Mechanisms of Pro-Environmental Behavioural Spillover

Abstract

If one pro-environmental behaviour (PEB) influences subsequent behaviours, this spillover effect holds significant implications for environmental policies. This paper systematically examines the psychological mechanisms underpinning PEB spillover by testing a comprehensive set of mediators within the same empirical framework. Studies were conducted in Norway (N = 1,574) and Germany (N = 1,938), using correlational and experimental designs. The studies did not detect any significant spillover effects. However, correlational patterns of PEB were detected, revealing consistency between prior behaviours and policy support on the one hand and intentions to engage in other PEBs and supporting other policies on the other hand. Among a comprehensive set of mediators tested, environmental concern, pro-environmental self-identity, and global warming worry were identified as significant. Experimental interventions had limited effects on the initial behaviour. We recommend longitudinal studies to better capture the dynamics of behavioural change over time and preregistration protocols to address biases in research.

Keywords

pro-environmental behaviour behavioural spillover mediation environmental self-identity environmental concern

Introduction

Global environmental problems caused by human behaviour are intensifying (Intergovernmental Panel on Climate Change [IPCC], 2021) and changes to the behaviour of individuals are crucial to tackle these problems (Dietz et al., 2009). As a result, considerable resources are currently used to influence individuals’ pro-environmental behaviour (PEB). If PEB has a tendency to spill over to increase other behaviours aimed at the same superordinate environmental protection goal, more knowledge is needed about this effect to develop ways of boosting it. Alternatively, if the spillover effects are largely negative, leading to a reduction in other PEBs, more knowledge is needed about how to avoid such effects. Hence, a deeper insight into the relationships between PEBs could substantially help efforts to address current environmental crises. Therefore, in this paper, we aim to increase understanding of the prevalence and the underlying mechanisms of pro-environmental behavioural spillover by forwarding a comprehensive examination of PEB spillover.

We define pro-environmental behavioural spillover as a change in the performance of one PEB triggered by prior performance of another PEB by the same actor (Nash et al., 2017; Thøgersen & Crompton, 2009). The performance of the initial behaviour (PEB1) might either increase or decrease the likelihood of performing the second behaviour (PEB2), which is referred to as positive and negative spillover, respectively (Truelove et al., 2014). The presence and prevalence of PEB spillover has implications for policy design, as policy makers would benefit from a better understanding of how programs that target one PEB might begin a cascading effect to broader PEB performance and alternatively how such campaigns may inhibit or reduce future PEB uptake.

Previous research on behavioural spillover has found varied results that have so far not been explained by a unified theory. Some have found negative spillover (e.g., Geng et al., 2016; Lin & Chang, 2017), others have found positive spillover (e.g., Goetz et al., 2024; Margetts & Kashima, 2017), and yet others found that both were present and effectively cancelled each other out (e.g., Carlsson et al., 2021; Truelove et al., 2021). Meta-analyses of 77 (Maki et al., 2019) and 63 effect sizes (Geiger et al., 2021) from environmental behavioural spillover studies have found statistically negligible spillover effects following an experimental intervention targeting PEB1, although studies looking at behavioural intention (rather than behaviour) more often find positive effects. Longitudinal studies have had more success in detecting spillover (e.g., Juhl et al., 2017; Karmarkar & Bollinger, 2015; Thøgersen & Ölander, 2003; Thøgersen et al., 2024), but no meta-analysis has been conducted on non-experimental spillover studies. The stronger effects in longitudinal studies might be because they allow participants enough time for attitudes and behaviours to change. In longitudinal studies, if a change in one behaviour occurs, this change then has repeated opportunities to potentially stimulate changes in other behaviours over time (as illustrated for example in Juhl et al., 2017). To better understand when spillover may occur and how positive spillover could be strengthened, we believe it is important to understand how spillover manifests itself and under which circumstances it is more or less likely.

In this paper, we focus on the role of underlying processes explaining spillover effects. Several candidate psychological mechanisms for spillover effects have been postulated and tested in previous research (Burger et al., 2022; Carrico et al., 2018; Lacasse, 2016; Lauren et al., 2016, 2017; Manika et al., 2021; Margetts & Kashima, 2017; Sintov et al., 2019; Thøgersen et al., 2024; Truelove & Nugent, 2020; Truelove et al., 2016, 2021; Van der Werff & Steg, 2018; Van der Werff et al., 2014; Wang et al., 2021; Werfel, 2017; Wolstenholme, 2020; Xu et al., 2018), but there have been few attempts to unify the scattered findings. Two studies reported in this paper address this gap by systematically testing a wider collection of both established and underexplored psychological mediators within a single empirical framework. This integrative approach enables a more comprehensive understanding of the psychological architecture underlying PEB spillover, allowing us to assess the relative importance and potential interrelations of different mechanisms. For example, it allows detecting mediations in opposite directions that may cancel each other out, leading to a reduced or even null total effect between behaviours (Lacasse, 2016). Our approach builds on a handful of previous studies discussed below and summarised in Table 1. Its ultimate goal is to develop a simple yet comprehensive model of PEB spillover.

Table 1.

Overview of Studies Testing Multiple Mediators at Once.

Article	Mediators	Summary of findings
Lacasse (2016)	Self-identity Guilt	Study 1: Self-identity positively mediated spillover and guilt negatively, leading to an overall null effect. Study 2: Self-identity positively mediated spillover, but guilt did not.
Truelove et al. (2016)	Self-identity Guilt Global warming worry	Spillover effects were not mediated by any of these mediators.
Lauren et al. (2017)	Self-identity Self-efficacy Contribution ethic	Self-identity was the only significant mediator.
Carrico et al. (2018)	Moral self-image Concern	Environmental concern positively mediated the effect of PEB condition on whether or not people donated. Moral self-image did not mediate the effect.
Xu et al. (2018)	Concern Self-identity Guilt	Spillover effects were not mediated by any of these mediators.
Truelove and Nugent (2020)	Guilt Self-identity Concern	Self-identity was most important.
Truelove et al. (2021)	Self-identity Guilt	There was a positive pathway through identity and negative pathway through guilt which cancelled each other out.

The following sections cover the mechanisms that have been used to explain spillover in previous research, describe attempts of combining more than one spillover mechanism, and introduce operationalisations used to measure spillover, before moving on to discuss the current research.

Mechanisms Explaining Positive Spillover

Environmental Self-Efficacy

To explain positive PEB spillover, self-efficacy is often mentioned (Lauren et al., 2017; Nash et al., 2017; Steinhorst et al., 2015). Successfully performing an action can lead one to believe that they can do it again, and can also boost confidence in one’s ability to take other, similar actions (Bandura, 1982). Consequently, a boost of environmental self-efficacy following PEB1 has been found to be a likely mechanism of positive spillover (Lauren et al., 2016, 2017; Nash et al., 2017; Steinhorst et al., 2015).

Environmental Self-Identity

Another commonly mentioned mechanism underlying positive spillover is based on self-perception theory (Bem, 1972). According to self-perception theory, an individual’s behaviour influences the way they see themselves, and therefore PEB1 can lead to a shift in self-perception (self-identity) that is consistent with the behaviour. If PEB1 is motivated by a pro-environmental goal, it is expected to strengthen one’s environmental self-identity or at least make it more salient (Van der Werff & Steg, 2018; Van der Werff et al., 2014). Since an individual’s behaviour is partly determined by the way they see themselves, a boosted environmental self-identity may subsequently lead to more environmentally friendly choices. Many studies have found this effect (Elf et al., 2019; Lacasse, 2016; Lauren et al., 2017; Truelove & Nugent, 2020; Van der Werff et al., 2014). However, others have failed to confirm an identity pathway for spillover (Poortinga et al., 2013), or even found environmental self-identity to mediate negative spillover for certain groups (Truelove et al., 2016).

Environmental Concern

Self-perception theory also predicts that people use their actions to gauge their attitudes (Bem, 1972), such as environmental concern, in our case. Action for environmental reasons has been found to serve as a reminder of the threats the environment is facing, thereby increasing environmental concern and potentially pro-environmental behaviour (Carrico et al., 2018; Xu et al., 2018).

Goal Importance

The relative salience and perceived importance of a goal can be primed or reinforced by goal-directed actions, which could in turn lead to further actions for the same goal (Fishbach & Dhar, 2005; Stangherlin et al., 2023). For example, PEB spillover has been shown to occur as a result of priming environmental goals (Devezer et al., 2014; Margetts & Kashima, 2017).

Mechanisms to Explain Negative Spillover

Moral Licensing

Successful execution of actions can lead to an increased moral self-concept that could provide a “licence” to act less pro-environmentally or even anti-environmentally. This idea of a moral licence has been extensively tested in health contexts (e.g., Fishbach & Dhar, 2005) and in pro-social contexts (e.g., Conway & Peetz, 2012; for overview see Blanken et al., 2015) but it has also been found in environmental contexts (Lin & Chang, 2017; Meijers et al., 2015, 2019). Others have measured moral self-concept as an outcome of behavioural interventions and found that it could not explain spillover effects (Carrico et al., 2018; Werfel, 2017).

