Do Self-Incentives and Self-Rewards Change Behavior? A Systematic Review and Meta-Analysis

Abstract

Encouraging people to self-incentivize (i.e., to reward themselves in the future if they are successful in changing their behavior) or self-reward (i.e., prompt people to reward themselves once they have successfully changed their behavior) are techniques that are frequently embedded within complex behavior change interventions. However, it is not clear whether self-incentives or self-rewards per se are effective at bringing about behavior change. Nine databases were searched alongside manual searching of systematic reviews and online research registers. One thousand four hundred papers were retrieved, spanning a range of behaviors, though the majority of included papers were in the domain of "health psychology". Ten studies matched the inclusion criteria for self-incentive but no studies were retrieved for self-reward. The present systematic review and meta-analysis is therefore the first to evaluate the unique effect of self-incentives on behavior change. Effect sizes were retrieved from 7 of the 10 studies. Analysis of the 7 studies produced a very small pooled effect size for self-incentives (k = 7, N = 1,161), which was statistically significant, d = 0.17, CI [0.06, 0.29]. The weak effect size and dearth of studies raises the question of why self-incentivizing is such a widely employed component of behavior change interventions. The present research opens up a new field of inquiry to establish: (a) whether or not self-incentivizing and self-rewarding are effective behavior change techniques, (b) whether self-incentives and self-rewards need to be deployed alongside other behavior change techniques, and, (c) when and for whom self-incentives and self-rewards could support effective behavior change.

Investigating which behaviour change techniques work for whom in which contexts delivered by what means: Proposal for an international collaboratory of Centres for Understanding Behaviour Change (CUBiC)

“…We have been using systematic review and meta‐analysis to explore what is known about the unique properties of behaviour change techniques (Figure 1) and how they interact with other behaviour change techniques. Thus far, we (Brown, Smith, Epton, & Armitage, 2018; Epton, Currie, & Armitage, 2017) have completed systematic reviews of BCTTv1 1.1 (goal setting [behaviour]), BCTTv1 1.3 (goal setting [outcome]), BCTTv1 10.7 (self‐incentives), and BCTTv1 10.9 (self‐rewards), and are currently working on BCTTv1 6.2 (social comparison), BCTTv1 6.3 (information about others’ approval), and BCTTv1 13.5 (identity associated with changed behaviour). These systematic reviews go some way to addressing the issue of which behaviour change techniques work best, for whom, in which contexts, delivered by what means, but each has identified yawning gaps in knowledge that require further primary research.…”

“…For example, we were unable to find a single study in which the behaviour change technique of self‐reward (BCTTv1 10.7) had been uniquely tested in a randomized trial (Brown et al, 2018). Similarly, despite extracting 384 effect sizes, Epton et al (2017) were unable to answer fully one of their key research questions, namely: ‘which behaviour change techniques enhance the effects of goal setting on behaviour change?’ due to the dearth of primary studies on the topic.…”

The Behavioral Medicine Research Council: Its origins, mission, and methods.

“…The BMRC also evaluates strategic preclinical research questions that, if resolved, would facilitate research on important clinical research questions. Some examples of potential SPRQs include needs for refined phenotypes or endophenotypes in research on behavioral and psychosocial risk factors (e.g., Brody, Yu, Barton, Miller, & Chen, 2017; Cuthbert, 2014; Huppertz et al, 2016), questions about biobehavioral mechanisms linking risk factors to medical outcomes (e.g., Cole et al, 2015; Lutgendorf & Andersen, 2015; Miller, Chen, & Parker, 2011; Suls, Green, & Davidson, 2016; Wirtz & von Kanel, 2017), and questions about behavior change processes as applied to behavioral or psychosocial risk factors for chronic medical illnesses (e.g., Brown, Smith, Epton, & Armitage, 2018; Epton, Currie, & Armitage, 2017; E. James et al, 2016; Larsen et al, 2017; Nielsen et al, 2018; Tate et al, 2016; Whelan, Morgan, Sherar, Orme, & Esliger, 2017; Winter, Sheats, & King, 2016).…”

Self-Incentives Uniquely Boost Cessation in Community-Based Stop Smoking Programs: Randomized Controlled Trial

“…For example, in one study, just 11% of participants who were asked to self-incentivize each month actually did so [19]. The question arises as to how to encourage people to self-incentivize [9].…”

“…Thus, self-incentive administration was dichotomized (i.e., 0 assigned to participants that had not incentivized at all; 1 assigned for participants that had incentivized at least once between baseline and 3month or 6-month follow-up, regardless of condition allocation) to tap whether participants had administered an incentive at least once versus not incentivizing at all.Data AnalysisA power calculation was undertaken to predict the sample size required for this study using self-reported measures of smoking status at follow-up, analyzed by chi-square. Based on previous research assessing the effects of single behavior change techniques[9,12,21,38,39] in order to detect a medium sized effect (d = 0.50,[40]) at a power of 80% and a significance level of 0.05, 108 participants were required across all three conditions at follow-up.Success of the randomization procedures was checked using MANOVA to verify equivalence of age, validated carbon monoxide outputs and nicotine dependence across conditions at baseline. Chi-square was used to assess randomization for gender, ethnicity, occupation and educational attainment.…”