Effort and Accuracy in Social Preferences

  Copyright: © FCN Overview of experimental results for between-subject treatments; social index is the average number out of 8 adapted-dictator-game tradeoffs that was chosen most beneficially for the partner

Funded by the Excellence Initiative of the German federal and state governments

Social preferences, i.e. that people take the effect of their choices on others into account, are often observed in lab experiments. E.g., in dictator games participants tend to voluntarily share a large part of an endowment with a passive co-participant. However, this behavior may not always correctly reflect real-world preferences. If the causes for such inconsistency are identified, these can be used to improve both the validity of further experiments and the efficacy of real-world applications. In this experiment, I investigate a new potential source of inconsistency: the larger complexity of real-world settings, e.g., by which it may not be obvious what the effect of individuals’ choices is on themselves and on others. This may lead to a tradeoff between accuracy in making the right choices and saving effort in finding out the exact choice effect by using simpler choice rules. My hypothesis is that the willingness to sacrifice accuracy in order to save effort is significantly larger when it concerns others' payoff than when it concerns own payoff. Thus, in more complex situations, people tend to focus on the effect of their choices on themselves. The experiment explicitly controls for the extent that shifts in social preferences are explained by strategic ignorance and real effort choice. It involves 8 adapted dictator games with random partners and varied tradeoffs between own and partner’s payoff and uses a payoff table to manipulate complexity between subjects:

0. payoff information in a covered decision matrix is revealed by a single click to reflect the transparency of a typical lab setting (baseline treatment)

1. payoff information in the covered boxes of a decision matrix is revealed by separate clicks (strategic ignorance treatment)

2. payoff information in the covered boxes of a decision matrix is revealed by performing separate counting tasks sufficiently well (real effort treatment)

3. payoff information in the boxes of a decision matrix is presented by means of a simple function of the outcomes of separate counting tasks to reflect real-world opacity of choice effects (complexity treatment)

The experiment with an extensive debriefing has been executed with z-Tree in the local infrastructure in November/December 2015 with 204 subjects, and the following additional control treatments have been performed in March/April 2016 with 146 subjects:

2b. payoff information in the covered boxes of a decision matrix is revealed in green or red resp. by performing separate counting tasks well enough or not (performance motivation)

2c. payoff information in the covered boxes of a decision matrix is revealed by performing separate slider tasks sufficiently well (performance type)

3b. payoff information in the boxes of a decision matrix is presented by means of a simple function of the outcomes of separate counting tasks, which can be explicitly revealed by performing these tasks sufficiently well (registration effect)

An overview of the main results for all treatments is given in Fig. 1. The results of the experiment confirm my hypothesis that in complex choice settings, the willingness to sacrifice accuracy in order to save effort is significantly larger when it concerns others' payoff than when it concerns own payoff. This is due to the already known phenomenon of strategic ignorance (treatment 1). Interestingly, registering performance in some real effort task leading to payoff revelation significantly reduces strategic ignorance and therefore supports social choice (treatments 2, 2c, and 3b), which has obvious marketing implications. However, when such registration does not take place, like in real-world complexity, strategic ignorance increases again, which makes the level of social choice in treatment 3 not statistically distinguishable from its level in treatment 1.

Acknowledgement

This project is supported by research assistance of Jana Baur and Jannik Wendorff and funded by the Excellence Initiative of the German federal and state governments. Help from Zachary Grossman as well as lab management and seminar participants at the School of Business and Economics at RWTH Aachen University is appreciated.

Project Publications

Harmsen – van Hout M.J.W. (2017). Effort and Accuracy in Social Preferences, FCN Working Paper No. 15/2017, Institute for Future Energy Consumer Needs and Behavior, RWTH Aachen University, November.