Contribution Ethic

Similar to a moral licencing effect, a contribution ethic reflects the situation where performing a PEB makes a person feel that they have done their fair share for solving a specific environmental problem. This is expected to lead individuals to “rest on their laurels” or in other words to lead to negative PEB spillover (Thøgersen & Crompton, 2009). As opposed to moral licencing, which is usually connected to a general moral self-concept, contribution ethic has usually been operationalised as specific to the cause in question (Truelove et al., 2014). However, contribution ethic has rarely been measured as a mediator of PEB spillover and some research has even found that contribution ethic mediated a positive spillover on public sphere action intentions (Lauren et al., 2017).

Alleviation of Guilt

Many people experience guilt in relation to environmental problems (Marczak et al., 2023) and the more guilt people feel about their environmental actions the more they engage in PEB (Shipley & Van Riper, 2022). PEB that is motivated by emotional regulation can take the form of legitimate action, but it can also lead people to take a single action that provides emotional relief (Hansen et al., 2006). Alleviation of guilt experienced from engaging in PEB1 has been found to motivate negative spillover (Burger et al., 2022; Lacasse, 2016; Truelove et al., 2021), although other studies have found this not to be a relevant factor (Truelove & Nugent, 2020; Xu et al., 2018).

Single-Action Bias and Global Warming Worry

While environmental concern has been found to explain positive spillover effects (see Section “Environmental Concern”), there is some evidence that global warming worry can be reduced by an initial action, leading to less propensity for further actions (Hansen et al., 2006; Weber, 1997), coined single-action bias by Weber (1997). However, a more recent test of this hypothesis did not find global warming worry to mediate negative PEB spillover (Truelove et al., 2016). The concepts of environmental concern and global warming worry are similar and clearly tied together, but have been discussed in different parts of the literature using different measures and therefore we treat them as distinct variables. While concern is generally regarded as a rational cognitive response, worry is an emotion, that can arise with or without rational justification. Emotional reactions are considered to be more transient, and more likely to be alleviated by a single action. In contrast, concern is typically more persistent and may remain despite taking a small action, reflecting its basis in reasoned judgement.

Combining Mediators

The potential mechanisms outlined above are not mutually exclusive, that is, while an initial pro-environmental action may boost one person’s environmental self-identity, it may make another feel like they have acquired a licence to not act pro-environmentally. If such conflicting mechanisms are at play, spillover may be happening, but positive and negative pathways to spillover cancel each other out, leading to reduced effect sizes or the conclusion that no spillover is found (Lacasse, 2016; Truelove et al., 2021). If multiple positive pathways are at play simultaneously, they may work in tandem or one may be stronger, “crowding out” the effect of the other (Lauren et al., 2017).

To the best of our knowledge, only seven papers have simultaneously measured more than one mediator to assess their relative importance in explaining environmental behavioural spillover (see Table 1 for a summary). At most, they have co-tested three mediators. The current research adds to previous research on spillover processes by systematically testing a wider collection of psychological constructs that can be expected, theoretically or empirically, to be pathways of spillover. In particular, in the cross-sectional part of two studies described below, we concurrently measure six to seven psychological mediators of spillover, and we use a set of eight to nine mediators in the experimental part of these studies. This allows for the possibility of finding conflicting or competing mechanisms, and for testing which mediators have more influence than others.

Piecing together a coherent theory of PEB spillover from separate studies each examining a small number of mediators can be difficult. In contrast, our approach holds more promise for cumulative theory building. Given certain limitations of the empirical data we collected (discussed in Section “General Discussion”), our contribution should be seen more as a proof of concept than a final word on this topic.

Current Research

Testing PEB spillover empirically is not straight-forward. The simplest approach is to measure past behaviour retrospectively and test for correlations between past behaviour and behavioural intentions (e.g., Lauren et al., 2016; Manika et al., 2021; Wang et al., 2021). Another way of testing spillover is by using longitudinal designs to capture correlations between different behaviours enacted at different time points (e.g., Ek & Miliute-Plepiene, 2018; Stangherlin et al., 2023; Thøgersen et al., 2024). Finally, spillover can be evaluated experimentally by applying an intervention to shift PEB1 and measuring whether this shift in PEB1 leads to a change in PEB2 that was not targeted by the intervention. In the present work, we assess spillover in three ways: based on cross-sectional, longitudinal, and experimental data.

We follow the recommendations of Galizzi and Whitmarsh (2019) for measuring spillover with an experimental study, by using a between-subjects randomly assigned intervention targeting PEB1, observing differences in PEB1 across conditions, and subsequently measuring PEB2 as behavioural intentions across conditions. In our first study, we give some participants the opportunity to engage in PEB1 and compare them with those who do not get that opportunity, in terms of their subsequent PEB intentions and policy support intentions. We also compare those who get the opportunity to engage in PEB1 to a third group, who get the same opportunity but with an additional default nudge. See Figure 1a for a schematic depiction. In the second study, we use defaults again, as well as a simple descriptive norm message (see Figure 1c).

Figure 1.

Overview of study designs: (a) Study 1 experimental design, (b) Study 1 correlational design, (c) Study 2 experimental design, and (d) Study 2 longitudinal design.

We additionally measure PEB and policy support retrospectively as PEB1 (see Figure 1b). While this correlational method cannot be used to prove that spillover is happening, spillover is a plausible interpretation of associations between behaviours.

In Study 2, we also included a follow-up measure 2 weeks after the first survey, allowing us to test the effect of PEB1 at one time on PEB2 at a later time (see Figure 1d).

Study 1

Method

Participants

The study was conducted online in June 2022, with a quota sample of Norwegian residents recruited by a market research company (n = 1,574 after preregistered data exclusions), see socio-demographic characteristics of the sample in Table 2. The studies received approval from Sikt (Norwegian Agency for Shared Services in Education and Research, reference numbers 669080 and 649631). Our institutions did not require an additional approval for the studies. The studies were conducted according to the principles expressed in the Declaration of Helsinki, as well as with ethics guidelines by the American Psychological Association and with national and institutional regulations. Written informed consent was provided by all participants.

Table 2.

Demographic Characteristics of Both Study Samples.

Variable	Category	Study 1 (%)	Study 2 (%)
Gender	Female	51.9	51.1
	Male	47.9	48.8
	Non-binary/other	0.2	0.1
Age	18–29	15.8	14.9
	30–39	17.9	11.4
	40–59	35.3	34.3
	60+	30.9	39.4
Education	8 years or less	0.6	5.0
	9–13 years	37.5	57.5
	Undergraduate degree	37.5	13.8
	Graduate degree	23.7	23.4
	Don’t want to say	0.7	0.4
Income (yearly NOK/monthly €)	Less than 400,000 NOK	26.5
	400,000–699,999 NOK	42.6
	700,000–1,000,000 NOK	14.0
	More than 1,000,000 NOK	5.5
	Less than €2,000		37.5
	€2,000–3,000		24.0
	€3,000–5,000		23.2
	More than €5,000		8.4
	Don’t want to say/don’t know	11.4	6.9

Study Design and Procedure

Participants were first asked about socio-demographics, possible moderators of spillover (analyses presented in Supplemental Appendix A), and PEB and policy support.

Participants were then randomly assigned to one of three experimental groups. The opportunity group was offered the opportunity to donate a portion of their participation fee to an environmental organisation (WWF). They could freely choose to donate 0, 5, 10, 15, 20, or 25 NOK (approximately 2.5 EUR). The default group got the same opportunity, but the 25 NOK donation was (provisionally) pre-selected. The control group did not get the chance to donate and directly moved to the next part of the questionnaire.

Lastly, participants answered questions about mediators, PEB intentions, and policy support intentions, which were used as PEB2 measures. The full questionnaire and a preregistered analysis plan can be found on OSF (anonymised).

Measures

All multi-item scales were computed by taking the mean of the relevant items, following preregistered criteria. Question order was randomised within each conceptual block (presented as paragraphs below). The order of the conceptual blocks was also randomised within the categories mediators, PEB and policy support, and moderators. See Supplemental Appendix B for means, standard deviations, and Cronbach’s alpha of all scales.

