Shifting horizons: Significant life events and pro-environmental behaviour change in early adulthood

Kaloyan Mitev; Lorraine Whitmarsh; Merideth Gattis; Nicholas Nash

doi:10.1371/journal.pclm.0000709

Abstract

Young people’s daily routines are especially malleable during significant life events, offering opportunities for pro-environmental behaviour (PEB) change. We investigated how PEBs shifted during two Moments of Change (MoCs), an exogenous disruption (COVID-19) and a biographical transition (starting university). We also looked at whether values and attitudes explained these shifts. We conducted two longitudinal studies with 16–24-year-olds. Study 1 (exogenous MoC) tracked behaviour across three waves during 2020 (n = 146) using multilevel latent growth models. Study 2 (biographical MoC) (n = 256) used paired-samples t-tests examining change across two time points, i.e., pre and post the start of university. Both studies used path-analytic structural equation models to test a values – attitudes – behaviour pathway and regressions to examine the self-activation hypothesis. In Study 1, we found positive changes to food waste and the consumption of animal products, and a negative change in environmental activism and active travel. In Study 2, we found positive changes in domestic PEBs, active travel, and the consumption of animal products, and negative changes in environmental activism and ethical consumption. Self-transcendence values positively predicted activism (Study 1) and domestic PEBs (Study 2) and related to lower animal-product consumption in both studies, while environmental attitudes mediated the link between self-transcendence and consumption only in Study 2. These findings suggest that targeted interventions timed to MoCs can leverage values and attitudes to support lower-impact diets and home practices. Structurally constrained domains (e.g., activism opportunities, infrastructure-dependent travel) may require contextual changes in tandem with MoCs to yield benefits to PEBs. Our study is one of the first to look at COVID-19 as a MoC and one of the first to examine the transition from school to university and its effects on multiple PEBs. Some limitations include reliance on self-report measures with retrospective baselines and short follow-up periods.

Citation: Mitev K, Whitmarsh L, Gattis M, Nash N (2025) Shifting horizons: Significant life events and pro-environmental behaviour change in early adulthood. PLOS Clim 4(10): e0000709. https://doi.org/10.1371/journal.pclm.0000709

Editor: Md Wahid Murad, University of South Australia, AUSTRALIA

Received: March 29, 2025; Accepted: September 1, 2025; Published: October 15, 2025

Copyright: © 2025 Mitev et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The data associated with our study has been deposited into the publicly available repository of the Open Science Framework. The data can be accessed here: https://doi.org/10.17605/OSF.IO/GJPDZ.

Funding: This work was supported by the European Research Council (ERC), under the project “Understanding and leveraging ‘moments of change’ for pro-environmental behaviour shifts” [grant number: 820235 to LW; KM; MG; NN] The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Climate change, caused by greenhouse gas emissions (GHGs) from human activities, is one of the most urgent and immediate problems challenging modern day societies [1]. Tackling climate change requires urgent, substantive change in societal patterns of consumption and the adoption of pro-environmental behaviours (PEBs), defined as everyday actions minimising their environmental impact or even benefiting the environment [2]. Many PEBs (e.g., purchasing, travel, etc.) include habitual actions, as well as more deliberate ones, e.g., buying appliances [3].

Once formed, habits can be difficult to break; however, they can be weakened during life events or moments of change (MoCs) which represent significant changes to one’s environment and provide opportunities to foster pro-environmental consumption patterns. MoCs can also affect people differently; for example, they may activate pre-existing values, such that people who value the environment are more likely to adopt PEBs during a MoC (i.e., ‘self-activation’). The present study investigates the effects of two MoCs, i.e., the COVID-19 pandemic and the start of university, on young people’s PEBs, values and attitudes.

Habits and moments of change

Much human behaviour is habitual and has its origins in pursuing a goal, for example to satisfy a need or attain a reward [4]. Habits are triggered directly by a stimulus in the environment, i.e., contextual cues [5]. These can include time of day, physical location, and preceding actions in a sequence [6].

