The effect of consumer lockdown on the relationship between environmental beliefs and pro-environmental behaviors

This study investigates the disruptive effect of the COVID-19 lockdown on the relationship between environmental beliefs and pro-environmental behaviors. We used a survey conducted in the four major cities of Colombia before COVID-19 arrived in December 2019 and then repeated it in October 2020, after most of the major restrictions on mobility and economic activity had been lifted. The survey captured ecocentric and anthropocentric beliefs using the New Ecological Paradigm (NEP) and Pro Environmental Behaviors (PEBS) using 18 questions about energy and water savings, recycling, and sustainable purchasing. In the 2020 survey, we included a measure of the severity of the individual’s experience of disruption, including economic, family, and health-related aspects. Controlling for demographic variables, we analyzed (1) descriptive changes in NEP and PEBS, (2) the moderating effect of the disruption on the effect of NEP on PEBS, (3) the direct effect of disruption severity on PEBS, and (4) the moderating effect of severity on the effect of NEP on PEBS in 2020. We found that disruption caused by the lockdown crisis decreased anthropocentric beliefs and slightly increased ecocentric beliefs. This disruption also modified the effects of NEP on PEBS, thereby ameliorating the capacity of NEP to activate PEBS. The severity of experience moderated most of these effects. We discuss the implications of our results for the theory of environmental beliefs and pro-environmental behaviors and offer recommendations to reverse the potential negative effect of the COVID-19 crisis on the promotion of sustainable consumption.


Introduction
The COVID 19 pandemic has caused the most salient disruptive event in recent history, which manifests in different ways (Chiu et al 2020). The daily habits and routines of virtually all human beings were forced to change. Not only were the patterns of consumption exogenously affected, but some deeper structures of beliefs about the world may have also changed. As the crisis receded, the world wonders what truly changed in people's behavior and which behaviors will rebound now the pandemic is finally over. In particular, we are uncertain about the extent to which the disruption in consumption may have brought both threats and opportunities to promote sustainable consumption and pro-environmental behaviors (Trujillo 2022). During the lockdown, a wide range of consumption habits were altered. The effects of such changes on the environment have already been documented, including the reduction in greenhouse gas (GHG) emissions (Gillingham et al 2020) and changes in the patterns of plastic use and disposition due to safe packaging (Klemeš et al 2020).
Nonetheless, the experience of the disruption was long enough to open the door for structural changes in deeper environmental beliefs and attitudes. Furthermore, there may be changes in the way such beliefs are connected to Pro-Environmental Behaviors (PEBS). This is the focus of the present study. Based on theories of habit discontinuation and place attachment,we posit that the experience of the disruption not only altered restrictions were lifted. Following place attachment mostly to home and the habit discontinuity hypothesis, we hypothesize the following: 1. Behaviors related to energy and water use are influenced by the intensification of cleaning and health protection needs, causing an permanent increase in water and energy consumption (e.g., washing clothes more frequently with higher water temperatures). Moreover, because people became habituated to stay at home, the increase in energy and water use will continue to post-lockdown.
2. During confinement, people depended on home delivery to get groceries, food, and other essential products. Furthermore, health-promoting policies encourage extra-packaging for these products, which increases disposables at home. Place attachment might trigger heightened attention to waste increase, which combined with diminished opportunities for reducing it, could turn people's focus on improving recycling practices as an alternative. Our hypothesis was that recycling behavior may increase.
3. Behaviors associated with purchasing depended on the offerings available on e-commerce channels and were constrained by the mobility restrictions, time, and distance required to reach local shops. These constraints may relegate sustainability criteria to second place in people's decision-making processes, suggesting that using sustainable criteria for purchases may decrease.

Environmental beliefs and pro-environmental behaviors
Once expectations regarding specific changes in PEBS have been posited, we now turn our attention to the role of beliefs. Environmental-related beliefs have been firmly established as relevant originators of PEBS. From general models of behavior such as the Theory of Planned Behavior (Ajzen 1991) to more specific models of proenvironmental behavior, such as the Value Belief Norm Model (VBN) (Stern 2000), beliefs play a key role in the determination of both behavioral intentions and behaviors. Beliefs about the relationship between humans and nature are particularly pertinent. The New Environmental Paradigm (Dunlap et al 2000), both as a concept and as a measure, has been used to incorporate specific human-nature beliefs into the prediction of different types of pro-environmental behaviors (Stern 2000). The repertoire of PEBSs related to NEP is very wide. This includes the acceptance of CO2 reduction policies (Steg, Dreijerink and Abrahamse 2005) and indirect behaviors, such as activism and consumption avoidance (Ghazali et al 2019). NEP is also related to PEBS classified as either impact or intent, as well as media consumption, regarding global warming (Huang 2016). It predicts pro-environmental behavioral intentions when incorporated as a belief component within the TPB (Fielding, McDonald and Louis 2008). NEP is also correlated with other constructs that affect PEBS, such as attitudes towards recycling and donating toward environmental causes (de Groot and Steg 2008), altruistic tendencies, and the perception of environmental risks (Slimak and Dietz 2006). The NEP is appropriate for capturing beliefs about anthropocentrism and ecocentrism (Casey and Scott 2006), which are of particular relevance to this work. Anthropocentrism refers to the belief that nature exists primarily for human use and has no inherent value of its own (Dunlap et al 2000). Ecocentrism refers to beliefs in which people are the only participants in an environment (Fornara et al 2020). Thus, changes in PEBS may also be the result of modifications in anthropocentric and ecocentric beliefs triggered by consumption disruption. We will now examine this possibility.