PEB in the past was measured with nine items: “In the past month I have eaten a meat dish” with response options (1) “More than four times a week”, (2) “Three or four times a week”, (3) “Once or twice a week”, (4) “Less than once a week”, (5) “Never”. “In the past month I turned off the tap while brushing teeth”; “In the past month I turned off lights when not in use”; “In the past month I waited until the washing machine was full before running it”; “In the past month I walked or cycled instead of driving to places within 5 km”; “I recycle batteries and other electronic waste”; “In the past month I used reusable shopping bags”; “In the past month, the dairy products I purchased were organic”; “In the past month, the fruits and vegetables I purchased were organic” with response options (1) “Never”, (2) “Rarely”, (3) “Sometimes”, (4) “Often” and (5) “Always”. Cronbach’s alpha was slightly below the preregistered threshold of .70, but since item removal did not increase scale reliability, a Cronbach’s alpha of .69 was deemed acceptable.

Policy support was measured with two items, “I support tax rebates for people who purchase electric vehicles” and “I support an increase in tax on carbon emissions” on a scale from (1) “Strongly disagree” to (5) “Strongly agree”. Spearman–Brown scale reliability was too low for these items (.55) and therefore the items were kept separate for further analyses.¹

PEB intentions were measured with 11 items: “In the following month I will eat meat”; “In the following month I will only take showers shorter than 5 minutes”; “In the following month I will reduce my highway driving speed”; “In the following month I will turn off the tap while soaping up in the shower”; “In the following month I will recycle glass”; “In the following month I will recycle plastic”; “In the following month I will donate to an environmental organisation”; “In the following month I will purchase some organic dairy products”; “In the following month I will purchase some organic fruits and vegetables”; “In the following month I will lower the temperature in my apartment or house by 1 degree Celsius, compared to what I am used to”; “In the following month I will subscribe to a “green” electricity tariff from my energy provider” with the response options “Yes” and “No” (Cronbach’s α = .69).

Policy support intentions were measured with three items: “In my future voting decisions, I will support providing tax rebates for people who purchase solar panels”; “In my future voting decisions, I will support providing tax rebates for people who purchase heat pumps”; and “In my future voting decisions, I will support stricter regulation of the fishing industry to prevent over-fishing” with response options “Yes” and “No”. The third question did not form a reliable scale with the other two items and was therefore dropped from analysis (Spearman–Brown reliability = .71).

Contribution ethic was measured with two items adapted from Lauren et al. (2017): “How much have you done to protect the environment today?” with response options ranging from (1) “Nothing” to (5) “More than enough” for the first item and response options ranging from (1) “Nothing” to (5) “A lot more than my share” for the second item (Spearman–Brown reliability = .79).

Environmental concern was measured with two items: “How concerned are you about environmental damage?” and “How concerned are you about global warming?” adapted from Carrico et al. (2018) with response options ranging from (1) “Not at all concerned” to (5) “Extremely concerned” (Spearman–Brown reliability = .82).

Environmental self-identity was measured with three items such as: “I see myself as a pro-environmental person” (Van der Werff et al., 2013) with a response scale ranging from (1) “Strongly disagree” to (5) “Strongly agree” (Cronbach’s α = .87).

Global warming worry was measured with two items, asking how serious a threat global warming is to nature and humans with response options ranging from (1) “Not serious at all” to (5) “Extremely serious” (Spearman–Brown reliability = .93). Items adapted from Leiserowitz (2006) and Truelove et al. (2016).

Moral self-concept was measured with two items, “I am a moral person” and “I am an altruistic person” with a 5-point agreement scale (Truelove et al., 2016). The two items did not form a reliable scale, and since a large portion of participants responded “I don’t know” to the question about altruism, this item was dropped.

Self-efficacy was measured with two items “I feel confident that I can protect the natural environment” and “I feel capable of behaving in a climate-friendly way in everyday life” with a 5-point agreement scale (Spearman–Brown reliability = .65) adapted from Lauren et al. (2017).

Guilt was measured with three items asking about guilt, regret, and disappointment about the participant’s donation decision with response scales ranging from (1) “Not at all” to (5) “A lot” (Cronbach’s α = .91). Items based on Truelove et al. (2016) and Truelove and Nugent (2020).

Reactance was measured with two items: “Did you see the donation task as a demand?” and “Did you feel the donation task encroached on your freedom in any way?” with response options ranging from (1) “Not at all” to (5) “Very much so” (Spearman–Brown reliability = .84).

Preregistered Hypotheses

We preregistered a total of 194 hypotheses in Study 1 (link to preregistration: osf.io/anonymised). The first four hypotheses test spillover experimentally. First, we test whether being in the opportunity group influences further PEB intentions (H1) and policy support intentions (H2), compared to the control group. Next, we test whether being in the default group increases PEB intentions (H3) and policy support intentions (H4) through increased donations, compared to the opportunity group.

The next four hypotheses look for behavioural correlations. Specifically, we hypothesise that PEB predicts PEB intentions (H5) and policy support intentions (H6) and that policy support predicts PEB intentions (H7) and policy support intentions (H8). As preregistered, we control for experimental group in these analyses.

For each of these eight hypotheses we test whether the spillover effect can be explained by any of six mediating variables: contribution ethic, environmental concern, environmental identity, global warming worry, moral self-concept, and self-efficacy. In H3 and H4, where donation is already a mediating factor, the psychological mediator is included as a second mediator in a serial path. In these cases, the additional mediators guilt and reactance are also tested. See pre-registration on OSF (osf.io/anonymised, Section “Study 1”) for precise phrasing of all hypotheses.

Analysis

All hypotheses are tested with the R package “lavaan” and 5,000 bootstrap samples (Rosseel, 2012). The package “semhelpinghands” was used to get standardised coefficients and bootstrapped confidence intervals (Cheung, 2023). Most variables have some missing data, up to 18.3% for moral self-concept. The full information maximum likelihood method was used to account for missing data. To account for false discovery rate (FDR) across the hypothesis tests reported in this paper and Supplemental Appendix A we applied the Benjamini–Yekutieli correction (Benjamini & Yekutieli, 2001) for FDR, which is effective under arbitrary dependence among tests, making it suitable given the overlap in variables across models. The final number of hypothesis tests after some adjustments (see Section “Results”) was 166 tests.

Mediation hypotheses were tested as indirect effects of the independent variable on the dependent variable either through one mediator or through PEB1 and another mediator in a series.

The market research company used two panels to recruit participants and donation amounts were significantly different between the panels. Therefore, all tests control for the panel from which participants were recruited (not preregistered).

We also conducted exploratory analysis of multiple mediators in one model, using the XMed function of the regsem package in R (Serang et al., 2017), which offers the possibility of using regularisation methods that can be used to determine which mediator variables can be reduced to a zero effect and therefore excluded from analysis. For each of the mediation hypotheses listed (H1 to H8) the XMed procedure was used to select from the entire set of tested mediators those with specific indirect effects greater in absolute value than 0.001. Although multicollinearity was not a concern (all VIF < 3.0 and all r < .80; James et al., 2017) some mediators are moderately correlated, and therefore we used the Elastic Net regularisation method (Li et al., 2021).

A post hoc power sensitivity analysis using G*Power revealed that, given N = 1,040 (control group and one experimental condition), and an adjusted alpha level of α = .00026 (Bonferroni correction for 166 hypotheses), the study had 80% power to detect effects of f² = .02. This corresponds approximately to Cohen’s d ≈ 0.28 and to r ≈ .14, a small to medium effect size. Note that this is a conservative estimate, as the Bonferroni correction is more conservative than the FDR correction we used.

Results

Experimental Spillover Tests (Tests of H1–H4)

Being in the opportunity group did not influence behavioural and policy support intentions, compared to the control group. H1 and H2 were therefore rejected (see Table 3 – row 1, columns 1 and 2). The default group (M = 8.55 NOK, SD = 10.70) donated slightly more compared to the opportunity group (M = 7.34 NOK, SD = 10.40), but this was not significant and the effect size is negligible (t[1,030] = 1.83, p = .068, Cohen’s d = 0.11). Indirect effects on PEB intentions and policy support intentions through donations were not significant after FDR correction and therefore H3 and H4 were rejected (see Table 3 – row 1, columns 3 and 4).

Table 3.

Size of the Standardised Indirect Effect Through Each Mediator in Models Including All Mediators Selected by an Elastic Net Regularisation Test.