When the context is stable, behaviours do not rely on conscious deliberation [7] as their performance is guided by context-driven habit processes, however a change in the contextual cues should result in attenuating habitual responses [8]. This is the basis of the Habit Discontinuity Hypothesis [9]. A recent systematic review on MoCs found that most of the studies in the domain tend to focus on biographical events (e.g., moving house; becoming a parent) and to a lesser extent on exogenous events (e.g., financial crashes, natural disasters), both of which result in significant changes to one’s environment [10].

Biographical MoCs usually affect the individual aspects of one’s life and are predominantly planned for, voluntary and preceded by some degree of preparation (although there are some exceptions, such as serious illness). Research has focused on relocation [11]; pregnancy [12]; retirement [13]; and changes in employment circumstances [14]; holiday breaks [15]; New Year’s resolutions [16]; or biographical live events in general [17] as drivers for pro-environmental behaviour change. For example, reduction in driving after relocation [18] and lower meat consumption after becoming a parent [19].

Exogenous MoCs are driven by societal change and are usually unplanned. Studies have found changes in energy conservation due to extreme climate events [20]; in travel behaviours due to the financial crisis [21,22]; and in food consumption due to experiencing a tsunami [23]. The effects of exogenous MoCs have received less attention, perhaps due to practical issues such as planning and data collection. This calls for further research to investigate these important gaps in the domain.

COVID-19 as a MoC.

One exogenous MoC is the Coronavirus (COVID-19) pandemic which started in 2020 in the UK and led to huge behavioural and lifestyle shifts because of lockdowns and other restrictions. The pandemic led to various restrictions which were introduced in March 2020 and went through several changes. For a detailed summary of these restrictions please see Fig 1. Mitigating climate change is one area of human life that COVID-19 seemed to influence in a rather positive way. In 2020, the UK registered a 9.5% decrease in carbon emissions [24] with the biggest changes coming from the transport sector. For example, cycling trips increased up to 200% in the period April – June 2020 compared to 2019, probably due to big decreases in train use (95%) and bus use (85%) [25]. People also reported walking more (38%), and the proportion of people switching from public transport use to walking increased by 31% [26].

Download:

Fig 1. Timeline of COVID-19 restrictions in the UK between March 2020 and December 2021.

https://doi.org/10.1371/journal.pclm.0000709.g001

There were also changes in consumption. Data from the Vegan Society in the UK reported that 21.5% of people reduced their meat consumption during COVID-19, while 15.4% reduced the amount of dairy/eggs in their diet [27]. CAST [28] reported that people threw away less food, with food waste levels remaining lower than pre-COVID levels as restrictions relaxed [29] (see also [30]). Organic food sales also increased by 12.5%, while sales of clothes made from natural materials increased by 9% [31]. Finally, it is important to note that young people were one of the most affected groups by the COVID-19 restrictions. Research has identified that for those aged between 15 and 25 the pandemic had an adverse impact on well-being and mental health [32]. This furthers the notion that COVID-19 served as an important MoC and should be explored in relation to the habit discontinuity hypothesis and its influence on young people’s PEBs. While research has focused on the effects of COVID-19 on certain PEBs, in the present article we explore this through the prism of MoC which is one of the major contributions of the current work.

Starting university as a MoC.

In addition to going through the COVID-19 pandemic, in 2020 more than half a million people started their undergraduate degree as full-time students at universities across the UK [33]. This makes the start of university a pivotal MoC for young people. However, most of the literature in the environmental domain has focused on investigating biographical events happening later in life (e.g., relocation, becoming a parent). There have been fewer studies on MoCs occurring earlier in life. This makes the transfer from school to university an understudied but interesting MoC. Studies in this domain usually explore students’ overall experiences at university and the general trend in their green actions (e.g., [34]), for example, how information about environmentalism and membership of campus-based green groups impact PEBs but without a specific focus on new students [35]. Conversely, when past research has looked at the start of university as a MoC, studies have usually concentrated on health behaviours such as dieting and exercising. For instance, a decrease in physical activity after starting university [36], a change in diet and subsequently weight gain [37], and a decrease in meat and milk consumption [38,39]. This makes the effect of starting university as a MoC on PEBs in an understudied phenomenon with important implications for tackling climate change.

