Mean, Kind, or Monetary: Does Feedback Form Affect Task Performance?
This paper contributes to the active line of literature on feedback and performance by questioning whether the feedback form matters. While there is evidence that explicit (monetary) incentives can improve performance (Audi and Locke, 2003; Eriksson et al., 2009; Schmidt et al., 2020), non-explicit incentives also matter (Fontaine et al., 2007). We thus design an experiment to examine how explicit and non-explicit incentives affect task performance.
The experiment consists of three treatments, in addition to a control, in a between- subjects design. In each treatment, subjects play a real-effort task where they encode letters to numbers for 20 rounds. In all three treatments, subjects play the first 10 rounds without any form of feedback. After the 10th round, audio feedback plays every 2 rounds. In the first treatment (FinT), the payout is correlated with performance, and the audio feedback refers to whether the subject won or missed a payout of x in the preceding round. In the other two treatments, subjects receive a flat payout irrespective of their performance in the game. The audio feedback still plays every two rounds after the 10th.
In the second treatment (KindT), the audio refers to whether or not the subject has performed the preceding task correctly using encouraging words. The same applies to the third treatment (MeanT) except that the audio uses discouraging or scolding words. In the control group, subjects play the 20 rounds without any feedback and receive a flat payout irrespective of their performance. The average payout for all four groups is the same.
Post-experiment, subjects respond to a relatively long questionnaire (52 questions) on their demographics and personality traits. We include attention checks after the 23rd, the 33rd, and the 45th questions. Response to the questionnaire was not remunerated.
We hypothesize the following. H1: Pay-by-performance positively affects performance, i.e. FinT is correlated with higher performance relative to the control group. H2: Given a flat reward, feedback, regardless of its form, enhances performance, i.e. KindT and MeanT enhance performance relative to the control group. H3: Given a flat reward, encouraging feedback KindT enhances performance relative to discouraging feedback MeanT.
Employing a sample of 348 students and non-students from Egypt, we find that: (a) subjects in FinT perform significantly better than the control group in terms of accuracy but not with respect to the time taken to perform the task. (b) We also find that in KindT and MeanT subjects significantly got more correct answers relative to the control. Interestingly enough, subjects in the MeanT are also significantly faster. This result of positive effects of feedback on performance extended to the non-remunerated post-experiment questionnaire, where subjects in both groups significantly did better in the attention checks questions compared to the control group. (c) In contrast to H3, we find that performance is better in MeanT than in KindT.
We conclude that feedback matters. While explicit rewards for performance are important, non-explicit rewards can also play a role and their effects may last longer.