		1	2	3	4	5	6	7	8	9	10
	Mediators of spillover	Op → PEBI	Op → PSI	Def → PEBI^a	Def → PSI^a	PEB → PEBI	PEB → PSI	PS_EV → PEBI	PS_CT → PEBI	PS_EV → PSI	PS_CT → PSI
1	Partial main spillover effect (no mediator)	0.017	0.031	−0.048	0.018	0.415***	−0.013	0.080***	0.091***	0.228***	0.088**
2	Contribution ethic	−0.006	−0.010	−0.001	−0.000	NS	0.018*	−0.006*	0.006*	0.004	0.004
3	Env. concern	−0.000	NS	NS	NS	0.083***	0.032	0.061***	0.110***	0.014	0.029
4	Env. self-identity	NS	NS	NS	NS	0.084***	0.083***	0.057***	0.113***	0.030***	0.064***
5	Global warming worry	−0.005	−0.013	NS	NS	NS	0.042**	0.016	0.027	0.019	0.044*
6	Guilt	—	—	NS	NS	—	—	—	—	—	—
7	Moral self-concept	−0.003	NS	NS	NS	0.005	NS	0.019*	0.032**	NS	NS
8	Self-efficacy	0.005	NS	NS	NS	0.011*	NS	0.004	0.008*	NS	NS
9	Reactance	—	—	NS	NS	—	—	—	—	—	—
10	Total effect	0.009	0.009	−0.015	0.036	0.599***	0.163***	0.243***	0.387***	0.295***	0.224***

Note. ^aSerially mediated effect of the default intervention on PEBI/PSI through donations and then the proposed mediators. Mediation pathways that could be reduced to zero according to the regularisation tests are marked as NS.

p < .050, **p < .010, ***p < .001.

Behavioural Correlation (Tests of H5–H8)

Prior PEB positively related to behavioural intentions (see Table 3 – row 1, column 5) and policy support intentions (see Table 3 – row 1, column 6) controlling for the experimental group to rule out unintended experimental effects. H5 and H6 were supported.

Because the two items used to measure policy support did not form a reliable scale (as noted in Section “Measures”), the effect of policy support was tested twice, once for each policy support variable. Both supporting tax rebates for electric vehicles and supporting increased carbon emissions tax predicted increased behavioural intentions (see Table 3 – row 1, columns 7 and 8) and policy support intentions (see Table 3 – row 1, columns 9 and 10). H7 and H8 were supported.

Mediation Analysis (Tests of H1–8_med1–8)

Since the main experimental spillover hypotheses H1 to H4 were not supported, there was no theoretical possibility of finding simple mediations of these effects. Therefore, we deviated from the pre-registration and rejected all mediation hypotheses related to H1 to H4 without further testing (i.e., H1–2_med1–6 and H3–4_med1–8).

As hypothesised, environmental concern (H5–8_med2), environmental identity (H5–8_med3), and global warming worry (H5–8_med4) consistently mediated all behavioural correlations positively. Self-efficacy positively mediated the effect of PEB on PEB intentions (H1_med6) as well as e-car policy support on PEB intentions (H7_med6). Moral self-concept also mediated the effect of both types of policy support on PEB intentions (H7_med5) positively. See Supplemental Appendix C for an overview of mediation test results.

Exploratory Analysis of Multiple Mediators at Once

We exploratorily tested models with multiple mediators. This is important because when influential mediators are omitted, the importance of others may be over or under-estimated (as discussed in Section “Combining Mediators”). To do so, we first used a regularisation method to select mediators and then ran structural equation models with all selected mediators (following Ammerman et al., 2018). In Table 3, each column represents one path model. The first row shows the coefficient for the partial main effect of the independent variable on the outcome, and rows 2 to 9 present the indirect effect through the specified mediator. Row 10 shows the total effect of the independent variable on the outcome directly and through all included mediators. It should be noted that correction for false discovery rates was not applied for the exploratory analysis, but instead we only considered p-values below .001 as significant.

This analysis revealed that, while other mediators are held constant, environmental self-identity consistently and positively mediates behavioural correlations. Environmental concern positively mediates the effect of prior PEB and both types of policy support on PEB intentions.

Exploratory Analysis of Correlations Between Individual Behaviour Items

At a suggestion from an anonymous peer reviewer, we repeated the behavioural correlation analysis with individual behaviour items rather than with behavioural indexes. This analysis, which is summarised in Supplemental Appendix Table D1, confirms positive correlations between all prior PEBs on the one hand and all behavioural intentions and policy support intentions on the other hand. The standardised coefficients of these tests range from 0.02 to 0.69, with most coefficients between .10 and .40. These results also provide suggestive evidence that correlations are stronger between more similar behaviours such as buying organic dairy and buying organic fruit and vegetables (β = .69, p < .001) and recycling electronics and recycling glass (β = .41, p < .001). This is consistent with previous literature finding that similarity of behaviours increases chances of positive correlation (Margetts & Kashima, 2017; Thøgersen, 2004).

However, we also find significant correlations across categories. For instance, bringing bags to the grocery store is correlated with intentions of slowing driving speeds (β = .30, p < .001) and buying organic fruit and vegetables is correlated with intentions of donating to an environmental organisation (β = .40, p < .001).

Discussion

In Study 1 we looked for spillover from donating to PEB intentions and policy support intentions as well as correlations between PEB and policy support in the past and intentions. We find that the opportunity to donate does not influence PEB intentions, and although the default for donating increases donations slightly and higher donations correlate with higher intentions, the indirect effects are very small and not significantly different from zero. This is at least partially due to the manipulation not affecting the donations strongly enough.

Correlational tests show positive associations between PEB and policy support on the one hand and PEB intentions and policy support intentions on the other hand. This suggests positive spillover, rather than negative. As hypothesised, these effects are positively mediated by environmental concern, self-identity, global warming worry, and in some cases also by self-efficacy. These findings are fairly consistent with previous research on spillover mediators (Carrico et al., 2018; Lauren et al., 2017; Truelove & Nugent, 2020), but our study is the first that measures all these mediators concurrently and shows that each of the commonly tested mediators environmental concern and self-identity can uniquely explain part of the correlation between behaviours and policy support. It should be noted that since this is a correlational analysis, equivalent models cannot be ruled out. Therefore, each of these mediators could also be potentially conceptualised as predicting both PEB1 and PEB2.

Following the typology of spillover conceptualisations presented in Section “Current Research”, we conclude that our data support the possibility of spillover correlationally (see hypotheses H5–H8) but not experimentally (see hypotheses H1–H4). While we attempted to use an experimental design to evaluate the studied mechanisms as well, the interventions we used were not strong enough to have the intended effects. The opportunity to donate did not affect behavioural intentions or policy support and the default for donating the highest amount did not increase donations significantly.

Study 2

Method

The second study was performed to expand on the findings of the first study by testing our hypotheses in a different sample, with a partly modified study design. Because the experimental manipulation in Study 1 did not increase PEB1, we changed the intervention in two ways. First, the design was adjusted by modifying the “default” manipulation so that there were only two options for donation, either the full amount or nothing. While we were not certain this change would be effective, we wanted to see whether a stark choice between donating something and donating nothing (compared to a more gradual choice between donating a little bit more or less as in Study 1) could potentially lead to more pronounced spillover effects due to its possible greater salience. Second, we added a new treatment group that received a manipulation appealing to descriptive social norms, which have been shown to influence donations (e.g., Croson et al., 2009; Vesely et al., 2022). Finally, a follow-up survey was added two weeks after the main survey, enabling longitudinal tests of PEB spillover. For the longitudinal tests, to strengthen the claims of PEB spillover, we controlled for prior behaviour of the same kind as the outcome.

Participants

The study was conducted online in March and April of 2023, with a quota sample of German residents recruited by a market research company (n = 1,938 in the first wave and n = 1,321 in both waves after preregistered data exclusions). For sample characteristics see Table 2 above.

Study Design and Procedure

Participants were randomly assigned to one of three experimental groups. Every group was given the chance to donate their 2€ reward to Bund für Umwelt und Naturschutz Deutschland (German Federation for the Environment and Nature Conservation). The opportunity group and the default group correspond to those in Study 1, the opportunity group simply being offered the option to donate and the default group seeing the option to donate with the 2€ box already ticked. In addition, the norm group was nudged to donate with a descriptive norm message that stated how many people donated in Study 1 (44.2%).²

Participants first answered questions about socio-demographics and PEB in the last 2 weeks. After that they were randomly assigned to one of the experimental groups and offered the opportunity to donate. They then answered questions about hypothesised mediators and finally PEB intentions and policy support intentions.

In the follow-up survey, approximately 2 weeks after the first survey, participants reported their PEB over the last 2 weeks.

Measures

PEB in the past was measured in the categories food, resource saving and mobility. The food category was measured with three questions: “Thinking of your main meal each day (typically lunch or dinner), how many of your meals included meat?”; “What percentage of the dairy products you purchased were organic?”; “What percentage of the fruits and vegetables you purchased were organic?” with the answering scale (1) “Less than 20%”, (2) “20%–40%”, (3) “40%–60%”, (4) “60%–80%”, (5) “More than 80%” (the scale was reversed for the first question).

The resource saving category was measured with five questions: “How often did you turn off the tap while brushing teeth?”; “How often did you turn off lights when leaving a room for more than 10 minutes?”; “How often did you wait until the washing machine was full before running it?”; “How much of your plastic waste did you recycle?”; “How often did you buy new shopping bags when shopping?”. These had the answering scale (1) “Less than 20% of the time”, (2) “20%–40% of the time”, (3) “40%–60% of the time”, (4) “60%–80% of the time”, (5) “More than 80% of the time”. Cronbach’s alpha was below the preregistered threshold of 0.70, but since item removal did not increase scale reliability, a Cronbach’s alpha of 0.63 was deemed acceptable.