Starting university is also accompanied by moving out of one’s family home and expanding one’s social circle [40]. The ages between 16 and 24, known as late adolescence, are an essential stage of a person’s life characterised by identity formation [41], exploration, and self-focus [42] as well as legal adulthood. Identity formation is important as young people’s values could have a crucial role in their pro-environmental norms and behaviours [43]. In addition, some of the behaviours young people adopt soon after they move out of the parental home and become independent might stick with them in the long term. Therefore, they could influence their active participation in society [44]. A key reason for this is that puberty is marked by extensive remodelling of the brain, particularly in areas involved in decision-making, risk assessment, and social interactions. Brain regions, such as the prefrontal cortex (responsible for executive functions) and the limbic system (associated with emotions and rewards), undergo significant maturation. The interplay between these regions during puberty can lead to heightened sensitivity to social cues and rewards, influencing behaviour choices [45]. Thus, the experiences and habits formed during this period can have lasting effects. For example, positive behaviours such as regular exercise, healthy eating, and constructive social interactions can lead to a healthier, more engaged adult life. Conversely, negative behaviours adopted during this time, such as substance abuse or social withdrawal, can also persist, potentially leading to detrimental effects on one’s health and social participation. Adolescents exposed to civic education and encouraged to participate in community activities are more likely to become engaged citizens. This involvement can include voting, volunteering, advocacy, e.g., through engaging in more pro-environmental behaviours later in life [46]. Finally, educational programs during late adolescence that integrate environmental themes have been found to help young people develop a strong environmental identity, which once formed, tends to influence behaviour in a consistent manner into adulthood [47]. Therefore, starting university might be one of the most impactful MoCs. The current article is one of the few with a specific focus on exploring the start of university as a MoC and its effects on young people’s PEBs. This makes the present research pivotal in enriching the theoretical framework surrounding MoCs.

In the present article we chose two theoretically distinct MoCs that young people commonly experienced in 2020: (i) the COVID-19 pandemic as an exogenous, society-wide disruption that reconfigured daily routines, and (ii) the transition to university as a biographical one that typically entails relocation, new schedules, and rebuilding everyday practices. Both events can reshape everyday life, therefore, creating brief periods when routines are disrupted, and new habits can be formed. Juxtaposing these MoCs permits a comparison between (a) change driven primarily by externally imposed constraints and affordances and (b) change driven by role transition and practice reassembly, both of which are central mechanisms in habit-discontinuity hypothesis. In addition, the selection is also methodologically advantageous: both events were highly prevalent in the target cohort in the year 2020, and were also salient and temporally well-defined, i.e., supporting clear measurement windows. Finally, they were also ecologically valid by capturing behaviours under real-world disruption rather than laboratory settings. While MoCs are shown to contribute to pro-environmental behaviour changes (e.g., [18]), there are also other important psychological constructs which might mediate the connection between MoCs and the change in PEBs (e.g., values, attitudes). For example, values, which are relatively stable guiding principles, are expected to be especially consequential as they provide higher-order priorities when routines are being reconsidered, while attitudes work at a more immediate level, translating those priorities into judgments about specific behaviours. We therefore explore the values – attitudes – behaviour pathway in more detail in relation to MoCs to paint a clearer picture of life events and their effects on habitual sustainable actions.

Values and attitudes

One important psychological construct that might mediate the relationship between a MoC and PEBs is human values which have been identified as lying along two dimensions: self-enhancement to self-transcendence, and openness to change to conservatism [48]. Self-transcendence values are characterised by a concern for the interests and welfare of others but are also considered important for promoting human growth and wellbeing [49]. People with higher levels of self-transcendence values are typically more environmentally-concerned and engage in more environmentally-friendly behaviours [50,51]. In addition, a meta-analysis concluded that people for whom self-transcendent values were more important than self-enhancement experienced greater obligation to act in an environmentally-friendly way [52]. Studies have also reported positive associations between self-transcendence and various categories of pro-environmental behaviours [53], environmental activism [54], energy conservation [55], recycling [56], and consumption [57,58].

