ArticlePDF Available

Abstract and Figures

When people engage in a first pro-environmental behaviour (PEB1; for example, conserving energy at home), are they more or less likely (positive and negative spillover, respectively) to engage in other pro-environmental behaviours (‘PEB2’; for example, conserving water at home)? We examined evidence for spillover using a meta-analysis of interventions. We coded 22 studies and unpublished data that fulfilled the following criteria: used experimental or quasi-experimental design, showed change in a PEB1 and measured at least one PEB2. Analysis of the 77 effect sizes found in these studies showed that the overall spillover from a PEB1 was positive, though small, on the intention to perform a PEB2 (pooled mean effect size estimate d+ = 0.17). However, the spillover effect was negative and small for actual behaviour (d+= −0.03) and policy support (d+ = −0.01) for PEB2. Positive spillover was most likely when interventions targeted intrinsic motivation and when PEB1 and PEB2 were similar. Future research in the area should target and measure spillover processes, collect larger samples and statistically test for spillover in more consistent ways.
This content is subject to copyright. Terms and conditions apply.
Meta-analysis of pro-environmental behaviour spillover
Alexander Maki, Vanderbilt University
Amanda R. Carrico, University of Colorado Boulder*
Kaitlin T. Raimi, University of Michigan*
Heather Barnes Truelove, University of North Florida*
Brandon Araujo, University of North Florida
Kam Leung Yeung, University of North Florida
Author note: The second, third, and fourth authors contributed equally, and their order was
determined alphabetically.
This work was supported by the National Science Foundation (grant #: SES-1325660),
Vanderbilt Institute for Energy and Environment, the Climate Change Research Network, and
the Vanderbilt University Trans-Institutional Program.
Correspondence concerning this article may be addressed to Alexander Maki, Vanderbilt
Institute of Energy and Environment, Vanderbilt University, 159 Buttrick Hall, PMB 407702,
2301 Vanderbilt Place, Nashville, TN 37240. E-mail:
Conflict of interest statement: The authors declare no competing interests.
Acknowledgements: The authors would like to thank Roland Davis and Nathan Lee-Ammons for
their assistance with data collection/coding, and all of the generous researchers who helped us
with the coding of their data/manuscripts/articles.
Author contribution statement: A.M. contributed study conceptualisation, data collection/coding,
data analysis, results interpretation, and drafting of the manuscript. A.R.C. contributed funding,
study conceptualisation, results interpretation, and drafting of the manuscript. K.T.R. contributed
funding, study conceptualisation, results interpretation, and drafting of the manuscript. H.B.T.
contributed funding, study conceptualisation, results interpretation, and drafting of the
manuscript. B.A. contributed data collection/coding and drafting of the manuscript. K.L.Y.
contributed study conceptualisation, data collection/coding, results interpretation, and drafting of
the manuscript.
When people engage in a first pro-environmental behaviour (“PEB1”; e.g., conserving energy at
home), are they more or less likely (positive and negative spillover respectively) to engage in
other pro-environmental behaviours (“PEB2”; e.g., conserving water at home)? We examined
evidence for spillover using a meta-analysis of interventions. We coded 22 studies and
unpublished data that fulfilled the following criteria: used experimental or quasi-experimental
design, demonstrated change in a PEB1, and measured at least one PEB2. The analysis of the 77
effect sizes found in these studies showed that the overall spillover from a PEB1 was positive,
though small, on the intention to perform a PEB2 (d+= .17). However, the spillover effect was
negative and small for actual behaviour (d+= -.03) and policy support (d+= -.01) for PEB2.
Positive spillover was most likely when interventions targeted intrinsic motivation and when
PEB1 and PEB2 were similar. Future research in the area should target and measure spillover
processes, collect larger samples, and statistically test for spillover in more consistent ways.
Keywords: Spillover, environmental behaviour, rebound, gateway, meta-analysis
Meta-Analysis of Pro-environmental Behaviour Spillover
Scholars and decision-makers have increasingly applied insights from the social and
behavioural sciences (e.g., “nudging”) to motivate voluntary pro-environmental actions1-3. These
approaches have been successful across many contexts4-6. Yet, the vast majority of this literature
considers only the effects of interventions on targeted pro-environmental behaviours (PEBs),
without examining effects on non-targeted PEBs. Like health-related decisions (e.g., seatbelt use
and driving7), environmental decisions may be prone to behavioural spillover effects whereby
the adoption of one PEB (PEB1) reduces (negative spillover) or increases (positive spillover) the
likelihood of adopting other PEBs (PEB2)8,9.
Spillover effects may stem from several distinct psychological processes9-14. Negative
spillover is often attributed to moral licensing, such that an individual feels morally “off the
hook” after a prosocial act and is less inclined to adopt other prosocial acts. Alternatively,
positive spillover may result from a desire for consistency across behaviours or because PEBs
prime environmental concern. Regardless of the mechanism, spillover effects could dramatically
alter the net effects and cost efficiency of behavioural interventions. If negative spillover is
common, we may overestimate the environmental benefits of behavioural interventions. If
positive spillover is common, we may underestimate such benefits and miss opportunities to
engage the public in more impactful secondary behaviour changes.
Despite recent interest in PEB spillover, there is no quantitative synthesis of this
literature. Existing research has found mixed effects, with some studies suggesting negative
spillover, some suggesting positive spillover, and some finding no evidence of spillover9,12,13,15.
Measurement inconsistencies may be one cause of conflicting findings—studies vary in whether
they measure actual behaviour15,16, behavioural intentions17, self-reported behaviour14,18, or
policy support19,20. Though all are useful indicators for evaluating environmental interventions,
intentions and policy support are theoretically and practically distinct from actual behaviour21.
Further, although evidence that spillover is non-existent would be practically significant, null
effects are sometimes met with scepticism from editors and reviewers and many researchers
never pursue publication22-24. Meta-analysis is a valuable tool for integrating results (published
or not) to generate more precise and generalizable statistical conclusions25,26.
This meta-analysis is the first to assess the presence and magnitude of spillover effects
within the context of pro-environmental behaviour. We examine spillover to pro-environmental
behaviour, intentions, and policy support by testing nine pre-registered hypotheses and three
exploratory questions.
H1: Spillover to intentions and policy support will be stronger than to behaviour
Behavioural intentions often precede behaviour21, and meta-analytic evidence suggests
that it is easier to predict and influence intentions than behaviour27,28. Thus, we expect spillover
effects will be stronger when PEB2 is measured as behavioural intention rather than actual
behaviour. Because policy support is often conceived as an attitude20 rather than a behaviour
(e.g., voting15), we expect spillover to policy support will also be stronger than to behaviour.
H2: Environmental identity
People look to their behaviour to infer their self-concept29 and strive to maintain
consistency across behaviours and beliefs30. Reminding people of past PEBs leads to increased
environmental identity19,31, which in turn relates to increased preference for pro-environmental
products31, pro-environmental intentions32, and policy support19,20. Thus, we expect that
explicitly linking a PEB1 to environmental identity will cause more positive spillover relative to
other types of interventions9,13,16.
H3: Intrinsic motivation
Intrinsic motivation entails engaging in an action because it is inherently important or
interesting33. Empirical evidence suggests that environmental values and intrinsic motivation are
closely linked to PEBs34, and several theories and some evidence suggest that intrinsic
motivation should lead to greater positive spillover9,12,17. We expect that interventions targeting
intrinsic motivation will cause more positive spillover relative to other interventions.
H4: Guilt
Because PEBs are morally significant behaviour36, they may be affected by moral
licensing—when an initial moral behaviour (e.g., PEB1) boosts one’s perceived morality (or
reduces guilt), which then licenses harmful behaviour9,37. Thus, guilt can mediate PEB
spillover19. Guilt-based interventions should encourage an initial PEB, but we predict that
reduced feelings of guilt after engaging in that PEB will undermine motivation for other PEBs,
resulting in greater negative spillover relative to other interventions9.
