Incentivized experiments are frequently used to learn about individuals’ social, political, and economic behavior. However, public officials and other individuals are sometimes barred from accepting payment for anything related to their position, so money cannot be used in experiments (e.g., Butler and Kousser Reference Butler and Kousser2015). We assess whether donations to charity can be used to incentivize public officials, as an alternative to traditional monetary inducements.

We conducted our tests at the National Conference of State Legislatures (NCSL) and the National League of Cities (NLC), where we rented a booth in the exhibit hall and recruited attendees to take a survey.Footnote ¹ For the study, we randomized what, if any, incentives the officials received. One-third of officials were not given any incentive for their performance, another third were told that they would be paid based upon their performance, and the final third were told that money would be donated to a charity of their choice and that the amount of money would depend upon their performance.Footnote ²

Table 1 presents the effect of different incentives on the two outcomes we looked at.Footnote ³ Column 1 looks at the results from a quiz designed to measure political knowledge. The dependent variable measures, for each individual, the average distance from the correct answers.Footnote ⁴ The coefficients for both types of incentives were statistically significant at the 0.05 level and had a negative coefficient around −0.02, showing that individuals who received the incentives were closer to getting the correct answers. This effect is smaller than what previous researchers found when studying the effect of incentives on voters (Bullock et al. Reference Bullock, Gerber, Hill and Huber2015; Prior et al. Reference Prior, Sood and Khanna2015), but it is still substantively meaningful. The average value of the dependent variable was 0.29. Thus, receiving the incentives caused respondents to reduce their error rate by 7–9% ($\frac{-0.021}{0.29}=-0.07$ and $\frac{-0.027}{0.29}=-0.09$). These findings are consistent with previous results showing that respondents reduce their error rate when they are incentivized (see review in Morton and Williams Reference Morton and Williams2010).

Table 1 The Effect of Incentives on Performance and Risk Aversion

Note. Entries are coefficients with SEs in parentheses. *p < 0.05 (two-sided).

We also tested whether the types of incentives caused a change in how risk averse the participants were (see Column 2). We measured risk aversion using the survey item developed by Eckel and Grossman (2002). The results show that the incentives have no statistically distinguishable effect on officials’ risk aversion.

There is interest in expanding the types of subject pools we use for experiments (Kam et al. Reference Kam, Wilking and Zechmeister2007; Grose Reference Grose2014). As we expand the subject pool for incentivized experiments, we need to ensure that the type of incentives adopted are effective for the population we are studying (Morton Reference Morton2012). Our results suggest that when direct payments are not possible (see e.g., Butler and Kousser Reference Butler and Kousser2015), donations to charity are a promising alternative that can be used to incentivize public officials or other participants who cannot accept direct payments.