Mobility was measured with one item: “How often did you walk or cycle instead of driving to places within 5 km?” using the same answering scale as the resource saving.

Goal importance was measured with “Acting in environmentally friendly ways is an important a goal for me” on a scale from (1) “Strongly agree” to (5) “Strongly disagree”.

All other measures were identical to those in Study 1. See Supplemental Appendix B for descriptive statistics of all measures.

Preregistered Hypotheses

The first eight hypotheses explored experimental evidence for spillover. As detailed below, we test whether a default for donating influences three domains of PEB intention as well as policy support, through donation as a mediating factor. We similarly test whether these effects can also be found from the norm manipulation.

Being in the default group (vs. the opportunity group) shifts PEB intention in the domains of food (H1), resource saving (H2), and mobility (H3), as well as policy support intention (H4), the effect being mediated by donation.

Being in the norm group (vs. the opportunity group) shifts PEB intention in the domain of food (H5), resource saving (H6), and mobility (H7), as well as policy support intention (H8), the effect being mediated by donation.

The next twelve hypotheses assessed longitudinal evidence for spillover from PEB in one of four domains in the 2 weeks leading up to Wave 1 to PEB in other domains the next 2 weeks, measured at Wave 2.

PEB in the domain of food predicts PEB in wave 2 in the other domains: resource savings (H9), mobility (H10), and energy consumption (H11).

PEB in the domain of resource savings predicts PEB in wave 2 in the other domains: food (H12), mobility (H13), and energy consumption (H14).

PEB in the domain of mobility predicts PEB in wave 2 in the other domains: food (H15), resource savings (H16), and energy consumption (H17).

Energy consumption at wave 1 predicts PEB in wave 2 in the other domains: food (H18), resource savings (H19), and mobility (H20).

Crucially, for each of these twenty main hypotheses we tested whether the spillover effect could be explained by mediating variables. These were environmental concern, environmental identity, global warming worry, moral self-concept, self-efficacy, and goal importance. Additionally, we tested whether the experimental hypotheses H1–H8 were mediated by contribution ethic, guilt, or reactance (in the experimental hypotheses, where donation was already a mediating factor as noted above, the attitudinal mediator was included as a second mediator in a serial path). See preregistration with precise phrasing of each hypothesis (osf.io/anonymous).

Analysis

Hypotheses were tested largely in the same way as in Study 1, with some exceptions. All hypotheses were tested with Lavaan. Most variables have some missing data, up to 4.0% for policy support in wave 1 and 7.5% for energy use in wave 2. In case of mediations with a binary mediator, the DWLS estimator was used and therefore a method other than FIML was needed to account for missing data. This was done with stochastic regression imputation. For longitudinal tests, the ML estimator was used. To account for FDR across the 210 hypothesis tests, we applied the Benjamini–Yekutieli correction.

A sensitivity analysis revealed that, given N = 1,288 (control group and default condition), and adjusted α = .00024 (Bonferroni correction for 210 hypotheses), the study had 80% power to detect effects of f² = .02. This corresponds approximately to Cohen’s d ≈ 0.24 and to r ≈ .13, a small effect. Note again that this is a conservative estimate, as Bonferroni correction being more conservative than the Benjamini–Yekutieli correction.

Results

Both interventions were effective at increasing donations in the first wave. A chi-square test shows that the default group was significantly more likely to donate (37.1%) than the opportunity group (27.1%; χ² [1] = 16.52, p < .001, φ = 0.11³). And the norm group (32.4%) was also significantly more likely to donate compared to the opportunity group (χ² [1] = 7.33, p = .007, φ = 0.08). However, no experimental, longitudinal, or mediation tests were significant after correcting for false discovery rate, and therefore all hypotheses were rejected. As in Study 1, we repeated the analyses with individual behaviour items (see Supplemental Appendix Table D2). Again, these analyses show that prior PEB is most strongly related to PEB intentions within the same category of behaviours, and that these spillovers are all positive. However, we also find some longitudinal spillovers between categories. Waiting until the washing machine is full before running it had a negative spillover effect on the number of meals without meat (β = −.07, p < .001) and walking or cycling short distances positively spilled over to fewer meals with meat (β = .06, p < .001).

We also tested the effects of including multiple mediators in the same way as in Study 1, but again, we found no indirect effects that were significant at the .001 level. Correlational evidence for spillover was not evaluated in Study 2 as this was not part of the preregistered analysis plan.

Discussion

The finding of no spillover effects could be due to the short duration of the longitudinal tests, low effectiveness of the experimental intervention or a variety of other reasons. However, this adds to an increasing number of studies that either fail to find spillover or find it to be of a size that is of little interest for the purpose of changing PEB in the ways that are called for (Geiger et al., 2021; Maki et al., 2019).

General Discussion

In the two studies described in this paper, potential for spillover was only found using correlational data. That is, those who reported more engagement with PEBs and policy support in the past were more likely to report intentions to engage in further PEBs and policy support. Experimental and longitudinal ways to detect spillover did not yield any significant outcomes. The experimentally amplified PEB1 of donating to an environmental NGO did not result in increased intentions to engage in PEBs or policy support (Study 1), nor did it lead to increased levels of PEBs or policy support in the following 2 weeks (Study 2). Additionally, prior PEBs reported in the first wave of data collection did not impact further PEBs recorded 2 weeks later (Study 2).

Behavioural spillover is an inherently causal process, although it has often been studied in the past without employing causal methodologies. We set out to study this process with two methods that support causal conclusions: by experimentally manipulating PEB1 before measuring PEB2 and by measuring PEB at two different time-points. Since these methods did not yield significant results, the bulk of the results we report pertain only to correlations between behaviours.

Conceptualisations of spillover based on correlations have been criticised on the grounds that correlations between behaviours could be either caused by spillover across those behaviours and/or by joint antecedents (Sharpe, 2022). The current research is informative about the correlations between prior behaviour and policy support on the one hand and behavioural and policy support intentions on the other hand, which may suggest consistent pro-environmental actions, but cannot be used to verify causal direction. However, behavioural correlations can give valuable information about relevant psychological variables that are of interest for consistent PEB (Carrico, 2021).

Theoretical Implications

Results suggest moderate consistency effects in Study 1. These were not necessarily stronger between items that have many shared attributes, compared to seemingly disparate items, meaning that consistency in pro-environmental behaviour does not seem to be limited to similar behaviours. More research is needed to determine the domain specificity of spillover.

Environmental concern, pro-environmental self-identity, and global warming worry all consistently mediate consistency effects. While environmental concern and global warming worry share so much variance that they might be considered to be indicators of one latent construct, it is clear that pro-environmental self-identity is a distinct psychological construct and uniquely contributes to the explanation of the relations between behaviours. Those who are more concerned about the environment and global warming are more likely to show more intentions to continue to act pro-environmentally, after doing so in the past. The same is true for those who see acting pro-environmentally as a part of their identity. These findings inform spillover theorising by showing that a model of spillover should include both the changes in self-identity experienced when engaging in PEB, as well as the emotional and cognitive assessments of risk.

Methodological Implications

We have followed the suggestion of previous spillover researchers (Sintov et al., 2019) to correct for Type 1 error, since we tested a large number of hypotheses which often relied on some of the same variables. The studies had larger sample sizes than are typically seen in spillover research and were powered to detect small effect sizes (r ≈ 0.14). Therefore, we can confidently say that the lack of significant effects cannot be attributed solely to lack of power. However, other methodological limitations may have affected the spillover results, as discussed further in Section “Limitations and Future Research Directions”.

Despite not obtaining many statistically significant results, especially in Study 2, a clear strength of the present work is that it has been conducted within the framework of a formal preregistration protocol. If future preregistered studies on environmental behaviour spillover also find weak or null spillover effects, this could suggest that the understanding of spillover processes that emerges from the current literature (typically not based on preregistered analyses) may be in part influenced by publication bias, file-drawer problem, and related issues.

Default Interventions

The introduction of defaults for the highest donation option did not increase donations as much as expected, resulting in a negligible increase of donations in Study 1 and a small effect size in Study 2. This was surprising, since default interventions have shown very promising results, even when no additional effort is needed to change the choice (Ghesla et al., 2019). A meta-analysis of defaults found that on average they have a medium effect size, although they are less effective in environmental domains compared to consumer domains (Jachimowicz et al., 2019).