H5: Incentives
Extrinsic motivation entails engaging in an action to earn or avoid an external reward or
punishment (e.g., financial incentives33). Targeting PEB1 with extrinsic motivators may “crowd
out” intrinsic motivation to engage in other PEBs, leading to negative spillover9,12,33. Prior work
shows that financial incentives sometimes reduce positive spillover17 but sometimes increase
positive spillover38. We predict that incentive-based interventions will be more likely produce no
spillover or negative spillover relative to other interventions.
H6: Difficulty
In line with self-perception theory29 and research on prosocial behaviours39, one is more
likely to attribute a difficult (vs. easy) PEB to an internal identity, leading to future PEBs.
Reminders of difficult and unique past PEBs strengthens environmental identity31 , which is
associated with increased pro-environmental concern and behaviour 19,31,32. Thus, a difficult
PEB1 should lead to more positive spillover than an easy PEB1.
H7: Perceived similarity of behaviours
Lay perceptions of PEBs cluster along a number of dimensions41, and some PEBs are
perceived as more closely related than others42. Positive spillover resulting from self-perception
and identity processes (discussed above) may be more likely when behaviours are conceptually
linked9,13,40. Therefore, we predict that positive spillover will increase as perceived similarity of
PEB1 and PEB2 increases.
H8: Publication bias
Evidence for the “file drawer effect” demonstrates that null or weak effects are less likely
to be published than significant or strong effects23. Thus, published spillover effect sizes should
be larger than unpublished effect sizes.
H9: Spillover to self-reported behaviours will be stronger
People often overestimate the frequency in which they engage in socially-desirable
behaviours43. Thus, spillover effects to self-reported behaviour should be stronger than
objectively-measured behaviour.
This meta-analysis included 77 effect sizes generated from 25 experimental or quasi-
experimental studies that successfully manipulated a first pro-environmental behaviour (PEB1)
and measured support for or adoption of a second pro-environmental behaviour (PEB2). Three
types of PEB2 effect sizes were included, 30 for behavioural intentions (total N = 2,856), 30 for
observed or self-reported behaviour (total N = 20,253), and 17 for policy support (total N =
1,338; see Table 1).
There was significant variability among PEB2 intentions (Q = 90.81, p < .001, I2 =
68.1%) and policy support effect sizes (Q = 27.66, p = .04, I2 = 42.2%) but not PEB2 behaviour
(Q = 31.08, p = .36, I2 = 6.7%). Forest plots (see Figures 1-3) indicate that the overall mean
effect size was significantly different from zero for intentions and behaviour, not policy support.
Specifically, adopting PEB1 had a small, positive effect on PEB2 intentions, a near-zero negative
effect on PEB2 behaviour, and a negligible effect on PEB2 policy support. Egger’s tests assessed
relationships between study effect sizes and the sample size from which the effect was derived.
None were significant (intentions: p = .35; behaviour: p = .87; policy support: p = .68),
suggesting that small samples were not biasing the estimate of the overall effect sizes. Across all
effect sizes, mean statistical post-hoc power was 28% (SD = 27%), suggesting many studies were
severely underpowered. Only 10% of effect sizes were derived from studies with 70% or greater
Confirmatory Hypothesis Testing
Results from all confirmatory analyses are shown in Table 1 and Supplementary Table 2.
We consider both p-values and Cohen’s d effect sizes when evaluating support for
hypotheses44,45, and consider a hypothesis partially supported when either 1) one or two of the
three outcomes (i.e., intentions, behaviour, and policy support) confirms the hypothesis based on
significant p-values, or 2) when effect sizes are consistent with the hypothesis but the p-values
are not significant.
H1. The effect on PEB2 intentions (reported above) was significantly more positive than
behaviour (QB = 36.82, p < .001, τ2 = 0.02) and policy support (QB = 8.52, p = .004, τ2 = 0.01); the
effect on behaviour and policy support did not significantly differ (QB = 0.14, p = .71, τ2 = 0.00).
Therefore, we found partial support for H1: spillover effects for behavioural intentions were
stronger (and in an opposite direction) than effects for behaviour, but effects for policy support
were not stronger than behaviour.
H2. Targeting identity (vs. not targeting identity) led to more positive effect sizes for
PEB2 behaviour and policy, but not PEB intentions, though the differences were not significant
(intentions: QB = 0.19, p = .67, τ2 = 0.00; QB = 0.44, behaviour: p = .51, τ2 = 0.00; policy support:
QB = 0.08, p = .77, τ2 = 0.00). Thus, we found no support for H2.
H3. For all three types of PEB2, targeting intrinsic motivation (vs. not doing so) led to
more positive spillover. However, only the effect for intentions was significant (intentions: QB =
5.23, p = .02, τ2 = 0.01; behaviour: QB = 0.04, p = .85, τ2 = 0.00; policy support: QB = 0.08, p =
.77, τ2 = 0.00). Thus, H3 was partially supported.
H4. For PEB2 intentions, targeting guilt (vs. not targeting guilt) led to more negative
spillover (QB = 11.65, p = .001, τ2 = 0.12). For PEB2 behaviour, no interventions targeted guilt,
so no comparison could be made. For PEB2 policy support, targeting guilt led to slightly more
positive spillover (QB = 2.66, p = .10, τ2 = 0.01). H4 was therefore partially supported: guilt-
based interventions increased the magnitude of negative spillover for intentions, but led to
marginal positive spillover for policy support.
H5. For PEB2 intentions, incentive interventions reduced positive spillover relative to
non-incentive interventions (QB = 2.74, p = .10, τ2 = 0.01). For PEB2 behaviour, incentive
interventions led to less negative spillover compared to non-incentive interventions (QB = 1.61, p
= .21, τ2 = 0.00). No studies using incentive interventions measured PEB2 policy support. Thus,
there was partial support for H5.
H6. No studies tested the effect of a difficult PEB1 on PEB2 behaviour and only one
study examined the effect of an easy PEB1 on policy support. However, a moderately difficult
PEB1 led to less positive spillover on PEB2 intentions compared to an easy PEB1 (QB = 3.51, p
= .06, τ2 = 0.01). Alternatively, an easy PEB1 led to more negative spillover on PEB2 behaviour
compared to a moderately difficult PEB1 (QB = 1.62, p = .20, τ2 = 0.001). Thus, H6 was not
supported: PEB1s that were more difficult reduced positive spillover for intentions and produced
no spillover on behaviour.
H7. For PEB2 intentions, high perceived similarity (vs. low-similarity) between PEB1
and PEB2 led to a more positive effect (QB = 5.88, p = .02, τ2 = 0.09; there were no effects for
medium perceived similarity). For PEB2 behaviour, PEBs that were perceived as highly similar
led to more positive spillover, compared to those at medium (QB = 2.24, p = .14, τ2 = 0.02) or low
perceived similarity levels (QB = 2.25, p = .13, τ2 = 0.02); there was no difference between
medium and low perceived similarity levels (QB = 0.00, p = 1.00, τ2 = 0.00). For PEB2 policy
support, highly similar PEBs led to slight negative spillover overall, but less negative spillover
compared to low-similarity PEBs (QB = 8.03, p = .01, τ2 = 0.03). Overall, H7 was supported,
more perceived similarity between PEB1s and PEB2s led to more positive spillover to intentions
and behaviour and less negative spillover to policy support.
H8. For PEB2 intentions, published effect sizes were the same as effect sizes from theses
(QB = 0.00, p = 1.00, τ2 = 0.00) and larger than effect sizes from unpublished data (QB = 14.08, p
< .001, τ2 = 0.06). For PEB2 behaviour, published effect sizes were larger than those from theses
(QB = 0.47, p = .49, τ2 = 0.00; no unpublished studies were found). For PEB2 policy support,
published effect sizes were smaller than effect sizes from theses (QB = 0.19, p = .66, τ2 = 0.00),
but larger than effect sizes from unpublished data (QB = 0.03, p = .87, τ2 = 0.00). Thus, H8 was
partially supported. Only one effect was significant, but trends suggest stronger spillover effects
in published vs. unpublished data.