In addition, there may be a link between values and MoCs. While values are thought to be quite stable over time [59], there is some evidence that a life-changing events (e.g., moving to a new culture or having a child) could make them malleable [60,61]. Indeed, going to university can change values [62], and different values might change according to the programme of study [60]. Similarly, self-transcendence values were especially important during COVID-19 as they were reported to be related to complying with restrictions and helping vulnerable people struggling with the pandemic [63]. This furthers the notion that human values could be vital for behaviour change during MoCs.

Interestingly, values could also play another role as MoCs may facilitate the expression of pre-existing values; this is known as the self-activation hypothesis [18]. It states that a MoC could prime pre-existing values leading to a change in behaviour. Hence, self-transcendence could be activated by the MoC and in turn influence behaviour change. For example, value activation would mean travel restrictions (context change) leading one to consider travelling more sustainably (e.g., cycling; [64]). Previous work has already explored values in relation to the self-activation hypothesis during MoCs [14] but predominantly during relocation. To the best of our knowledge, no research to date has explored the self-activation hypothesis during the start of university as a MoC or during an exogenous MoC such as COVID-19. Consequently, the results from the present article could enrich the MoCs literature and offer new perspectives on the role of values in changing PEBs while going through a MoC.

Another important psychological construct which might play a role in the relationship between MoCs and PEBs is attitudes, which could also have a role in shaping PEBs. Evidence shows that environmental attitudes are positively predicted by universalism which is one of the key characteristics of self-transcendence [65]. Attitudes are evaluative responses toward an object of thought embodying cognitive, behavioural, and affective responses [66]. Thus, holding attitudes towards a particular event or object can be connected to the emergence of behaviours [67]. Previous studies have shown an association between environmental attitudes and behaviours [68] and attitudes and behaviour change [69]. Those with stronger pro-environmental attitudes show higher engagement in PEBs [70,71]. Furthermore, research has shown that there is a relationship between values and PEBs, stipulated that mediating variables (e.g., attitudes) are needed for the reasonable prediction of environmental actions by self-transcendence [72]. Thus, it could be that the effect of values on behaviours is mediated by environmental attitudes [73]. This makes attitudes another vital element in the behavioural domain which should be explored in more detail when investigating MoCs.

Hence, given the well-established perspective in the field that environmental attitudes might serve as mediators guiding the relationship between self-transcendence values and PEBs [72,73], in the present article we test this through the prism of moments of change. We apply a mediation analysis which directly tests this proximal pathway, allowing us to examine whether attitudes statistically carry (part of) the association between values and PEBs during periods of routine disruption. Modelling attitudes as a mediator between self-transcendence values and PEBs therefore provides process evidence about how value orientations become enacted in concrete choices precisely when new behavioural patterns are formed.

Present research

The present article explores how habit discontinuity affects PEBs – more specifically sustainable consumer behaviours, and environmental activism – when going through an exogenous MoC (COVID-19) and a biographical MoC (starting university). The article also looks at the role of values and attitudes in the relationship between MoCs and a change in PEBs and tests the self-activation hypothesis.

Hypothesis 1. In line with previous studies in the domain of MoCs [18], we hypothesised that:
1. Hypothesis 1a: The COVID-19 pandemic will result in changes to young people’s PEBs.
2. Hypothesis 1b: The transition from school to university will lead to changes in young people’s PEBs.

These hypotheses are exploratory in their nature due to the lack of studies in the domain; therefore, it is difficult to predict the direction of changes.

Hypothesis 2. Consistent with past research on the positive influence of self-transcendence values on PEBs [53], we hypothesised that:
1. Hypothesis 2a: Self-transcendence would serve as a positive predictor of PEBs during COVID-19 as a MoC.
2. Hypothesis 2b: Self-transcendence would serve as a positive predictor of PEBs during the transition from school to university as a MoC.
Hypothesis 3. Based on the previously established mediating role of attitudes [73], we hypothesised that:
1. Hypothesis 3a: Environmental attitudes would serve as a mediator in the relationship between values and PEBs during the COVID-19 pandemic.
2. Hypothesis 3b: Environmental attitudes would serve as a mediator in the relationship between values and PEBs during the transition from school to university.
Hypothesis 4. Studies 1 and 2 further tested the self-activation hypothesis. Previous research has evaluated the effects of MoCs on activating values (e.g., [14]) however only during relocation; in the present research we test the self-activation hypothesis and hypothesise that:
1. Hypothesis 4a: Going through COVID-19 as a MoC (Study 1) could activate self-transcendence values which could then lead to positive pro-environmental behaviour change.
2. Hypothesis 4b: Going through the start of university as a MoCs (Study 2) could lead to the activation of self-transcendence values which could then lead to a positive change in pro-environmental behaviour.