Descriptive Norm Intervention

Descriptive norms have also been shown to be quite effective in increasing donations to environmental causes (Vesely et al., 2022) and PEB more broadly (Bergquist et al., 2019), but the norm intervention in Study 2 only marginally increased the number of people who chose to donate. This may be due to the nature of the descriptive norm presented to participants – since we did not want to mislead participants, we used the real numbers from a previous study, where 44.2% of participants donated to an environmental organisation. It is possible that a stronger norm would have resulted in higher donation levels. To improve the research design in future studies on spillover we suggest pre-testing any interventions with the intended sample to make sure that the intervention has at least a medium effect size.

Limitations and Future Research Directions

One possible explanation for the small effects of the experimental interventions may be that before their administration all participants answered questions about their prior PEB. Some studies have used answering such questions as a means to make pro-environmental self-identity more salient (e.g., Van der Werff et al., 2014) or alter their self-perception (e.g., Lauren et al., 2017), and perhaps the donation intervention “drowned” in the effect of answering behavioural questions. This could be considered a limitation of the experimental design, and we recommend that future research either clearly separates measures of previous behaviour and manipulation of PEB1, for example by administering them at different time points, or employs a stronger intervention to manipulate PEB1. Stronger interventions should, however, be used cautiously when the aim is to study spillover, since extrinsic motivations can hinder spillover (Thomas et al., 2016; Yang et al., 2021).

Behavioural change takes time and longitudinal studies that have found changes have more often been carried out over months or even years (Stangherlin et al., 2023; Thøgersen et al., 2024). Practical restrictions precluded such a design in this study, but for further research into these phenomena we would recommend specifically targeting change that can happen in shorter periods of time with interventions or employing longer time intervals to allow spontaneous behavioural change to occur. Additionally, to allow for more causal clarity in longitudinal designs, we suggest measuring mediators at all time points, making it possible to more firmly establish whether the changes in a mediator are a consequence of PEB1.

Practical Implications

Based on these studies, we can conclude that simple interventions, such as choice defaults and descriptive norm messaging, have potential to increase donations to NGOs, but campaigners should not expect large effects of such interventions. Studies finding small or null effects of behavioural interventions, such as the present ones, can increase realism of expected effect sizes in field applications (e.g., online fundraising). Although we did not find much evidence of positive PEB spillover occurring, these studies also do not raise concerns for negative PEB spillover or single action bias resulting from easy PEB or simple interventions such as the ones studied here.

Conclusion

While this study did not yield definitive conclusions about the prevalence or magnitude of pro-environmental behavioural spillover, it nonetheless contributes meaningfully to the broader effort of theory development in this area. By adopting a novel approach of including multiple mediators in one study, we demonstrated the importance of integrating the disparate pieces of theory into one empirical model. Importantly, our findings underscore the relevance of environmental self-identity and concern as key psychological constructs that may influence both behavioural consistency and spillover. Ultimately, this work should be viewed as a proof of concept – laying the groundwork for further empirical investigations into pro-environmental behaviour change.

Supplemental Material

sj-docx-1-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental material, sj-docx-1-eab-10.1177_00139165261447970 for Mechanisms of Pro-Environmental Behavioural Spillover by Ragnheiður “Heather” Torfadóttir, Gloria Amaris, Christian Klöckner, John Thøgersen, Heather Barnes Truelove, Ellen van der Werff and Stepan Vesely in Environment and Behavior

Supplemental Material

sj-docx-2-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental material, sj-docx-2-eab-10.1177_00139165261447970 for Mechanisms of Pro-Environmental Behavioural Spillover by Ragnheiður “Heather” Torfadóttir, Gloria Amaris, Christian Klöckner, John Thøgersen, Heather Barnes Truelove, Ellen van der Werff and Stepan Vesely in Environment and Behavior

Supplemental Material

sj-docx-3-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental material, sj-docx-3-eab-10.1177_00139165261447970 for Mechanisms of Pro-Environmental Behavioural Spillover by Ragnheiður “Heather” Torfadóttir, Gloria Amaris, Christian Klöckner, John Thøgersen, Heather Barnes Truelove, Ellen van der Werff and Stepan Vesely in Environment and Behavior

Supplemental Material

sj-docx-4-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental material, sj-docx-4-eab-10.1177_00139165261447970 for Mechanisms of Pro-Environmental Behavioural Spillover by Ragnheiður “Heather” Torfadóttir, Gloria Amaris, Christian Klöckner, John Thøgersen, Heather Barnes Truelove, Ellen van der Werff and Stepan Vesely in Environment and Behavior

Supplemental Material

sj-docx-5-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental material, sj-docx-5-eab-10.1177_00139165261447970 for Mechanisms of Pro-Environmental Behavioural Spillover by Ragnheiður “Heather” Torfadóttir, Gloria Amaris, Christian Klöckner, John Thøgersen, Heather Barnes Truelove, Ellen van der Werff and Stepan Vesely in Environment and Behavior

Footnotes

ORCID iDs

Ragnheiður “Heather” Torfadóttir

Gloria Amaris

Christian Klöckner

John Thøgersen

Heather Barnes Truelove

Stepan Vesely

Author Contributions

Ragnheiður “Heather” Torfadóttir: Conceptualisation; methodology; investigation; data curation; formal analysis; writing – original draft, writing – review and editing; visualisation. Gloria Amaris: Methodology; investigation; writing – review and editing. Christian Klöckner: Methodology; writing – review and editing; supervision. John Thøgersen: Methodology; writing – review and editing. Heather Barnes Truelove: Methodology; writing – review and editing. Ellen van der Werff: Methodology; writing – review and editing. Stepan Vesely: Conceptualisation; Methodology; investigation; writing – review and editing; supervision; funding acquisition.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Norwegian Research Council. Project Number: 313642.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data will be made available on osf.io.

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Ragnheiður “Heather” Torfadóttir is a PhD candidate at the Department of Psychology at NTNU. In her PhD project she focuses on understanding the processes involved in pro-environmental behavioural spillover. She has co-authored work on experiencing climate consequences through virtual reality, typology of efficacy in environmental action, and applying COVID-19 communication lessons to climate change messaging.

Gloria Amaris, PhD, is a researcher specialising in mathematical modelling to understand human decision-making. Her research integrates principles from choice modelling, psychology, and environmental sciences, with a focus on sustainability, climate change, water and energy systems, and new markets and policy.

Christian Klöckner is a Professor in Social Psychology and Quantitative Methods at the Norwegian University of Science and Technology. He leads the research group for “Citizen, Environment, and Safety” which is focusing on researching individual, social and contextual drivers of environmental decisions (including food, energy, mobility, housing). His personal research interests are modelling of environmentally relevant decisions and behaviour in the social and structural context, innovative environmental communication methods, societal disruption and transition, and psychological trade-offs in sustainability issues.

John Thøgersen is a Professor of Economic Psychology at Aarhus University, Denmark, Department of Management. His research focuses on consumer willingness to reduce their climate footprint, responses to eco- and climate labelling, the establishment of sustainable lifestyles, and drivers and impediments for energy renovation and for buying organic food.

Heather Barnes Truelove is a Professor of Psychology and a joint faculty member in the Institute of Environmental Research and Education at the University of North Florida. Her research centres on the mechanisms underlying individual pro-environmental behaviour and policy support.

Ellen van der Werff is Associate Professor in Environmental Psychology at the University of Groningen. She studies factors influencing pro-environmental behaviour and interventions to change behaviour.

Stepan Vesely works as a senior researcher at the Department of Psychology at NTNU. He specialises in research on sustainable behaviour and its determinants and regularly publishes in top field journals, including Ecological Economics, Environment and Behaviour, Global Environmental Change, and Journal of Environmental Psychology.

References

Ammerman

B. A.

Serang

Jacobucci

Burke

T. A.

Alloy

L. B.

McCloskey

M. S.

(2018). Exploratory analysis of mediators of the relationship between childhood maltreatment and suicidal behavior. Journal of Adolescence, 69(1), 103–112. https://doi.org/10.1016/j.adolescence.2018.09.004

Bandura

(1982). Self-efficacy mechanism in human agency. American Psychologist, 37(2), 122–147.

Bem

D. J.

(1972). Self-perception theory. Advances in Experimental Social Psychology, 6, 1–62.

Benjamini

Yekutieli

(2001). The control of the false discovery rate in multiple testing under dependency. The Annals of Statistics, 29(4), 1165–1188. https://doi.org/10.1214/aos/1013699998

Bergquist

Nilsson

Schultz

W. P.

(2019). A meta-analysis of field-experiments using social norms to promote pro-environmental behaviors. Global Environmental Change, 59, 101941. https://doi.org/10.1016/j.gloenvcha.2019.101941

Blanken

Van De Ven

Zeelenberg

(2015). A meta-analytic review of moral licensing. Personality and Social Psychology Bulletin, 41(4), 540–558. https://doi.org/10.1177/0146167215572134

Burger

A. M.

Schuler

Eberling

(2022). Guilty pleasures: Moral licensing in climate-related behavior. Global Environmental Change, 72, 102415. https://doi.org/10.1016/j.gloenvcha.2021.102415

Carlsson

Jaime

Villegas

(2021). Behavioral spillover effects from a social information campaign. Journal of Environmental Economics and Management, 109, 102325. https://doi.org/10.1016/j.jeem.2020.102325

Carrico

A. R.

(2021). Climate change, behavior, and the possibility of spillover effects: Recent advances and future directions. Current Opinion in Behavioral Sciences, 42, 76–82. https://doi.org/10.1016/j.cobeha.2021.03.025

10.