H9. H9 was not supported; for PEB2 behaviour, observed behaviour effect sizes were
stronger (though more negative) than self-reported effect sizes (QB = 5.12, p = .02, τ2 = 0.01).
Exploratory analyses
Exploratory analyses tested three additional research questions (Table 2).
RQ1. We tested whether studies that manipulate perceptions of a past PEB1 have
stronger spillover effects compared to studies that manipulate an actual PEB1. For PEB2
intentions, interventions inducing a PEB1 led to more positive spillover compared to
manipulating perceptions (QB = 11.65, p = .001, τ2 = 0.12). For PEB2 behaviour, no studies
manipulated perceptions. For PEB2 policy support, manipulating perceptions of past PEB led to
more positive spillover (QB = 2.02, p = .16, τ2 = 0.01). Thus, there was mixed evidence.
RQ2. We examined whether an easy (rather than difficult) PEB2 lead to positive
spillover. For PEB2 intentions, an easy PEB2 led to more positive spillover than a moderately
difficult PEB2 (QB = 6.38, p = .01, τ2 = 0.06). For PEB2 behaviour spillover, there was no
difference between easy and moderate PEB2s (QB = 0.00, p = 1.00); yet, an easy or moderately
difficult PEB2 led to less negative spillover compared to a highly difficult PEB2 (QB = 0.49, p =
.48, τ2 = 0.00 and QB = 0.51, p = .48, τ2 = 0.00 respectively). For PEB2 policy support, an easy
PEB2 led to more positive spillover compared to a moderately difficult PEB2 (QB = 4.10, p =
.04, τ2 = 0.05). The results were mixed for behaviour, but for intentions and policy support an
easier PEB2 led to more positive spillover.
RQ3. Finally, we tested differences in spillover between types of samples. Most studies
used general population or student samples. Across all three types of PEB2, spillover effects
were identical (i.e., QB = 0.00, intentions, p = 1.00, τ2 = 0.00), more positive (i.e., policy support,
QB = 0.20, p = .65, τ2 = 0.00), or less negative (i.e., behaviour, QB = 2.13, p = .15, τ2 = 0.01) in
general population samples compared to student samples.
Based on a meta-analysis of 77 effect sizes, we draw several conclusions about the
conditions that lead to PEB spillover and highlight areas for future research. First, when
individuals are induced to adopt an initial PEB, intentions to adopt another PEB increase slightly
but actual adoption decreases very slightly. These effects are best understood in the common
language of effect size derived from Cohen’s d 46,47. A Cohen’s d of 0.17 for intentions and -0.03
for behaviour mean that if 100 people were successfully encouraged to adopt a pro-
environmental action via an intervention, approximately five would develop stronger intentions
to adopt a second pro-environmental action, and fewer than one would reduce engagement in a
second action because of their experience performing PEB1. In contrast, we find no evidence of
spillover from pro-environmental behaviour to policy support.
Although spillover effects are small, this does not mean that spillover is unimportant or
that practitioners should ignore downstream effects of environmental interventions. The
importance of these effect sizes should be evaluated within the context of a particular program or
policy decision. However, these data refute the concern that motivating voluntary pro-
environmental actions will undermine public support for more robust policies48, or the prospect
that “simple and painless” behaviours serve as a gateway to higher-impact actions49.
Perhaps most importantly, the direction and magnitude of spillover effects vary as a
function of the types of interventions used and behaviours targeted. These results suggest there
may be ways to trigger larger positive spillover effects or mitigate negative spillover effects.
Those hoping to leverage or avoid spillover effects need to be aware of the complex ways in
which these effects operate. Despite these complexities, the current research presents several
clear lessons and highlights gaps in the existing spillover literature.
Factors Affecting Spillover
Across all measures of PEB—behavioural intentions, actual behaviour, or policy
support—interventions that targeted intrinsic motivations resulted in more positive spillover,
though it was only statistically significant for intentions. Alternatively, interventions that
increased guilt induced negative spillover for intentions, and incentives tended to limit positive
(and negative) spillover. Program managers and policy designers should therefore consider the
consequences of these approaches on secondary intentions and behaviours, and may consider
also targeting intrinsic motivation to maximize the chances of positive spillover. Similarly, there
is mixed evidence that difficult PEB1s will lead to more positive spillover, suggesting that even
relatively easy behaviours may be efficient targets for PEB1 interventions. Positive spillover is
also more likely between similar behaviours. Thus, practitioners should focus interventions on
clusters of PEBs that are perceived by their audience as similar in the type of behaviour (e.g.,
types of energy behaviours) or educate audiences about behaviours that are connected to a
common goal behaviour (e.g., climate-friendly behaviours).
Finally, spillover effects varied widely based on how PEB2 is measured. In general,
spillover to behavioural intentions was positive and stronger than spillover to actual behaviour
and policy support, which was negative and weaker. Furthermore, no overall spillover effects
emerged when PEB2 was measured as policy support. Measures of intentions are often used as a
proxy for actual behaviour, which is both difficult and expensive to assess. However, this
analysis suggests that these two methodological approaches purporting to measure the same
phenomenon lead to two very different conclusions. We caution researchers against relying on
self-reported intentions to draw conclusions about behaviour.
Limitations and Areas for Future Research
A number of limitations should be considered when interpreting these data and designing
future studies. First, most of the studies included were underpowered; future research should aim
for sample sizes based on expected effect sizes of second-order effects on behaviour. The small
average effect sizes revealed here (Cohen’s d: -0.03 to 0.17), suggest that large sample sizes are
needed to achieve acceptable levels of statistical power. This also means that highly-powered
studies were strongly weighted when computing average effect sizes across studies; this was
particularly true for PEB2 behaviour, as four effect sizes accounted for 80% of the weighted
effect size. We occasionally found larger effect sizes, such as the effect of targeting identity on
PEB2 intentions. Greater focus on theory and measurement of mediators, such as identity, will
help determine why larger effects occur, and will make achieving adequate levels of statistical
power more feasible.
Additionally, researchers vary substantially in their measurement and testing of PEB
spillover. For example, some define spillover as a change in PEB2 based on some experimental
intervention on PEB1 (regardless of whether participants successfully performed PEB1),
whereas others look at correlations between PEB1 and PEB2, or even correlations between
change in PEB1 and change in PEB2. To progress as a field, researchers must agree on what
constitutes spillover. We have taken the position that spillover effects can only be measured
when there is an experimental or quasi-experimental intervention in which participants are
successfully induced to take on PEB1, to increase our confidence in causal effects from PEB1 to
Scholars also vary in their conceptualization of moderators such as behavioural difficulty,
perceived behavioural similarity, or even whether an intervention targets a process such as
identity or intrinsic motivation. We used best practice to code studies on these dimensions,
including use of two independent coders, computing inter-rater agreement, and resolving
disagreements. However, the sooner the field can reach consensus on how to characterize
behaviours and interventions, the more reliable future tests of moderation will become.
Likewise, scholars should test spillover effects between contexts and across longer time
horizons13,50. Most of the studies included here measured PEB2 immediately after PEB1 and in
the same context, limiting what the current meta-analysis can tell us about spillover. For
example, researchers should explore whether interventions targeting intrinsic motivation also
lead to positive spillover between behaviours in different settings and over time. If the answer is
yes, small effects may accrue to create meaningful amounts of behaviour change across
numerous downstream outcomes. Relatedly, scholars should also explore how the net effects of
interventions that are designed to produce positive spillover compare to those that target a single
Finally, published studies and theses tended to find more positive and stronger effects
than unpublished data sets, suggesting that the spillover literature is not immune to the file
drawer problem. This is a major problem in any area of research, but especially in a field that has
direct implications for policymaking—practitioners need to know if spillover effects are likely to
be weak or non-existent. Therefore, we urge spillover researchers to pursue publication of null
results and for reviewers and editors to evaluate publications on the merits of the research design
rather than significant results. Preregistering methodologies and hypotheses can improve the
integrity of the research process and create avenues for publishing null results.