Study 1

Materials and methods

Ethics.

The Ethics Committee at the Department of Psychology, University of Bath reviewed and approved the study (Reference number: 20-197).

Sample.

We recruited participants by advertising an online study on Facebook with boost posts targeting young people (aged 16–24) living in all areas of the United Kingdom. We collected data at two time points. During the first time point of data collection there were 829 people who took part in the UK survey. However, 468 were excluded as they completed less than 70% of the survey. Another three participants were excluded from the analysis as they only spent between two and six minutes completing the survey which was deemed an insufficient as the median time of completion was 10 minutes. Thus, 358 participants completed part one of the study. From the 358 people, 268 left details to be re-contacted. When contacted about the second part of the survey, 181 agreed to participate. However, 25 people were excluded as they completed less than 70% of the survey, and a further 10 people were excluded as the identifiers they had provided could not be matched with identifiers from the first round of data collection. Therefore, the final sample of the study consisted of 146 participants.

Design.

Our study had a longitudinal within-subjects correlational design investigating the relationship between COVID-19 and young people’s pro-environmental behaviours. The participants reported their pro-environmental behaviours at three time points. They reported on the first two time points in August 2020. First, they were asked to think about their PEBs before the coronavirus outbreak (i.e., March 2020) (time point one), thus providing retrospective answers. Second, they had to report on their PEBs ‘at the moment’ of completing the survey, data was collected between 20 and 27 August 2020 (time point two), therefore after experiencing the COVID-19 pandemic and its restrictive measures. The third time point of data collection was between 30 November and 4 December 2020, during which the participants once again reported their PEBs while experiencing further lockdown measures.

Procedure.

Participants were able to access the online survey on Qualtrics directly by clicking on the Facebook post they saw. After opening the survey, they were presented with an information sheet giving details about the study followed by a consent form. While completing the first survey, respondents were asked to provide contact details (e-mail address) and indicate if they were willing to take part in further studies. Participants were then recontacted three months later via e-mail containing a link to the second part of the study also hosted on Qualtrics. The survey once again included an information sheet and a consent form. Each part of the study was completed voluntary, and participants were entered into a prize draw for a shopping voucher worth £50 for each time point of data collection in which they participated. The participants were debriefed about the purposes of the study at the end of the second survey.

Materials.

Demographics: The participants completed demographic questions about their gender, age, type of place of residence, and employment status.

Pro-environmental behaviours: We used 20 items (3) to measure pro-environmental behaviours. The participants had to rate how often they engaged in different pro-environmental actions, e.g., “encourage other people to save energy”; “use a bike instead of using a car”, etc. The scale was from “Not at all” (1) to “At least once a day” (7). These were combined into four sub-scales using a principal component analysis (see S1 Text): environmental activism (seven items) (α = .81 in Study 1; α = .88 in Study 2); ethical consumption (three items) (α = .57 in Study 1; α = .74 in Study 2); travel behaviours (three items) (α = .41 in Study 1; α = .52 in Study 2); domestic behaviours (five items) (α = .52 in Study 1; α = .70 in Study 2). Due to the low reliability score of the domestic behaviours scale in Study 1, we excluded it from the analysis. Instead, we included a single question asking participants how much of the food they buy was thrown away for the three time-points on a scale from None (1) to More than 30% (5). This applied only to Study 1. Due to the low reliability scores of the scale for travel PEBs in both Study 1 and Study 2, we decided to focus only on active travel which was measured with two items on the use of a bike instead of a car and walking somewhere instead of using a car. This was applied to both Study 1 and Study 2.