Carrico

A. R.

Raimi

K. T.

Truelove

H. B.

Eby

(2018). Putting your money where your mouth is: An experimental test of pro-environmental spillover from reducing meat consumption to monetary donations. Environment and Behavior, 50(7), 723–748. https://doi.org/10.1177/0013916517713067

11.

Cheung

S. F.

(2023). semhelpinghands: Helper functions for structural equation modeling [Manual]. https://CRAN.R-project.org/package=semhelpinghands

12.

Conway

Peetz

(2012). When does feeling moral actually make you a better person? Conceptual abstraction moderates whether past moral deeds motivate consistency or compensatory behavior. Personality and Social Psychology Bulletin, 38(7), 907–919. https://doi.org/10.1177/0146167212442394

13.

Croson

Handy

Shang

(2009). Keeping up with the Joneses: The relationship of perceived descriptive social norms, social information, and charitable giving. Nonprofit Management and Leadership, 19(4), 467–489. https://doi.org/10.1002/nml.232

14.

Devezer

Sprott

D. E.

Spangenberg

E. R.

Czellar

(2014). Consumer well-being: Effects of subgoal failures and goal importance. Journal of Marketing, 78(2), 118–134. https://doi.org/10.1509/jm.11.0599

15.

Dietz

Gardner

G. T.

Gilligan

Stern

P. C.

Vandenbergh

M. P.

(2009). Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions. Proceedings of the National Academy of Sciences, 106(44), 18452–18456. https://doi.org/10.1073/pnas.0908738106

16.

Miliute-Plepiene

(2018). Behavioral spillovers from food-waste collection in Swedish municipalities. Journal of Environmental Economics and Management, 89, 168–186. https://doi.org/10.1016/j.jeem.2018.01.004

17.

Elf

Gatersleben

Christie

(2019). Facilitating positive spillover effects: New insights from a mixed-methods approach exploring factors enabling people to live more sustainable lifestyles. Frontiers in Psychology, 9, 2699. https://doi.org/10.3389/fpsyg.2018.02699

18.

Fishbach

Dhar

(2005). Goals as excuses or guides: The liberating effect of perceived goal progress on choice. Journal of Consumer Research, 32(3), 370–377. https://doi.org/10.1086/497548

19.

Galizzi

M. M.

Whitmarsh

(2019). How to measure behavioral spillovers: A methodological review and checklist. Frontiers in Psychology, 10, 342. https://doi.org/10.3389/fpsyg.2019.00342

20.

Geiger

S. J.

Brick

Nalborczyk

Bosshard

Jostmann

N. B.

(2021). More green than gray? Toward a sustainable overview of environmental spillover effects: A Bayesian meta-analysis. Journal of Environmental Psychology, 78, 101694. https://doi.org/10.1016/j.jenvp.2021.101694

21.

Geng

Cheng

Tang

Zhou

(2016). Can previous pro-environmental behaviours influence subsequent environmental behaviours? The licensing effect of pro-environmental behaviours. Journal of Pacific Rim Psychology, 10, e9. https://doi.org/10.1017/prp.2016.6

22.

Ghesla

Grieder

Schmitz

(2019). Nudge for good? Choice defaults and spillover effects. Frontiers in Psychology, 10, 178. https://doi.org/10.3389/fpsyg.2019.00178

23.

Goetz

Mayr

Schubert

(2024). One thing leads to another: Evidence on the scope and persistence of behavioral spillovers. Journal of Public Economics, 236, e105166. https://doi.org/10.1016/j.jpubeco.2024.105166

24.

Hansen

Marx

Weber

(2006). The role of climate perceptions, expectations, and forecasts in farmer decision making. International Research Institute for Climate Prediction. Final Report of an IRI Seed Grant Project.

25.

IPCC. (2021). Climate change 2021: The physical science basis. Cambridge University Press. https://doi.org/10.1017/9781009157896

26.

Jachimowicz

J. M.

Duncan

Weber

E. U.

Johnson

E. J.

(2019). When and why defaults influence decisions: A meta-analysis of default effects. Behavioural Public Policy, 3(02), 159–186. https://doi.org/10.1017/bpp.2018.43

27.

James

Witten

Hastie

Tibshirani

(2017). An Introduction to Statistical Learning: With Applications in R. Springer.

28.

Juhl

H. J.

Fenger

M. H. J.

Thøgersen

(2017). Will the consistent organic food consumer step forward? An empirical analysis. Journal of Consumer Research, 44(3), 519–535. https://doi.org/10.1093/jcr/ucx052

29.

Karmarkar

U. R.

Bollinger

(2015). BYOB: How bringing your own shopping bags leads to treating yourself and the environment. Journal of Marketing, 79(4), 1–15. https://doi.org/10.1509/jm.13.0228

30.

Lacasse

(2016). Don’t be satisfied, identify! Strengthening positive spillover by connecting pro-environmental behaviors to an “environmentalist” label. Journal of Environmental Psychology, 48, 149–158. https://doi.org/10.1016/j.jenvp.2016.09.006

31.

Lauren

Fielding

K. S.

Smith

Louis

W. R.

(2016). You did, so you can and you will: Self-efficacy as a mediator of spillover from easy to more difficult pro-environmental behaviour. Journal of Environmental Psychology, 48, 191–199. https://doi.org/10.1016/j.jenvp.2016.10.004

32.

Lauren

Smith

L. D. G.

Louis

W. R.

Dean

A. J.

(2017). Promoting spillover: How past behaviors increase environmental intentions by cueing self-perceptions. Environment and Behavior, 51(3), 235–258. https://doi.org/10.1177/0013916517740408

33.

Leiserowitz

(2006). Climate change risk perception and policy preferences: The role of affect, imagery, and values. Climatic Change, 77(1–2), 45–72. https://doi.org/10.1007/s10584-006-9059-9

34.

Jacobucci

Ammerman

B. A.

(2021). Tutorial on the use of the regsem package in R. Psych, 3(4), 579–592. https://doi.org/10.3390/psych3040038

35.

Lin

Y.-C.

Chang

C. A.

(2017). Exploring wasteful consumption. Journal of Environmental Psychology, 49, 106–111. https://doi.org/10.1016/j.jenvp.2017.01.001

36.

Maki

Carrico

A. R.

Raimi

K. T.

Truelove

H. B.

Araujo

Yeung

K. L.

(2019). Meta-analysis of pro-environmental behaviour spillover. Nature Sustainability, 2(4), 307–315. https://doi.org/10.1038/s41893-019-0263-9

37.

Manika

Antonetti

Papagiannidis

Guo

(2021). How pride triggered by pro-environmental technology adoption spills over into conservation behaviours: A social business application. Technological Forecasting and Social Change, 172, 121005. https://doi.org/10.1016/j.techfore.2021.121005

38.

Marczak

Wierzba

Zaremba

Kulesza

Szczypiński

Kossowski

Budziszewska

Michałowski

J. M.

Klöckner

C. A.

Marchewka

(2023). Beyond climate anxiety: Development and validation of the inventory of climate emotions (ICE): A measure of multiple emotions experienced in relation to climate change. Global Environmental Change, 83, 102764. https://doi.org/10.1016/j.gloenvcha.2023.102764

39.

Margetts

E. A.

Kashima

(2017). Spillover between pro-environmental behaviours: The role of resources and perceived similarity. Journal of Environmental Psychology, 49, 30–42. https://doi.org/10.1016/j.jenvp.2016.07.005

40.

Meijers

M. H. C.

Noordewier

M. K.

Verlegh

P. W. J.

Willems

Smit

E. G.

(2019). Paradoxical side effects of green advertising: How purchasing green products may instigate licensing effects for consumers with a weak environmental identity. International Journal of Advertising, 38(8), 1202–1223. https://doi.org/10.1080/02650487.2019.1607450

41.

Meijers

M. H. C.

Verlegh

P. W. J.

Noordewier

M. K.

Smit

E. G.

(2015). The dark side of donating: How donating may license environmentally unfriendly behavior. Social Influence, 10(4), 250–263. https://doi.org/10.1080/15534510.2015.1092468

42.

Nash

Whitmarsh

Capstick

Hargreaves

Poortinga

Thomas

Sautkina

Xenias

(2017). Climate-relevant behavioral spillover and the potential contribution of social practice theory. WIREs Climate Change, 8(6), 481. https://doi.org/10.1002/wcc.481

43.