Better understanding pro-environmental spillover would both enrich theoretical models
of spillover and help with the design of more effective behaviour change interventions that can
influence a wide range of behaviours, which in turn would play a key role in reducing GHG
emissions and thereby mitigating climate change.
Search Strategies
We conducted two searches for articles: an initial search in June 2016 and a follow-up
search to collect more recent publications in March 2017. Our searches used the following social
scientific article databases: Business Source, Digital Dissertations, Ecology Abstracts, Econlit,
GeoRef, GreenFile, PsychINFO, Scopus, and Web of Knowledge. We examined the abstract and
title of articles for search terms related to environmental behaviour and behaviour spillover. We
also reviewed reference lists of articles identified through the search as well as the reference list
in Blanken et al.’s meta-analysis on moral licensing37. We put out a call for papers on the Society
for Personality and Social Psychology (SPSP) forum board, the Facebook pages for the
American Psychological Association Division 34 (Society for Environmental, Population, and
Conservation Psychology) and the Sustainability Psychology Society for Personality and Social
Psychology group, and the Division 34 and the Association for Environmental Studies and
Sciences (AESS) listservs, all in August of 2016. We also directly contacted researchers who had
previously published multiple papers on the topic of environmental behaviour spillover. Authors
were also contacted if their study seemed promising but lacked key details necessary to evaluate
the study according to the inclusion criteria.
Pre-search decisions and inclusion criteria
We included studies that assessed a wide range of PEBs including: recycling, energy
conservation, water conservation, “green” purchases (e.g., energy efficient technologies, home
energy audits or retrofits), sustainable dietary/food decisions (e.g., eating vegetarian, purchasing
organic food), public transportation use, carpooling, environmental activism, donations to
environmental causes, and environmental policy support. This list was based, in part, on previous
lists used in other meta-analyses and reviews of pro-environmental behaviour51. Behaviours such
as wildlife conservation were excluded because few studies by environmental social scientists
have focused on changing these types of behaviours in experimental contexts while considering
spillover. The complete list of search terms can be found in the Supplementary Information.
The six inclusion requirements were: (1) the study investigated one of the focal
environmental behaviours, intention, or policy support, (2) the study contained an intervention
condition and at least one comparison condition (i.e., experimental or quasi-experimental design)
(3) the study included at least one primary and one secondary PEB, (4) the study demonstrated
that at least one of the manipulations changed PEB1 compared to a control or comparison
condition, (5) the study included a measure of PEB2, and (6) the study reported statistical results
that allowed for the calculation of an effect size for PEB2 (or the necessary information was
provided by authors when requested).
Search results
In total, 14,981 articles were found in the initial database search and 67 other articles
were identified through other sources (e.g., reference lists of the articles or through listservs).
After the fifth author and a research assistant each independently reviewed half of the titles and
abstracts based on the inclusion criteria, they removed articles from the database that did not
qualify. At this point, 95 relevant articles remained. The first and fifth author analysed these 95
articles according to our six inclusion criteria based on a review of the full article. Figure 4 is the
PRISMA figure52 that describes how articles were selected and filtered through different phases
of the search process, including reasons for excluding articles during the in-depth review stage.
In the end, 25 articles met our inclusion criteria, and from those articles we were able to derive k
= 77 effect sizes (see Supplementary Table 1).
Additional coded study design features
Several additional features were coded that could influence the strength and direction of
environmental behaviour spillover effects. Of particular relevance here (for a complete list of all
coded variables, please view the coding framework on the project’s Open Science Framework
page:, we first coded for ease of PEB1 and PEB2, as well as their
similarity, both of which were subjectively categorized by the coders. Second, we coded for
characteristics of the intervention, particularly whether the interventions targeted environmental
identity, intrinsic motivation, guilt, or used financial incentives. Third, we coded for
characteristics of the study, intervention, or measures including: (1) whether PEB1 was an
intention, behaviour, or type of policy support, (2) whether PEB2 was an intention, behaviour, or
type of policy support, (3) whether PEB2, if it was a behaviour, was a self-reported or objective
behaviour (most analyses collapsed across these two types of outcomes, unless noted), and (4)
the type of sample (e.g., general population or college students). For interventions particularly, if
it was unclear whether a study was trying to influence a potential spillover process (i.e., identity,
intrinsic motivation, guilt), then we coded the study as being “unclear” in regards to that study
feature. Analyses of studies coded as “unclear” for a given study feature (e.g., whether the study
manipulated identity) are excluded from the present paper. To see how many studies were
excluded from the confirmatory and exploratory analyses, see Supplementary Tables 5 and 6 in
the Supplementary Information.
Reliability of coding
The coding framework was agreed upon prior to the coding process. For unpublished data
sets, either the first author or the researcher who collected that data set solely coded the data.
When the researcher associated with a data set coded that data set themselves for the meta-
analysis, the first author worked with the researcher to correctly code the data. For published
articles, the first author and a research assistant independently coded each study and the coders
discussed their codes and resolved all disagreements through conversations. Kappa coefficients
for categorical variables indicated a moderate level of agreement (median = 0.62, mean = 0.56),
and intra-class correlations for continuous variables indicated a high level of agreement (median
= 0.99, mean = 0.93).
When coding studies, part of the reason for the lower kappa coefficients were because of
the difficulty of coding for behaviour and intervention characteristics. We knew that it would be
difficult to code studies according to whether the interventions may have influenced identity,
intrinsic motivation, and guilt, as well as ease and similarity of PEBs, and thus we set those
variables aside to focus on them more deliberately after initial coding of articles was finished.
Before the two independent coders coded studies on behavioural (e.g., behavioural
difficulty and behavioural similarity) and intervention dimensions (e.g., whether the
interventions targeted identity, intrinsic motivation, guilt, or used incentives), the entire research
team agreed on coding guidelines. When possible, we relied on previous research relevant to the
behavioural dimension. For example, previous research has occasionally considered how pro-
environmental behaviours vary according to difficulty41,53, and we used these findings to guide
our classifications. For behavioural similarity, we considered “classes” of behaviours (e.g.,
energy behaviours, recycling behaviours) to be high in similarity, behaviours with somewhat
similar actions (e.g., energy and water conservation behaviours) as moderately similar, and
behaviours with quite different actions (e.g., recycling and taking the bus) as low in similarity.
For interventions, we classified interventions as clearly targeting a process if the authors stated
so in the paper or the intervention could be easily argued to have affected the process. Studies
were coded as unclear if, for example, coders were uncertain whether the intervention would
have affected a specific process (e.g., guilt). Ultimately, all codes were agreed upon by both
coders for all items, including behaviour and intervention characteristics.
Meta-analytic strategy
We calculated quantitative effect sizes using Cohen’s d as our effect size metric. If a
study had multiple conditions, we computed an effect size for any combination of conditions that
led to a significant change in PEB1. For example, if there was a message condition, an incentive
condition, and a control condition, and only the message condition significantly changed PEB1
compared to the control condition, then we only computed an effect size for the effect of the
message on PEB2 compared to the control condition. When multiple conditions had a significant
effect on PEB1 relative to a control or comparison condition, we reused samples and thus had
correlated effect sizes. Similarly, if articles had multiple PEB2s that were distinct, and for which
effect sizes could be computed independently (e.g., an effect size for recycling and an effect size
for energy conservation), we computed separate effect sizes that would also be expected to be
correlated. We initially conducted multilevel meta-analyses to take these dependencies into
account, but the differences between the multilevel and the single-level meta-analyses were
minimal, so we report only the single-level meta-analytic results for the sake of parsimony.
