Aim and theoretical background: The controversy over extrinsic reward effects on intrinsic motivation (IM) has persisted for 50 years in the field of experimental psychology. The debate continues even today due to partially mixed results of the past meta-analyses. While some meta-analyses have provided strong support for the undermining of intrinsic motivation by extrinsic rewards (e.g., Deci, Koestner and Ryan 1999; Tang & Hall 1995), others suggest that the negative effects are very limited and easily avoidable (e.g., Cameron et al. 2001; Eisenberger et al., 1999; Eisenberger & Cameron, 1996).

Recently, there has been a growing interest in this topic in the field of organizational behavior. However, a lack of robust evidence and a lack of quantitative synthesis of the extrinsic rewards–intrinsic work motivation relationship has caused some scholars (e.g., Gerhart & Fang 2015) to be highly critical about the undermining effect. Indeed, some scholars have openly questioned the relevance of this phenomenon in work settings.

Hence, the impact of rewards on IM remains highly debated, as noted, for example, by Van den Broeck et al. (2021). Drawing on cognitive evaluation theory (CET; Deci & Ryan 1985), the present meta-analysis tries to reconcile these debates by synthesizing experimental laboratory research and also observational research conducted in work settings.

Methodology: Meta-analysis was used to synthesize the published literature from the fields of experimental psychology and organizational behavior. The data was searched systematically from seven electronic databases. A total of 140 randomized controlled experiments and 42 observational studies were included in the analyses. These studies were drawn from 157 peer-reviewed articles published between 1971 and 2020.

Following Deci et al.’s (1999) approach, included studies were classified based on the reward contingency and reward type. The data was analyzed using a random-effects model and a hierarchical analysis framework (see Deci et al. 1999). Three separate meta-analyses were conducted. Two meta-analyses examined the effects of rewards on intrinsic motivation in controlled experiments using free-choice behavior/persistence and self-reported interest/enjoyment measures of IM. The third examined the association between rewards and self-reported intrinsic work motivation in observational studies. Comprehensive Meta-Analysis software was used to perform the meta-analyses (Borenstein et al. 2013).

Results:
For the meta-analysis of experimental studies using the free-choice behavioral measure of IM, the results showed that extrinsic rewards have a statistically significant negative effect on free-choice behavior after rewards are withdrawn. The undermining of free-choice IM by extrinsic rewards was evident for all rewards (d = –0.28), tangible rewards (d = –0.39), and expected rewards (d = –0.41). More specifically, extrinsic rewards significantly undermined free-choice behavior when rewards were contingent on task engagement (d = –0.57), task completion (d = –0.36), or task performance (d = –0.23), while having nonsignificant effects on free-choice IM when the rewards were task-noncontingent (d = 0.10) or unexpected (d = –0.04). For positive feedback, a positive effect was found (d = 0.33). A supplementary analysis showed that performance-contingent reward (PCR) nonattainment also undermined free-choice behavior (d = –0.66).

For the meta-analysis of self-reported interest, the results showed weaker composite effects. The only significant effects were found for positive feedback (d = 0.31), engagement-contingent rewards (d = –0.16) and performance-contingent rewards (d = 0.11). However, the overall pattern of effects was similar to the free-choice behavior analysis. A supplementary analysis on PCR nonattainment showed undermining of self-reported interest (d = –0.70).

For the meta-analysis of observational studies, no significant overall level association was found. More detailed analyses showed that both self-reported positive feedback (r = 0.19) and base salary (i.e., task-noncontingent rewards) (r = 0.19) correlated significantly with self-reported intrinsic work motivation. The relationship between intrinsic work motivation and performance-based rewards (PBRs) was nonsignificant (r = 0.05) and extremely heterogeneous. A supplementary analysis showed a weak negative correlation for the PBRs–intrinsic work motivation relationship when PBRs were perceived as being more controlling than informational (r = –0.10).

Limitations: The study excluded unpublished studies.

Research/practical implications: These findings advance our understanding of the undermining effect in experimental and work settings.

Originality/value: This is the first meta-analysis to replicate and extend Deci et al.’s (1999) findings. Moreover, previous meta-analyses have not examined the extrinsic rewards–intrinsic motivation relationship in organizational settings. In summary, this meta-analytical study provides support for CET and the existence of the undermining effect. However, the effects of rewards on IM are partially dependent on the utilized measure of IM (behavioral vs. self-report), type reward-contingency, and the context of research (laboratory vs. organization). Several factors were also identified that might help to explain why reward effects were stronger for experimental than observational studies.