Etiska normer och agerande i ett ekonomiskt experiment

Tullberg, J. Etiska normer och agerande i ett ekonomiskt experiment

This article investigates the connection between values and behavior and consists of two parts. First, an investigation of values is made using a questionaire. Then an investigation is made regarding the behavior of participants in an economic experiment consisting of three variations of the ultimatum game.

The questionnaire
The inquiry regarding values has two purposes. One is to classify those questioned into three theoretical groups: integrists, reciprocals and altruists. Later, a number of behavioral hypotheses for these groups are tested in an experiment. The second purpose of the inquiry is to improve the model for the different theoretical groups. The 90 individuals tested were presented with a minimum of 49 questions. Most were value statements for which they were asked to note their level of agreement on a six point scale. Did the value statements, expected to be in accordance with each other based on the theoretical model, also in fact result in strong correlations? Were values expected to be opposed to one another also perceived that way by the test persons? A further ambition in this examination of the model was to find which questions were most salient in distinguishing between the three different groups.

Results
The investigation of correlation between different value statements resulted in 35 significant correlations. All of these correlations were in line with the model. (Questions resulting in significant positive correlation were expected in the model to be in accordance, and those with significant negative correlations were expected for questions opposed to one another.)

After two thirds of the test persons had been analyzed, one apparent result was that the three ethical groups were not equidistant. Although test subjects with high composite altruist scores were generally discernable from integrist and reciprocals, subjects showing high integrist scores often had high reciprocal scores. An effort was therefore made to improve the means for separating reciprocals and integrists. For the last third of test persons, the inquiry was expanded with 10 additional questions directed at distinguishing differences between integrists and reciprocals. By including these 10 new questions the number of significant correlations increased to 55, although 10 were against the model.

Finally, a short list of 15 questions was selected that had been run on the whole test group and proven to give the most salient results. This list is provided and suggested to be used in future classifications of test persons into the three ethical groups.

The experiment
The experimental part of this research consisted of three games:
1/ An Ultimatum game with an anonymous co-player
2/ A Dictator game with an anonymous co-player
3/ An Ultimatum game with face to face negotiation possibilities with a co-player

In an ultimatum game one player, the proposer, suggests a split of an amount (100 sek in this experiment) between himself and the other player, the responder. If the offer is equal or higher than the responder’s demand the ’ultimatum’ is considered accepted and the players are payed accordingly. If the offer is lower, the ultimatum is considered rejected and the two players get no money.

An important step prior to this experiment was an analysis I made of an experiment by Carter and Irons (1991). The amount the proposer wanted to hold for himself, the P-value, was much lower than that stipulated by game theory, but rational given the demands of responders (the R-value). It is rational to have a buffer - a safety margin between the amount in the game not to be surpassed and the sum of the R- and P-value. The theory and empirical evidence for such a proposer behavior is strong enough to consider it rational - even if in conflict with game theory.

The Responder behavior is still a problem, however. Why anyone, though more specifically which type of person, would ask for a lot more than the minimum amount when this request will not contribute to their reward, but increase the risk of getting nothing at all?

Results
One result associated with game 1, and in line with the hypothesis, was that reciprocal individuals were demanding significantly more as reponders than the other two ethical groups.

The giving in the Dictator game was unusually limited, also in the altruist group, when compared with games conducted by other researchers. A more intuitive expectation, given the context of this experiment - the game following an ethical test and an ultimatum game - would be an influence for giving a greater amount. The differences in P-value between the three groups were not significant.

In the negotiation game it was expected that the buffer should go down and so it did. The integrists kept a rather high buffer (9 sek compared to 3 and 2 for the reciprocals and the altruists, respectively), a behavior in line with the model. One hypothesis was that the proposer should get the major part of the cut in the buffer. However, not only did the responder get all of the buffer, but he also got the proposer to cut down on his P-value. The result of this game was a more equivalent split between proposer and responder for all three groups.

Values and behavior
Respondents were asked to make normative statements and descriptive estimates of the ultimatum game. In deciding P-value in game 1 (P1), it seems rational for the proposer to take his descriptive estimate of the other player into account. So the correlation shown here was no surprise. For the decision on R-value (R1), there is no rationale for such a connection, but the expectation was a strong correlation to the normative position (RN). Surprisingly the R-value was more strongly correlated to the descriptive (RD) than to the normative (R1 - RD was 0.77, and R1 - RN was 0.40).

The same descriptive link appeared in the Dictator game. P2 was more highly correlated to RD than to RN. These results give an interesting, if disappointing, indication that behavior is more closely tied, or associated, with the behavior of others, even when there exists an excellent possibility to follow one’s own judgement.

The influence of interest upon values was investigated in two different ways. One hypothesis was that high performers would be more oriented towards integrist values, while low performers would have preferences for altruistic values. The performance of the three groups in the word-game, however, showed small and insignificant differences between groups, with altruists doing better than integrists and reciprocals being on top.

The second investigation of connection with interest gave a stronger result. Subjects were asked to evaluate three societies (each close to one of the three archetypes) relative to the degree that such a society would ’benefit my interests’. All three groups gave the highest rating to ’a mutually beneficial society’. Reciprocals and integrists also gave high ratings for an ’all by himself society’. The altruists gave a low evaluation of such a society and gave a high estimate of ’a caring for others society’. The question of what is in one’s own interest is of importance. Altruism is not only about giving, but about receiving, and what kind of return one can expect in different societies has no evident answer.

To reduce the theoretical influence of the investigation, categorizations were made from performed behavior rather than from the perspective of ethical groups. In game 1, equalitarians (keep 50 and give 50 as proposer and demand 50 as responder) were compared to non-equalitarians. In game 2, givers (those contributing anything over zero) were compared to non-givers. In game 3, differentiers (those ending up in something else than 50-50) were compared to non-differentiers.

- Equalitarians gave significantly different answers to 7 questions all having altruistic or reciprocal implications. They also showed a significant difference in index exhibiting a lower integrity value than non-equalitarians.
- Givers showed significant deviation for 13 questions and the 3 ethical indices. They exhibited a high value for altruism, and low values for reciprocity and integrity. All 13 specific questions showed accordance with a position classified in the model as exclusively altruistic.
- Differentiers showed only one significant result which can appear puzzling. They were negative to the proposition that contracts should be kept. The context of this question indicated flexibility, so the answer had been classified as integristic. The expectation had been that more answers of integristic and reciprocal character would show correlation with differentiers.

The paper ends with summarizing the main findings of this study. A few suggestions are made for further research, investigating the connection between norms and behavior by the use of economic experiments.