A few studies measured a change in PEB2 at multiple time points; however, because this
was rare we only consider the first, and often only, measure of PEB2. Some studies did not
report the effect of the intervention on PEB2, but instead looked at either the correlation between
PEB1 and PEB2 within the context of an experimental design, or the correlation between a
change in PEB1 and a change in PEB2. These designs were rare, and they were excluded from
the analyses. We briefly revisit these approaches in the discussion.
Consistent with previous meta-analyses in this area54-56, we examined comparisons
between studies (e.g., those targeting vs. not targeting intrinsic motivation) only if two or more
studies qualified for each comparison group. Differences between effect sizes across moderator
categories were tested (e.g., type of intervention) using the QB statistic and the corresponding p-
value, along with the τ2 for the amount of residual between-studies variance. The meta-analysis
was conducted using STATA Version 1557. Weighted average effect sizes were calculated based
on the mixed effects model, including for the tests of moderation. Cohen’s58 guidelines were
used for the interpretation of effect sizes; d = .20 is considered a “small” effect size, d = .50 is a
“medium” effect size, and d = .80 is a “large” effect size. Finally, all statistical tests were two-
tailed. For respective funnel plots, see Supplementary Figures 1-3.
Preregistration and data availability
The search strategy, coding framework, and hypotheses were all pre-registered at the
following site prior to data analysis: The data for all of the coding of the
articles, including variables not explored in the present paper, can be found publicly at the
following site:
1. Dietz, T., Gardner, G. T., Gilligan, J., Stern, P. C. & Vandenbergh, M. P. Household
actions can provide a behavioral wedge to rapidly reduce US carbon emissions. Proc.
Natl. Acad. Sci. USA 106, 18452-18456 (2009).
2. Vandenbergh, M. P., Carrico, A. R. & Bressman, L. S. Regulation in the behavioral era.
Minn. Law R. 95, 715-781 (2011).
3. Thaler, R. H. & Sunstein, C. R.. Nudge: Improving decisions about health, wealth, and
happiness (Penguin Books, 2009).
4. Abrahamse, W., Steg, L., Vlek, C. & Rothengatter, T. A review of intervention studies
aimed at household energy conservation. J. Environ. Psychol. 25, 273-291 (2005).
5. Carrico, A. R. & Riemer, M. Motivating energy conversation in the workplace: An
evaluation of the use of group-level feedback and peer education. J. Environ. Psychol. 31,
1-13 (2011).
6. Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. The
constructive, destructive, and reconstructive power of social norms. Psychol. Sci. 18, 429-
434 (2007).
7. Janssen, W. Seat-belt wearing and driving behavior: An instrumented-vehicle study.
Accid. Anal. Prev. 26, 249-251 (1994).
8. Thøgersen, J. Spillover processes in the development of a sustainable consumption
pattern. J. Econ. Psychol. 20, 53-81 (1999).
9. Truelove, H. B., Carrico, A. R., Weber, E. U., Raimi, K. T. & Vandenbergh, M. P.
Positive and negative spillover of pro-environmental behavior: An integrated review and
theoretical framework. Global Environ. Chang. 29, 127-138 (2014).
10. Gillingham, K., Kotchen, M. J., Rapson, D. S. & Wagner, G. Energy policy: The rebound
effect is overplayed. Nat. 493, 475-476 (2013).
11. Lauren, N., Fielding, K. S., Smith, L. & Louis, W. R. You did, so you can and you will:
Self-efficacy as a mediator of spillover from easy to more difficult pro-environmental
behavior. J. Environ. Psychol. 48, 191-199 (2016).
12. Nash, N., Whitmarsh, L., Capstick, S., Hargreaves, T., Poortinga, W., Thomas,
G…Xenias, D. Climate-relevant behavioral spillover and the potential contributions of
social practice theory. WIREs Clim. Chang. 8, (2017).
13. Nilsson, A., Bergquist, M. & Schultz, W. P. Spillover effects in environmental behaviors,
across time and context: A review and research agenda. Environ. Education R. 23, 573-
589 (2017).
14. Sintov, N. D., Geislar, S. & White, L. Cognitive accessibility as a new factor in
proenvironmental spillover: Results from a field study of household food waste
management. Environ Behav. (in press).
15. Carrico, A. R., Raimi, K. T., Truelove, H. B., & Eby, B. Putting your money where your
mouth is: An experimental test of pro-environmental spillover from reducing meat
consumption to monetary donations. Environ Behav. 50, 723-748 (2018).
16. Baca-Motes, K., Brown, A., Gneezy, A., Keenan, K. A. & Nelson, L. D. Commitment and
behavior change: Evidence from the field. J. Consum. Res. 39, 1070-1084 (2013).
17. Steinhorst, J., Klöckner, C. A. & Matthies, E. Saving electricity—For the money or the
environment? Risks of limiting pro-environmental spillover when using monetary
framing. J. Environ. Psychol. 43, 125-135 (2015).
18. Thomas, G. O., Poortinga, W. & Sautkina, E. The Welsh single-use carrier bag charge and
behavioural spillover. J. Environ. Psychol. 47, 126-135 (2016).
19. Lacasse, K. Don’t be satisfied, identify! Strengthening positive spillover by connecting
pro-environmental behaviors to an “environmentalist” label. J. Environ. Psychol. 48, 149-
158 (2016).
20. Truelove, H. B., Yeung, K. L., Carrico, A. R., Gillis, A. J. & Raimi, K. T. From plastic
bottle recycling to policy support: An experimental test of pro-environmental spillover. J.
Environ. Psychol. 46, 55-66 (2016).
21. Ajzen, I. The theory of planned behavior. Organ. Behav. Hum. Dec. 50, 179-211 (1991).
22. Ferguson, C. J. & Heene, M. A vast graveyard of undead theories: Publication bias and
psychological science’s aversion to the null. Perspect. Psychol. Sci. 7, 555-561 (2012).
23. Franco, A., Malhotra, N. & Simonovits, G. Publication bias in the social sciences:
Unlocking the file drawer. Sci. 345, 1502-1504 (2014).
24. Kühberger, A., Fritz, A. & Scherndl, T. Publication bias in psychology: a diagnosis based
on the correlation between effect size and sample size. PLOS One 9, e105825 (2014).
25. Cohn, L. D. & Becker, B. J. How meta-analysis increases statistical power. Psychol.
Methods 8, 243-253 (2003).
26. Hedges, L. V. Estimation of effect size from a series of independent experiments. Psychol.
Bull. 92, 490-499 (1982).
27. Armitage, C. J., & Conner, M. Efficacy of the Theory of Planned Behaviour: A meta-
analytic review. Br. J. Soc. Psychol. 40, 471-499 (2001).
28. Sheeran, P., Maki, A., Montanaro, E., Avishai-Yitshak, A., Bryan. A., Klein, W. M.
P…Rothman, A. J. The impact of changing attitudes, norms, and self-efficacy on health-
related intentions and behavior: A meta-analysis. Health Psychol. 35, 1178-1188 (2016).
29. Bem, D. J. Self-perception: An alternative interpretation of cognitive dissonance
phenomena. Psycholog. R. 74, 183-200 (1967).
30. Festinger, L., & Carlsmith, J. M. Cognitive consequences of forced compliance. J.
Abnorm. Soc. Psychol. 58, 203-210 (1959).
31. Van der Werff, E., Steg, L., & Keizer, K. Follow the signal: When past pro-environmental
actions signal who you are. J. Environ. Psychol. 40, 273-282 (2014a).
32. Van der Werff, E., Steg, L., & Keizer, K. I am what I am, by looking past the present: The
influence of biospheric values and past behavior on environmental self-identity. Environ.
Behav. 46, 626-657 (2014b).
33. Ryan, M. R., & Deci, E. L. Intrinsic and extrinsic motivations: Classic definitions and new
directions. Contemp. Educ. Psychol. 25, 54-67 (2000).