Further to this list, the young people also had to estimate their weekly consumption of each of the following: red meat, white meat, fish, and dairy. The scale ranged from Never (1) to Every day (6). These were combined into one sub-scale (four items) (α = .54 in Study 1; α = .64 in Study 2).

Values: We measured personal values using the Portrait Value Questionnaire PVQ (Short-Form) [74] consisting of 21 statements at time point two and time point three. Each gave a description of a person, e.g., “It’s important to her to show her abilities. She wants people to admire what she does.” At the beginning of the questionnaire, participants filled in their gender; thus, they saw either a female, male or non-conforming version of the questionnaire. For each statement, they rated how much each person was like them on a scale from “Not at all like me” (1) to “Very much like me” (6). The sub-scale of this questionnaire has been widely used in previous research and is divided into four sub-scales: Openness to change; Self-enhancement; Conservation; and Self-transcendence. Therefore, in our research we used the established and tested sub-scales as indicated by the authors of the scale, but focused only on self-transcendence values (α = .57 in Study 1; α = .74 in Study 2) due to their connection to environmental behaviour [75]. We collected data on values during timepoint two and again during timepoint three.

Environmental attitudes: We used six items from the New Ecological Paradigm (NEP; [76]) to measure environmental attitudes (α = .44 in Study 1; α = 62. in Study 2) e.g., “Humans have a right to modify the natural environment to suit their needs”. The scale ranged from Strongly disagree (1) to Strongly agree (5). Despite a low reliability, we proceeded with analysis using the NEP as it is one of the most widely-used measures of environmental attitudes [77]. We collected data on environmental attitudes only during timepoint two.

Data analysis plan: We applied multilevel latent growth modelling to test changes in behaviours over time. We built five models, one for each set of pro-environmental behaviours. Next, to investigate the direct effect of self-transcendence values on each set of behaviours and the mediating effect of attitudes we applied path-analytic structural equation models with observed composites. We used R for all estimations. Given that our study is exploratory in nature, we decided not to control for the number of comparisons (Type 1 error) as this could limit our ability to uncover novel relationships or patterns. This allowed us to interpret the findings as exploratory and use them to generate hypotheses for future research.

Results

Descriptive statistics.

Table 1 reports demographic data for the sample. Table 2 reports means and standard deviation for all study variables. Sample sizes of at least 100 people are preferred for growth curve models [78].

Download:

Table 1. Demographic data.

https://doi.org/10.1371/journal.pclm.0000709.t001

Download:

Table 2. Mean scores and standard deviations for all study variables.

https://doi.org/10.1371/journal.pclm.0000709.t002

Latent growth models.

To track changes over time for each of the five categories of PEBs we applied latent growth models (LGMs). LGMs are better equipped to deal with measurement error and missing data and allow for the modelling of change within individuals over time by treating growth parameters (e.g., intercepts and slopes) as latent variables. This approach is beneficial for understanding the underlying developmental processes that drive changes in observed variables [79]. We specified the LGMs to capture the growth trajectories of individuals over time, and the latent variables included the initial status (intercept) and the rate of change (slope). We considered several indices of the goodness-of-fit. These were: Chi-square, which should not yield a significant result [80]; the root mean square of approximation (RMSEA) where values below.05 show close fit but values up to.08 are also acceptable [81]. The Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) – values above.95 show very good fit [82], values over.90 provide evidence of acceptable fit [81]; and the Standardised Root Mean Squared Residual (SRMR) which should be up to.05 for a close-fitting model and up to.10 for an acceptable fit [81].

Our first LGM examined changes in the levels of environmental activism. The model had a good fit: x2 (1) = 0.63, p = .426; RMSEA = .001; CFI = 1.00; TLI = 1.05; SRMR = .016, indicating that linear growth was appropriate for the data. The slope of the model was significant (μ_s = -0.11, p < .05, 95% CI [-.195, -.033]). Thus, our participants’ levels of environmental activism declined after the start of the pandemic (See Table 2).

