A; C; D; E.