34. Bamberg, S. & Möser, G. Twenty years after Hines, Hungerford, and Tomera: A new
meta-analysis of psych-social determinants of pro-environmental behaviour. J. Environ.
Psychol. 27, 14-25 (2007).
35. Steg, L., Dreijerink, L. & Abrahamse, W. Factors influencing the acceptability of energy
policies: A test of VBN theory. J. Environ. Psychol. 25, 415-425 (2005).
36. Stern, P. C. New environmental theories: Toward a coherent theory of environmentally
significant behavior. J. Soc. Issues 56, 407-424 (2000).
37. Blanken, I., van de Ven, N. & Zeelenberg M. A meta-analytic review of moral licensing.
Pers. Soc. Psychol. Bull. 41, 540-558 (2015).
38. Lanzini, P. & Thøgersen, J. Behavioural spillover in the environmental domain: An
intervention study. J. Environ. Psychol. 40, 381-390 (2014).
39. Gneezy, A., Imas, A., Brown, A., Nelson, L. D. & Norton, M. I. Paying to be nice:
Consistency and costly prosocial behavior. Manage. Sci. 58, 179-187 (2012).
40. Thøgersen, J. A cognitive dissonance interpretation of consistencies and inconsistencies in
environmentally responsible behavior. J. Environ. Psychol. 24, 93-103 (2004).
41. Truelove, H. B. & Gillis, A. J. Perception of pro-environmental behavior. Global Environ.
Chang. 49, 175-185 (2018).
42. Margetts, E. A. & Kashima, Y. Spillover between pro-environmental behaviours: The role
of resources and perceived similarity. J. Environ. Psychol. 49, 30-42 (2017).
43. Affuso, O., Stevens, J., Catellier, D., McMurray, R. G., Ward, D. S., Lytle, L…Young, D.
R. Validity of self-reported leisure-time sedentary behavior in adolescents. J. Negat.
Results Biomed. 10, (2011).
44. Cumming, G. The new statistics: Why and how. Psychol. Sci. 25, 7-29 (2014).
45. Goh, J. X., Hall, J. A., & Rosenthal, R. Mini meta-analysis of your own studies: Some
arguments on why and a primer on how. Soc. Pers. Psychol. Compass 10, 535-549 (2016).
46. Magnusson, K. Interpreting Cohen’s d effect size: An interactive visualization. (2014).
47. Ruscio, J. A probability-based measure of effect size: Robustness to base rates and other
factors. Psychol. Method. 13, 19-30 (2008).
48. Maniates, M. F. Individualization: Plant a tree, buy a bike, save the world? Global
Environ. Polit. 1, 31-52 (2001).
49. DEFRA. A framework for pro-environmental behaviors (2008).
50. Maki, A. & Rothman, A. J. Understanding proenvironmental intentions and behaviors:
The importance of considering both the behavior setting and type of behavior. J. Soc.
Psychol. 157, 517-531 (2017).
51. Osbaldiston, R. & Schott, J. Environmental sustainability and behavioral science: meta-
analysis of pro-environmental behavior experiments. Environ. Behav. 44, 257-299 (2012).
52. Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & the PRISMA Group. Preferred
reporting items for systematic reviews and meta-analyses: The PRISMA Statement. PLOS
Med. 6, e100097 (2009).
53. Kaiser, F. G. A general measure of ecological behavior. J. Appl. Soc. Psychol. 28, 395-
422 (1998).
53. Abrahamse, W. & Steg, L. Social influence approaches to encourage resource
conservation: A meta-analysis. Global Environ. Chang. 23, 1773-1785 (2013).
54. Karlin, B., Zinger, J. F. & Ford, R. The effects of feedback on energy conservation: A
meta-analysis. Psychol. Bull. 141, 1205-1227 (2015).
55. Lokhorst, A. M., Werner, C., Staats, H., van Dijk, E. & Gale, J. L. Commitment and
behavior change: A meta-analysis and critical review of commitment-making strategies in
environmental research. Environ. Behav. 45, 3-34 (2013).
56. StataCorp. Stata Statistical Software: Release 15. (StataCorp LP, 2017).
57. Cohen, J. A power primer. Psychol. Bull. 112, 155-159 (1992).
Table 1
Confirmatory Results
PEB2 category d+
1. PEB2 spillover effects
will be stronger for
intentions and policy support
compared to behaviour.
2. Interventions targeting
environmental identity will
lead to positive spillover.
Identity Intervention
Other Intervention
3. Interventions targeting
intrinsic motivation will lead
to positive spillover.
Intrinsic Intervention
Other Intervention
4. Interventions targeting
guilt will lead to negative
Guilt Intervention
Other Intervention
5. Incentive interventions
will lead to negative
Incentive Intervention
Other Intervention
6. Difficult PEB1s will lead
to positive spillover.
High Difficulty
Moderate Difficulty
Low Difficulty
7. Similar PEB1s and
PEB2s will lead to positive
High Similarity
Moderate Similarity
Low Similarity
8. Published effect sizes
will be larger.
9. Spillover effects will be
stronger when PEB2
behaviour is self-reported,
rather than objective.
Note. Negative values mean negative spillover occurred, * means effect is different from 0 at
p < .05, ** means effect is different from 0 at p < .01. Confidence intervals, R2, and k can be
found in Supplementary Table 2.
Table 2
Exploratory Results
Note. Negative values mean negative spillover occurred, CI = confidence interval, * means effect
is different from 0 at p < .05, ** means effect is different from 0 at p < .01, R2 is only listed in the
first value of a given moderator.
d+ (95% CI)
Do interventions
targeting people’s
perceptions of their past
PEB1 have stronger
spillover effects
compared to studies
targeting another PEB1?
of PEB1
-0.29 (-0.58 to 0.00)
Policy Support
0.08 (-0.14 to 0.29)
0.22** (0.06 to 0.30)
Policy Support
-0.08 (-0.29 to 0.13)
Does an easy PEB2,
rather than a difficult
PEB2, lead to positive
-0.03 (-0.09 to 0.04)
Policy Support
0.10 (-0.07 to 0.27)
-0.05 (-0.12 to 0.03)
Policy Support
-0.31 (-0.89 to 0.28)
0.47 (-0.11 to 1.05)
-0.05* (-0.10 to -0.004)
Policy Support
0.04 (-0.08 to 0.16)
Are there differences in
spillover between types
of samples?
0.17** (0.003 to 0.33)
-0.03* -0.05 to -0.01)
Policy Support
0.01 (-0.14 to 0.16)
0.17 (-0.03 to 0.36)
-0.25 (-0.76 to 0.26)
Policy Support
-0.05 (-0.45 to 0.34)
Figure 1
Forest plot of PEB2 intention effect sizes
Note. Study numbers can be found in Supplementary Table 1, PEB2 = pro-environmental
behaviour 2, ES = effect size, CI = confidence interval.
Figure 2
Forest plot of PEB2 behaviour effect sizes
Note. Study numbers can be found in Supplementary Table 1, PEB2 = pro-environmental
behaviour 2, ES = effect size, CI = confidence interval.
Figure 3
Forest plot of PEB2 policy support effect sizes
Note. Study numbers can be found in Supplementary Table 1, PEB2 = pro-environmental
behaviour 2, ES = effect size, CI = confidence interval.
Figure 4
Flow of information through the phases of the present review
Records identified through
database searches
(n = 14,981)
Additional records identified
through other sources
(n = 67)
Studies included in
(n = 25; effect sizes =
Full-text manuscripts
excluded, with reasons
(n = 70)
Duplicate data (n = 3)
Not an experimental/
study with 2+
conditions (n = 38)
Less than two PEBs (n
= 18)
No reported/significant
change in PEB1 (n = 6)
No usable statistics for
PEB2 (n = 5)
Full-text manuscripts
assessed for eligibility
(n = 95)
Records excluded
(n = 14,953)
Records screened
(n = 15,048)
... The spillover effect is the mechanism through which an initial pro-environmental behavior (hereafter PEB) triggers other PEBs (Nilsson et al., 2016;Verfuerth & Gregory-Smith, 2018). However, this literature has generated mixed results, revealing that an initial PEB can promote either other PEBs or pro-environmental inactions and anti-environmental behaviors (hereafter AEBs; Nilsson et al., 2016; for a recent meta-analysis, see Maki et al., 2019). Understanding the impact and the direction of the spillover effect is fundamental to develop efficient environmental programs; thus, further research is needed to understand the circumstances under which an initial PEB triggers a virtuous circle. ...