The LGM exploring change in active travel behaviours had an acceptable fit: x2 (1) = 7.402, p < .05; RMSEA = .209; CFI = .960; TLI = .880; SRMR = .053. The slope of the model was significant (μ_s = -0.12, p < .05, 95% CI [-.220, -.028]). Thus, our participants engaged in less active travel after the onset of COVID-19 (See Table 2).

Our LGM looking at change over time of ethical consumption behaviours had a good fit: x2 (1) = 1.44, p = .230; RMSEA = .055; CFI = .997; TLI = .992; SRMR = .022, indicating that linear growth was appropriate for the data. The slope was not significant (μ_s = -0.06, p = .232, 95% CI [-.147,.036]). This revealed there was no change between the three time-points in participants’ ethical consumption behaviours (See Table 2).

We also estimated a LGM tracking changes in food waste levels which had an acceptable fit: x2 (1) =4.49, p < .05; RMSEA = .166; CFI = .938; TLI = .814; SRMR = .059. The slope (μ_s = -.100, p < .001, 95% CI [-.178, -.022]) was significant. The participants decreased their food waste levels, i.e., threw away less food after the start of the pandemic (See Table 2).

Our final LGM examined change over the first nine months of the pandemic in consumption of animal products. The model had a good fit: x2 (1) = 1.14, p = .285; RMSEA = .033; CFI = .999; TLI = .998; SRMR = .036, indicating linear growth was appropriate for the data. The model revealed the slope was significant (μ_s = -0.08, p < .05, 95% CI [-.128, -.022]). Our participants consumed less animal products after the start of the pandemic (See Table 2).

Path-analytic structural equation modelling: Mediation.

The next step in our analysis was to investigate the effect of self-transcendence and environmental attitudes on PEBs for which we observed a change over time. Thus, we applied path-analytic structural equation modelling (SEM) with observed composites including self-transcendence as predictor and environmental attitudes as mediator to environmental activism, consumption of animal products, active travel, and food waste in time point three. Models were fit in lavaan with a maximum-likelihood (ML) estimator and bias-corrected bootstrap CIs (10,000 resamples) for indirect effects. Our three-variable mediation models are saturated (just-identified), i.e., they reproduce the sample covariance matrix exactly. Therefore, global fit indices are trivial (χ²(0)=0; CFI/TLI = 1.00; RMSEA = 0; SRMR = 0), thus we focus on interpreting path estimates and R² rather than omnibus fit [83].

We collected data for people’s values twice – at time two (August 2020) and time three (December 2020). While values tend to be stable over time [59], to test this, we conducted a one-way ANOVA comparing participants’ scores for self-transcendence at time point two and time point three to test this. The results revealed there were no significant differences between the two time-points, F(1, 144) = 0.327, p = .568. Thus, we used the data for self-transcendence from time two as a predictor in our SEMs. The data on environmental attitudes was collected only once during time two.

Our first model included self-transcendence as a predictor, environmental attitudes as a mediator, and levels of environmental activism from time point three as a dependent variable. The results revealed self-transcendence predicted attitudes (b = 0.29, BootSE = 0.078, 95% BCa CI [0.115, 0.423], p = .001), and attitudes predicted activism (b = 0.25, BootSE = 0.142, 95% BCa CI [0.264, 0.822], p < .001). The direct effect of self-transcendence on activism remained significant (b = 0.27, BootSE = 0.117, 95% BCa CI [0.210, 0.678], p < .001). The indirect effect via attitudes was significant (b = 0.139, BootSE = 0.056, 95% BCa CI [0.055, 0.284], p = .013), yielding a total effect of b = 0.582 (BootSE = 0.115, 95% BCa CI [0.365, 0.816]). Variance explained was R² (attitudes) =.084 and R² (activism) =.178. (Standardised coefficients: a = .289, b = .273, c′ = .252; indirect = .079; total = .331). Therefore, values were connected with environmental attitudes which in turn were connected with engaging in environmental activism (Fig 2).