Poortinga

Whitmarsh

Suffolk

(2013). The introduction of a single-use carrier bag charge in Wales: Attitude change and behavioural spillover effects. Journal of Environmental Psychology, 36, 240–247. https://doi.org/10.1016/j.jenvp.2013.09.001

44.

Rosseel

(2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02

45.

Serang

Jacobucci

Brimhall

K. C.

Grimm

K. J.

(2017). Exploratory mediation analysis via regularization. Structural Equation Modeling: A Multidisciplinary Journal, 24(5), 733–744. https://doi.org/10.1080/10705511.2017.1311775

46.

Sharpe

(2022). Behaving pro-environmentally and supporting environmental policy: The common motivational basis of pro-environmental actions [Doctoral dissertation, University of Groningen]. https://doi.org/10.33612/diss.222302549

47.

Shipley

N. J.

Van Riper

C. J.

(2022). Pride and guilt predict pro-environmental behavior: A meta-analysis of correlational and experimental evidence. Journal of Environmental Psychology, 79, 101753. https://doi.org/10.1016/j.jenvp.2021.101753

48.

Sintov

Geislar

White

L. V.

(2019). Cognitive accessibility as a new factor in proenvironmental spillover: Results from a field study of household food waste management. Environment and Behavior, 51(1), 50–80. https://doi.org/10.1177/0013916517735638

49.

Stangherlin

I. D. C.

Thøgersen

De Barcellos

M. D.

(2023). Behavioral spillover in the circular economy: The importance of consumer goals. Journal of Environmental Psychology, 91, 102123. https://doi.org/10.1016/j.jenvp.2023.102123

50.

Steinhorst

Klöckner

C. A.

Matthies

(2015). Saving electricity – For the money or the environment? Risks of limiting pro-environmental spillover when using monetary framing. Journal of Environmental Psychology, 43, 125–135. https://doi.org/10.1016/j.jenvp.2015.05.012

51.

Thøgersen

(2004). A cognitive dissonance interpretation of consistencies and inconsistencies in environmentally responsible behavior. Journal of Environmental Psychology, 24(1), 93–103. https://doi.org/10.1016/S0272-4944(03)00039-2

52.

Thøgersen

Crompton

(2009). Simple and painless? The limitations of spillover in environmental campaigning. Journal of Consumer Policy, 32(2), 141–163. https://doi.org/10.1007/s10603-009-9101-1

53.

Thøgersen

Ölander

(2003). Spillover of environment-friendly consumer behaviour. Journal of Environmental Psychology, 23(3), 225–236. https://doi.org/10.1016/S0272-4944(03)00018-5

54.

Thøgersen

Vatn

Aasen

(2024). The chicken or the egg? Spillover between private climate action and climate policy support. Journal of Environmental Psychology, 99, 102434. https://doi.org/10.1016/j.jenvp.2024.102434

55.

Thomas

G. O.

Poortinga

Sautkina

(2016). The Welsh Single-Use Carrier Bag Charge and behavioural spillover. Journal of Environmental Psychology, 47, 126–135. https://doi.org/10.1016/j.jenvp.2016.05.008

56.

Truelove

H. B.

Nugent

M. R.

(2020). Straw wars: Pro-environmental spillover following a guilt appeal. Journal of Environmental Psychology, 72, 101521. https://doi.org/10.1016/j.jenvp.2020.101521

57.

Truelove

H. B.

Carrico

A. R.

Weber

E. U.

Raimi

K. T.

Vandenbergh

M. P.

(2014). Positive and negative spillover of pro-environmental behavior: An integrative review and theoretical framework. Global Environmental Change, 29, 127–138. https://doi.org/10.1016/j.gloenvcha.2014.09.004

58.

Truelove

H. B.

Carrico

A. R.

Yeung

K. L.

Wolff

J. M.

(2021). Identity and guilt as mediators of pro-environmental spillover. Frontiers in Psychology, 12, 659483. https://doi.org/10.3389/fpsyg.2021.659483

59.

Truelove

H. B.

Yeung

K. L.

Carrico

A. R.

Gillis

A. J.

Raimi

K. T.

(2016). From plastic bottle recycling to policy support: An experimental test of pro-environmental spillover. Journal of Environmental Psychology, 46, 55–66. https://doi.org/10.1016/j.jenvp.2016.03.004

60.

Van der Werff

Steg

. (2018). Spillover benefits: Emphasizing different benefits of environmental behavior and its effects on spillover. Frontiers in Psychology, 9, 1–16.

61.

Van der Werff

Steg

Keizer

. (2013). It is a moral issue: The relationship between environmental self-identity, obligation-based intrinsic motivation and pro-environmental behaviour. Global Environmental Change, 23(5), 1258–1265. https://doi.org/10.1016/j.gloenvcha.2013.07.018

62.

Van der Werff

Steg

Keizer

. (2014). Follow the signal: When past pro-environmental actions signal who you are. Journal of Environmental Psychology, 40, 273–282. https://doi.org/10.1016/j.jenvp.2014.07.004

63.

Vesely

Klöckner

C. A.

Carrus

Chokrai

Fritsche

Masson

Panno

Tiberio

Udall

A. M.

(2022). Donations to renewable energy projects: The role of social norms and donor anonymity. Ecological Economics, 193, 107277. https://doi.org/10.1016/j.ecolecon.2021.107277

64.

Wang

Yang

Bailey

Wang

(2021). Positive spillover of consumers’ sustainable behaviors: The mediating role of self-determination need satisfaction. Journal of Cleaner Production, 317, 128436. https://doi.org/10.1016/j.jclepro.2021.128436

65.

Weber

E. U.

(1997). Perception and expectation of climate change: Precondition for economic and technological adaptation. In Bazerman

M. H.

Messick

D. M.

Tenbrunsel

A. E.

Wade-Benzoni

K. A.

(Eds.), Environment, ethics, and behavior: The psychology of environmental valuation and degradation (pp. 314–341). The New Lexington Press/Jossey-Bass Publishers.

66.

Werfel

S. H.

(2017). Household behaviour crowds out support for climate change policy when sufficient progress is perceived. Nature Climate Change, 7(7), 512–515. https://doi.org/10.1038/nclimate3316

67.

Wolstenholme

(2020). Two birds, one stone: The effectiveness of health and environmental messages to reduce meat consumption and encourage pro-environmental behavioral spillover. Frontiers in Psychology, 11, 14.

68.

Zhang

Ling

(2018). Pro-environmental spillover under environmental appeals and monetary incentives: Evidence from an intervention study on household waste separation. Journal of Environmental Psychology, 60, 27–33. https://doi.org/10.1016/j.jenvp.2018.10.003

69.

Yang

Wei

Cheng

(2021). Spillover of different regulatory policies for waste sorting: Potential influence on energy-saving policy acceptability. Waste Management, 125, 112–121. https://doi.org/10.1016/j.wasman.2021.02.008

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.06 MB

0.11 MB

0.05 MB

0.00 MB

0.09 MB

0.07 MB

Mechanisms of Pro-Environmental Behavioural Spillover

Abstract

Keywords

Introduction

Mechanisms Explaining Positive Spillover

Environmental Self-Efficacy

Environmental Self-Identity

Environmental Concern

Goal Importance

Mechanisms to Explain Negative Spillover

Moral Licensing

Contribution Ethic

Alleviation of Guilt

Single-Action Bias and Global Warming Worry

Combining Mediators

Current Research

Study 1

Method

Participants

Study Design and Procedure

Measures

Preregistered Hypotheses

Analysis

Results

Experimental Spillover Tests (Tests of H1–H4)

Behavioural Correlation (Tests of H5–H8)

Mediation Analysis (Tests of H1–8med1–8)

Exploratory Analysis of Multiple Mediators at Once

Exploratory Analysis of Correlations Between Individual Behaviour Items

Discussion

Study 2

Method

Participants

Study Design and Procedure

Measures

Preregistered Hypotheses

Analysis

Results

Discussion

General Discussion

Theoretical Implications

Methodological Implications

Default Interventions

Descriptive Norm Intervention

Limitations and Future Research Directions

Practical Implications

Conclusion

Supplemental Material

sj-docx-1-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental Material

sj-docx-2-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental Material

sj-docx-3-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental Material

sj-docx-4-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Supplemental Material

sj-docx-5-eab-10.1177_00139165261447970 – Supplemental material for Mechanisms of Pro-Environmental Behavioural Spillover

Footnotes

ORCID iDs

Author Contributions

Funding

Declaration of Conflicting Interests

Data Availability Statement

Supplemental Material

Notes

Author Biographies

References

Supplementary Material

Mediation Analysis (Tests of H1–8_med1–8)