... Thus, a positive spillover effect among different PEBs is possible, with the first PEB affecting subsequent behaviors, behavioral intentions, and policy support. However, a recent meta-analysis (Maki et al., 2019) revealed that the positive spillover effect is stronger for behavioral intentions than actual behavior. Moreover, the virtuous effect is more likely when the previous and the subsequent PEB are perceived as highly similar. ...
... Regarding a specific within-domain spillover effect, in line with the results on behaviors' similarity (Maki et al., 2019), we hypothesized that recalling a past water-related behavior would prompt a positive spillover effect (Hypothesis 1). Thus, we expected that participants who recalled a PEB (i.e., water-saving vs. water-wasting) would express more willingness to engage in future water-related PEBs (Study 1) and reduce their subsequent water consumption (Study 2). ...
Full-text available
Research literature about the environmental spillover effect produced mixed results, revealing that an initial pro-environmental behavior (PEB) is likely to promote either other PEBs (i.e., positive spillover) or pro-environmental inactions and harming behaviors (i.e., negative spillover). Such inconsistency suggests a possible crucial role of moderating variables. In two experimental studies (N Study 1 = 141, N Study 2 = 124), we investigated whether the recall of past environmental behavior (water-saving vs. water-wasting) affects future intention to perform PEBs (Study 1) and actual PEBs (Study 2), depending on participants’ cognitive mindset (manipulated in Study 1 and measured in Study 2). Results showed that the cognitive mindset is a significant moderator of spillover effects. Compared to a holistic one, an analytical mindset is more likely to result in a greater willingness to engage in future PEBs (Study 1) and actual PEB (Study 2) when past PEB is salient. The main contributions of the studies, limitations and possible future research directions are discussed.
... For example, the inclusion of airbags in cars can lead to riskier driving behavior [24], and parents have indicated that they would allow children provided with safety gear to engage in riskier play [25]. In the environmental domain, research has found that after taking an initial pro-environmental action, individuals are sometimes less likely to engage in additional proenvironmental behavior [26][27][28]. Looking to climate change, some experimental research has found that discussing positive developments in reducing carbon dioxide emissions can lead to lower support for efforts to mitigate emissions [16]. ...
Geoengineering technologies are increasingly proposed as a necessary strategy to address climate change. How does the prospect of these technologies affect public support for traditional mitigation strategies that involve emission reductions? Recent research yields mixed results. Some evidence points to a moral hazard or risk compensation effect (i.e., less support for climate change mitigation). Other research finds evidence of heightened risk salience (i.e., more support for climate change mitigation), and still other work finds no effect at all. Focusing on carbon dioxide removal (CDR) technologies, we investigate whether these inconsistent findings result from differences in how the technologies are described against the backdrop of climate change. Specifically, we examine whether including explicit information (or not) about the impacts of climate change alters how information about CDR is received. Building on previous work, we designed two experiments to elucidate the circumstances under which information about CDR elicits a risk compensatory, risk salience, or null response among the public. Results predominantly align with a null effect from CDR information, although mixed results in Study 1 suggest the possibility of a very small risk compensation effect, depending on whether political ideology is controlled for. There was no evidence of a risk salience effect. However, information about the impacts of climate change increased perceived threat and, indirectly, policy support. We, thus, recommend that future studies of risk compensation and risk salience effects account for the provision of climate impacts information and political ideology.
... Residuals of purchasing intentions and activism intentions were moderately correlated in the model (r = 0.39); thus, they shared variance that was not explained by the predictors. As spillover refers to the activation of an intention by another intention (Maki et al., 2019), a willingness to buy less plastic might lead to a willingness to engage in this field (and the other way around), independent of other predictors. As activism intentions and policy support intentions shared a strong common source of variance over and above the predictors (r = 0.60), a spillover effect of these two intentions was also likely. ...
Full-text available
In the last few years, plastic has become an issue of current interest as tremendous ecological effects from plastic littering have become visible. Taking the role of consumers into account, activities comprising purchasing decisions and political engagement are expected to help prevent plastic pollution. The goal of this study was to examine antecedents of three potential plastic reduction activities: purchasing, activism, and policy support. Based on well-established psychological models of pro-environmental behaviour (i.e. theory of planned behaviour, norm activation model), an online survey (N = 648) was administered and analysed via structural equation modelling. Results revealed that personal norms were a relevant predictor of all three intentions. Whereas sufficiency orientation and collective efficacy predicted only activism intention and policy support intention, perceived behavioural control was the strongest predictor of purchasing intentions. Regarding behaviour, people with high activism intentions and sufficiency orientation were more likely to choose a plastic-free incentive instead of the conventional shopping voucher. This study highlights psychological antecedents of plastic reduction. An integrated model showed that rational cost–benefit considerations as well as morality serve as drivers of reducing plastic consumption. Implications for the promotion of plastic-free consumption are discussed.
Full-text available
To reduce food waste, many behavioural intervention experiments have been conducted worldwide, but their effectiveness remains unclear. To assess their impacts, we present a meta-analysis based on 58 studies, selected after screening 1143 papers, which were conducted between 2011 and 2021 covering 26 533 participants. We confirm that behavioural interventions have a moderate effect (z = 0.22) on food waste reduction, with education programs having the most significant impact and informational feedback having the least. We also show that interventions in elementary and middle school settings marginally improve the overall effect size (P < 0.1), and controlled experiments exhibit a higher effect size compared to pre-post experiments in education interventions (P < 0.05). Finally, we present a roadmap to guide future research in the next decade to further improve our understanding on the effects of behavioural interventions to reduce food waste.
Full-text available
The transition towards sustainable consumption and production requires public engagement and support. In this context, understanding the determinants of individual pro-environmental behavior can assist in sustainability policy design, and contribute to explaining cross-country and regional differences in its implementation and effectiveness. This paper examines the influence of local waste management culture on individual recycling behavior. To isolate the impact of location-specific norms, habits and traditions comprising waste management culture from the confounding effect of contemporaneous local economic and social conditions, we use data from over 40,000 domestic immigrants in Greece. Estimating models relating individual recycling activity in the region of current residence to recycling practices in the region of origin, we find robust evidence that region of origin waste management practices have quantitatively and statistically significant influence on individual recycling behavior: a 10 percentage point increase in the prevalence of recycling in the region of origin, increases the probability a subject recycles by 0.9 percentage points. The results suggest that locally prevailing waste management norms and practices influence individual recycling behavior independently of local economic, social and environmental circumstances. Designing effective sustainability policy may need to account for regional variation in norms and preferences, and encourage investment in the development of sustainable waste management culture.
New technological solutions can encourage lower household waste production and higher levels of waste separation. This paper focuses on analyzing the role of different behavioral factors, such as empowerment and pro-environmental behavior (PEB), have on citizens' intention to use a novel household waste management and separation system and how these interact with the financial incentives typically applied in this area, pay-as-you-throw (PAYT) and save-as-you-throw (SAYT). The proposed model was tested in Portugal using the structural equation modeling approach. Survey data from 400 respondents found that empowerment plays a vital role in adopting an innovative waste management system. The research discerns pro-environmental behavior (PEB) both as an antecedent and a moderator between system use and empowerment, system use and behavioral intention, and also between system use and financial incentives. We discovered that for people with low perceived PEB, PAYT actually reduces the use of the new waste management system, while SAYT can increase the use of the system. Furthermore, increasing the empowerment of users in the system can work exceptionally well at encouraging consumers that already have a high level of PEB. The paper concludes with a discussion section about the developed framework's application and implication in the waste management sector. This study is valuable for understanding how citizens will adopt a new waste management system and essential for encouraging citizens to engage in recycling behavior regularly.