Download:

Fig 2. Significant mediation model with self-transcendence as predictor, environmental attitudes as mediator and environmental activism as dependent variable.

https://doi.org/10.1371/journal.pclm.0000709.g002

Our second model included self-transcendence as a predictor, environmental attitudes as a mediator, and levels of consumption of animal products from time point three as a dependent variable. We found self-transcendence predicted higher environmental attitudes (b = 0.31, BootSE = 0.086, 95% BCa CI [0.134, 0.469], p = .001), but attitudes did not predict animal-product consumption (b = 0.09, BootSE = 0.158, 95% BCa CI [−0.164, 0.447], p = .361). The direct effect of self-transcendence on consumption was negative and significant (b = −0.26, BootSE = 0.107, 95% BCa CI [−0.594, −0.171], p < .001). The indirect effect via attitudes was not significant (b = 0.03, BootSE = 0.048, 95% BCa CI [−0.042, 0.151], p = .404), yielding a total effect of b = −0.23 (BootSE = 0.103, 95% BCa CI [−0.543, −0.136], p = .001). Variance explained was R²_attitudes = .095 and R²_consumption = .062 (Fig 3).

Download:

Fig 3. Non-significant mediation model with self-transcendence as predictor, environmental attitudes as mediator and consumption of animal products as dependent variable.

https://doi.org/10.1371/journal.pclm.0000709.g003

Our models for active travel and food waste did not find direct effects of self-transcendence or environmental attitudes; there were also no significant mediations in the two models. See S2 Text for full details of these SEMs.

Given the modest reliability of the environmental attitudes measure in Study 1 (α = .44), the indirect effects should be viewed as conservative, because measurement error in the mediator preferentially attenuates both constituent paths of the mediation. Consequently, the significant mediation for environmental activism is likely a lower-bound estimate of the underlying mechanism, and the null mediations for animal-product consumption, active travel, and food waste should be interpreted cautiously as potentially underpowered rather than definitive evidence of no indirect effect. We therefore refrain from fine-grained comparisons of indirect path magnitudes across domains and note that stronger inferences would require a higher-reliability attitude measure or repeated attitude assessment.

Linear regressions.

The final step in our analysis was to establish whether the MoC activated participants’ self-transcendence values which in turn led to a change in engagement with PEBs, i.e., environmental activism, consumption of animal products, active travel, and food waste. To test this, we subtracted the score of time one from time two, and of time two from time three, for each of the three PEB variables. Next, we checked whether self-transcendence was a significant predictor of these new variables. All our linear regressions showed non-significant results, indicating that self-transcendence values did not influence engagement in PEBs (See Table 3).

Download:

Table 3. Results of linear regressions.

https://doi.org/10.1371/journal.pclm.0000709.t003

Discussion

The findings from Study 1 offered some insight into the relationship between the COVID-19 pandemic and changes in young people’s PEBs. We found mixed results; in the nine months after the start of COVID-19 our participants reported consuming fewer animal products, and throwing away less food; thus, these behaviours increased, indicating a positive influence linked to the pandemic. However, we also observed reductions in PEBs; our participants engaged in less environmental activism and less active travel. These findings support H1a. The results for the changes in PEBs are in line with previous work exploring changes in meat consumption and food waste during the pandemic (e.g., [27,28]), but contrast findings on the relationship between COVID-19 and active travel (e.g., [25]). The latter could be due to the characteristics of the sample, i.e., living in smaller villages or towns, and the availability of infrastructure (e.g., bike hire schemes).

Next, self-transcendence values had a direct effect only on the consumption of animal products and environmental activism in partial support of H2a. The latter relationship was also mediated by attitudes in partial support of H3a. These findings offer important insights on the different factors which might guide young people’s engagement with everyday green actions. For example, the effect of values and attitudes could be more pronounced with actions which do not require additional effort (e.g., signing an online petition) compared to actions which might rely on external factors such as infrastructure (e.g., active travelling). In addition, we did not find support for the self-activation hypothesis in Study 1 in contrast to H4a. Previous research has only explored this during biographical MoCs; the present findings suggest the pandemic restrictions could have been the guiding factor in people’s decision-making rather than their values.

In Study 1, we focused on an exogenous MoC characterised by its sudden occurrence – the COVID-19 pandemic – whereas in Study 2, we examined a biographical MoC, the transition from school to university, and investigated their effects on young people’s PEBs.

Study 2