Expectations
Based on our hypothesis we measured two dependent variables: rational choice and self-interested behavior. To test the hypotheses, we examined how both rational decision making and self-interest can be measured by using similar experiments presented in prior research. We use the outcome of previous papers to compare results of our experiment to those expectations.
These experiments have gained attention because the empirical results differed dramatically from the predictions of game theory, which assumes self-interest (Camerer & Thaler, 1995).
Based on the findings of Kahneman et al. (1986) we should expect to see the majority of respondents to equally split their endowment in the ultimatum game in a normal economic situation. Furthermore, research by Bekkers (2007) based on the dictator game showed that donations increase with age and educational level. These factors are also taken into account in this studies’ data analysis. Additionally, in the dictator game the mean amount donated is 20%
of the endowment when studying research by Camerer (2003).
These choices clearly go against traditional economic theory. Therefore, we study whether the results differ in a crisis situation. In case of an increase in self-interested behavior we should see an increase in respondents to keep their entire endowment to themselves (Bekkers, 2007).
Furthermore, with regards to rationality, when the respondent is asked which amount they would minimally accept in the reversed ultimatum game, they should accept the least monetary offer. After all, a typical econ would always decide the possibility of having the highest net benefit and therefor avoid receiving nothing (Bankovic, 2019). Furthermore, subjects in ultimatum games tend to avoid ‘unfair’ offers and offer a fair split. This charitable behavior is considered as selfless by behavioral economists (Mullainathan & Thaler, 2000).
Survey results
In this section the survey results are revealed. The survey experiment has been completed by 115 individuals ranging from 20-55+ years old who are living in the Netherlands. The sample consists of people differing not only in age, but also gender and level of education. The gender split is equal with approximately 50% of the 115 respondents being man and 50% woman. We have taken in account the characteristics gender, age and education in our statistical analysis which will follow later in this results section. Table 1 displays the number of our sample based on each characteristic. In the following sub-sections we will discuss the results of our survey, followed by a statistical analysis to test the significance of our findings.
Age Education Gender
20-30 years 77 MBO 17 Man 57
30-40 years 26 Bachelor 56 Woman 58
40-55 years 10 Master 41
55+ 2 Doctorate 1
Total 115 Total 115 Total 115
Table 1. Sample
Dictator game results
In Figure 3 we find the results of the first question related to the dictator game. The respondents were asked how much money they would share with the person in room B who they do not know and cannot see or communicate with. Following the rules of a typical dictator game, the other person must accept any offer received. We notice that the results of this experiment are in line with statements by Camerer (2003), arguing that in the dictator game the mean amount donated is 20% of the endowment. In these results the mean lies between 20% and 25% of the endowment. We also test whether these results change when the stakes are increased from 100
euro’s to 500 euro’s. Figure 4 shows that the mean amount donated stays approximately 20%
of the endowment. All in all, our dictator game results reveal that the majority of the respondents (almost 55%) chose to share at least 20% of their endowment with a stranger that they cannot see and will not meet. This result is consistent with earlier dictator game experiments.
Figure 3. Dictator game results before crisis with a 100 euros stake
Figure 4. Dictator game results before crisis with a 500 euros stake
As explained, dictator game experiment is done in two scenarios. The results in Figure 3 and 4 display the results in a normal setting before an economic crisis. The results of the dictator game experiment ‘during crisis’ are displayed in Figure 5. We read that the mean stays between 20%
and 30% of the endowment. A similar mean result is displayed when the stakes are increased from 100 euros to 500 euros as seen in Figure 6. The absolute percentages however do slightly change compared to the dictator game in our first (no crisis) scenario. These results tell that during the crisis scenario, the percentage of people that give away any amount to a stranger increase to 60% in case of a 100euro’s stake and even 65% when there is a 500 euros stake.
Figure 5. Dictator game results during crisis with a 100 euros stake
Figure 6. Dictator game results during crisis with a 500 euros stake
Ultimatum game results
The ultimatum game has a similar setting as the dictator game in the experiment. However, the respondents are told that the rules are slightly different and that the stranger can either accept or reject their offer. In case of a rejection, both the respondent and the stranger will get 0 euros.
The results of our survey are consistent with the findings of Kahneman et al. (1986) who states that the majority of respondents equally split their endowment in the ultimatum game. Figure 7 clearly displays that over 60% of the respondents in our ultimatum game experiment split half of their endowment with a stranger. When the stakes are increased, we notice a slight decline in the amount of people to offer at least half of their endowment as seen in Figure 8. Looking at the ultimatum game results ‘during crisis’ in Figure 9 and Figure 10 we again notice that also the mean amount offered by the subject to the stranger declines as the stakes increase.
Figure 7. Ultimatum game results before crisis with a 100 euros stake
Figure 8. Ultimatum game results before crisis with a 500 euros stake
Figure 9. Ultimatum game results during crisis with a 100 euros stake
Figure 10. Ultimatum game results during crisis with a 500 euros stake
Reversed ultimatum game results
The difference between what is called the ‘reversed ultimatum game’ and the original ultimatum game can be found in its name. The roles of subject and stranger are reversed, making it possible for the subject to either accept or reject an offer. According to previous results, the majority in the ultimatum game tends to split the amount equally. However, in the reversed ultimatum game most respondents accepts less than half of the endowment as a sufficient offer.
This corresponds with the results discussed in Güth & Teitz (1990) and Camerer & Thaler (1995), where respondents frequently reject offers below 30% (Gintis, 2000). When the stakes are increased, we see that the outcomes stay almost the same in Figure 11 and 12. The same accounts for the choices during crisis, see Figure 13 and 14, which both show similar results as the ones before crisis.
Figure 11. Reversed ultimatum game results before crisis with a 100 euros stake
Figure 12. Reversed ultimatum game results before crisis with a 100 euros stake
Figure 13. Reversed ultimatum game results during crisis with a 100 euros stake
Figure 14. Reversed ultimatum game results during crisis with a 500 euros stake
Statistical analysis
Based on the survey results we observed differences between the decisions of our subject before and after crisis. Through a statistical analysis we can test whether these differences between choices made before and after crisis are significant. Below you can find three tables which correspond to each of our three experiments: dictator game, ultimatum game and reversed ultimatum game. Since we are specifically interested in analyzing differences in time, we have paired each set of questions before crisis (e.g., Q1-Q3) to a set of question after crisis (e.g., Q1b-Q3b). Looking at the p-value of the set of questions referring to the dictator game experiments in Table 2, one can see that its p-value is smaller than 0.05. The same accounts for both the ultimatum game results in Table 3 and reversed ultimatum game results in Table 4, which also indicate p<0.05.
Table 2. Statistical significance difference dictator game results
Table 3. Statistical significance difference ultimatum game results
Table 4. Statistical significance difference reversed ultimatum game results
It is also interesting to find out whether there are any significant variations in gender-based responses in the sample, next to age and education. We perform a mixed ANOVA analysis to see if this is true. Other than results from Havens et al. (2006) show us, we cannot say that age or education have any significant effect on the amount given away in our experiment. We can also find in Table 5 that p<0.05 in case of gender. Therefore, these findings are in line with Bekkers (2007) who states that in the analysis of the amount donated, females and males do not differ significantly.
Table 4. Statistical significance difference based on characteristics (Q19= gender, Q20= age, Q21= education)