Purpose: This study proposes a new agenda for research and practice on pro-environmental behaviours in organisational settings by exploring the intersection between technology innovations and pro-environmental initiatives. The goal is to demonstrate the utility of digital technology in promoting and achieving sustainability by addressing the complexity and inconsistency in pro-environmental behaviours. Design/methodology/approach: Using relevant literature on pro-environmental behaviours, this study explores the possibility of embedding technology innovations in pro-environmental initiatives to promote and enhance sustainability in organisational settings. Findings: This study argues that the recent technological advancement and open innovation provide new insights into understanding and implementing pro-environmental initiatives in organisational settings. While pro-environmental behaviours studies have advanced over the past decades, this study shows that many pro-environmental activities do not require employees to change behaviour. According to this study, psychology and technology innovations offer various opportunities for businesses to effectively and pragmatically embed sustainability into their operations without necessarily changing employees' behaviour. Research limitations/implications: This conceptual study offers opportunities to empirically explore the collaborative nexus between “psychology-based pro-environmental behaviour research and technology innovation”. Despite the plethora of studies on pro-environmental behaviours, results are mixed and inconclusive, raising questions about the dominant practice used for promoting pro-environmental initiatives and behaviours at the corporate level. This study, therefore, provides a new pathway for businesses to address their environmental aspects, demonstrating a pragmatic approach to resolving the complexity of pro-environmental behaviours. Originality/value: This study allows social investigators, policymakers, and technology developers to re-assess, revive and further investigate how they can collaborate to address practical environmental and social issues.
This study investigates the effect of two interventions aimed at reducing electricity consumption among branches of a large Italian bank. The first intervention consists in the technological renovation of 70 branch buildings through the installation of an automated energy management system. The second is an energy-saving competition that involved more than 500 branches for a year. Using two separate difference-in-differences estimations, we find that the technological renovation curbs electricity consumption by 15.8 percent overall, and by more than 25 percent outside the main work schedule. The behavioral intervention reduces electricity consumption, by around 6 percent outside the main work schedule, and by 2.4 percent overall, although not significantly so. The estimated cost-effectiveness ranges between 3.4 and 8.8 € cents per kWh saved for the technological intervention, and 9.8 and 17.8 € cents per kWh saved for the behavioral intervention. Our findings suggest that for both interventions, energy savings in the workplace are more easily obtained by reducing passive energy waste than through behavioral change during working hours.
In this Journal Club, Sandra Geiger describes a meta-analysis of intervention studies on spillover effects in environmental decisions.
Full-text available
An emerging body of literature has contributed to understanding behavioral spillover; however, a limited range of behaviors and psychological pathways have been studied. The current study investigates whether starting to compost, a relatively difficult behavior receiving limited attention in the spillover literature, results in spillover to household waste prevention behaviors, including food, energy, and water waste prevention. It also tests cognitive accessibility as a new mediator in the spillover process, and advances an integrative process model to address methodological inconsistencies in the spillover literature. Data are from a 2015 longitudinal field experiment to increase composting. Participants (N = 284) were residents of Costa Mesa, California, who received a structural intervention (i.e., curbside organic and BehaviorSintov et al. research-article2017 2 Environment and Behavior 00(0) waste bins) and procedural information about composting. Positive spillover was observed. Additionally, cognitive accessibility partially mediated the relationship between composting and energy and water waste-prevention behaviors. Future research should adopt a consistent definition of spillover and explore additional pathways.
Full-text available
Urgent and radical transition to lower-carbon forms of society is imperative to limit current and future climate change impacts. Behavioral spillover theory offers a way to catalyze broad lifestyle change from one behavior to another in ways that generate greater impacts than piecemeal interventions. Despite growing policy and research attention, the evidence for behavioral spillover and the processes driving the phenomenon are unclear. The literature is split between studies that provide evidence for positive spillover effects (where an intervention targeting an environmentally conscious behavior leads to an increase in another functionally related behavior) and negative spillover effects (where an intervention targeting an environmentally conscious behavior leads to a decrease in another functionally related behavior). In summarizing findings, particular attention is given to the implications for climate-relevant behaviors. While few examples of climate-relevant behavioral spillover exist, studies do report positive and negative spillovers to other actions, as well as spillovers from behavior to support for climate change policy. There is also some evidence that easier behaviors can lead to more committed actions. The potential contribution of social practice theory to understanding spillover is discussed, identifying three novel pathways to behavioral spillover: via carriers of practices, materiality, and through relationships between practices within wider systems of practice. In considering future research directions, the relatively neglected role of social norms is discussed as a means to generate the momentum required for substantial lifestyle change and as a way of circumventing obstructive and intransigent climate change beliefs. For further resources related to this article, please visit the WIREs website.
Full-text available
When implementing environmental education and interventions to promote one pro-environmental behavior, it is seldom asked if and how non-target pro-environmental behaviors are affected. The spillover effect proposes that engaging in one behavior affects the probability of engagement or disengaging in a second behavior. Therefore, the positive spillover effect predicts that interventions targeting one specific behavioral have the capacity to promote non-targeted and/or future pro-environmental behaviors. However, the negative spillover effect predicts that engaging in a first pro-environmental behavior will prevent or decrease a second pro-environmental behavior. Since the theoretical and empirical basis for positive and negative spillover effects are not sufficiently understood, the present paper (1) suggests a distinction between behavioral, temporal, and contextual spillovers (2) reviews the existing spillover research literature across a variety of environmental domains, (3) presents potential moderators governing the direction of spillover effects, and finally (4) discuss techniques to promote pro-environmental spillovers.
We studied publication bias in the social sciences by analyzing a known population of conducted studies—221 in total—in which there is a full accounting of what is published and unpublished. We leveraged Time-sharing Experiments in the Social Sciences (TESS), a National Science Foundation–sponsored program in which researchers propose survey-based experiments to be run on representative samples of American adults. Because TESS proposals undergo rigorous peer review, the studies in the sample all exceed a substantial quality threshold. Strong results are 40 percentage points more likely to be published than are null results and 60 percentage points more likely to be written up. We provide direct evidence of publication bias and identify the stage of research production at which publication bias occurs: Authors do not write up and submit null findings.
Psychological studies testing behavioral spillover—the notion that behavior change resulting from an intervention affects subsequent similar behaviors—has resulted in conflicting findings in the environmental domain. This study sought to further demarcate the spillover process by asking participants to engage in a difficult first pro-environmental behavior, reducing red meat consumption, for either health or environmental reasons. Evidence of spillover was tested via a subsequent monetary donation to an environmental organization. While there was no evidence of spillover for those in the green behavior condition, those in the health behavior condition were less likely to donate relative to controls. There was evidence that pro-environmental behavior led to an increase in environmental concern. In turn, environmental concern was associated with an increased likelihood of donating. Environmental concern may, thus, be one route to positive spillover in some subsets of the population.
Pro-environmental behavioural spillover – when performing one pro-environmental behaviour (PEB) increases the likelihood of performing another – has been identified as a possible way to increase the amount of environmentally friendly behaviours that individuals perform. The current research investigated this spillover process, the role of chronic environmental motivations, goal priming and behavioural similarity. Three studies (two conducted with students and one conducted with the general Australian public) provided evidence to suggest that positive spillover occurs between PEBs that are similar in terms of the resources required to perform them, but not between PEBs that are resource-dissimilar. There was no evidence to suggest that negative spillover (the instance where performing one PEB lessens the likelihood of subsequently performing another) occurred. Chronic environmental striving seems to independently influence the performance of PEBs, especially spending time to be more environmentally friendly. The role of priming goals in the spillover process remains unclear.