Consumption disruption and the beliefs-PEBS relationship
Despite being a health crisis, COVID-19´s global nature may affect other beliefs related to global phenomena such as the climate crisis. For instance, people may be driven to reconsider deep views of the interaction between humans and nature. Although these worldviews could be considered solid and stable, another approach points out that such views are affected by social structures (Stern et al 1995). The global nature of COVID-19 disruption, coupled with its severity, may have affected the social structure because it changed how people work, study, travel, and consume. Furthermore, disruption affected social interactions, imbuing them with fear and anxiety. Therefore, we posit that global consumption disruption may have been strong enough to shake those beliefs. Since the crisis submitted individuals to the evolution of a disease, the perception of the interaction between humans and nature should have shifted, disempowering humans. Consequently, we believe that COVID-19, as a disruptive event, could cause anthropocentric beliefs to decrease and ecocentric beliefs to increase. Up to this point, there is a basis for anticipating the direction and nature of changes in both PEBS and beliefs. However, we consider our questions further. We also investigate whether the disruption caused by the Covid lockdowns not only affects absolute beliefs, but also how beliefs influence PEBS. Verplanken and Roy claim through their habit discontinuity hypothesis (Verplanken and Roy 2016) that life course interruptions create a window of openness to new information and change of mindset within three months after the disruptive event, which is conducive to permanent behavioral changes. The Covid lockdowns surpassed these estimates. Thus, there could be an alteration in beliefs, leading to pondering and reconsidering individuals' thoughts and views, which may also affect the relationship between these beliefs and behaviors. Hence, we expect that the ways anthropocentric and ecocentric beliefs influence PEBS may change. Nonetheless, it is difficult to establish the nature of these changes. There is an implicit assumption in the literature regarding the direction of the effects of anthropocentric and ecocentric beliefs on PEBS. The influence of the former should be negative and the influence of the latter should be positive. However, it has been acknowledged that NEP may capture more than one dimension (Dunlap et al 2000), and an ample meta-analysis of its use has confirmed the possible multidimensionality of NEP (Hawcroft and Milfont 2010), hence, anthropocentrism and ecocentrism may be orthogonal instead of the ends of a single construct. This opens many possibilities in the way NEP affects PEBS, which, in turn, could influence the extent to which the context (and therefore the disruption) may change the structure of the Beliefs-PEBS relationship. Accordingly, we investigated possible changes in the influence of beliefs on PEBS. This may have triggered a re-evaluation of attitudes and beliefs about the world and led people to devise new ways of living (Bamberg 2006, Padilla-Walker et al 2015. The latter could potentially modify the relationship between environmental beliefs (anthropocentric or ecocentric) and PEBS.

Method
To investigate our research questions, we conducted an observational study based on a survey of several aspects of sustainable consumption that was first implemented in December 2019, before the COVID-19 crisis started, and then repeated in October 2020, when the majority of the lockdown and restrictive measures to economic activities were relaxed in Colombia (i.e., most of the consumption disruption was over) (see appendix A for a chronological summary of the events and restrictions in Colombia). In this way, we used two cross-sectional datasets taken from the same population but different samples. All research protocols followed the principles of the Declaration of Helsinki and the ethical standards of the authors' institution. The surveys were anonymous, and all participants included in the sample provided informed consent prior to answering the survey.

Sample and procedure
The 2019 survey was applied to a representative sample of 800 people from four major cities in Colombia (Bogota, Barranquilla, Cali, and Medellin), representing approximately 40% of the Colombian population. This sample had a 3.49% confidence interval with a 95% confidence level. A market research firm conducted the fieldwork through face-to-face interviews. The sampling procedures involved randomized geo-referenced starting points covering all cities. We targeted and achieved a 50% male-female balance and a representative sample of socioeconomic strata. Nineteen percent belonged to low-income areas, 65% to middle-income areas, and 16% to upper-middle and affluent strata. Sixty percent had completed high school, 35% had a college or technical education, and 5% had a post-graduate degree.
The 2020 survey was administered to a sample of 400 people in the same cities following the same sampling procedures and criteria. The same market research firm conducted the fieldwork. This sample offers a 4.9% confidence interval with a 95% confidence level. Again, we achieved a balance of 50% between men and women. Eighteen% belonged to low-income areas, 65% to middle-income areas, and 17% to upper-middle and affluent strata. Regarding education, 62% had completed high school, 30% had college or technical education, and 8% had postgraduate education. The following table (Table 1) shows the additional sociodemographic characteristics.

Measures
We measured pro-environmental behaviors using items adapted from Thøgersen and Olander's (2002) inventory of self-reported environmentally friendly behaviors. To the best of our knowledge, these questions have not been previously used in similar countries. We used eighteen 7-point questions. The questionnaire included seven items on energy and water use habits, six on recycling habits, and five purchase-related items (see appendix B for a full list of items). These three categories of PEBs are appropriate because they capture highly habitual behavior at participant's home, which make them closely related to our theoretical background (i.e., habit discontinuation and place attachment). The range of possible PEBs is obviously much wider, so we will discuss the limitations of focusing on this set of behaviors later in the final discussion. The seven points of the scale asked participants to report behavior frequency, ranging from 'never' to 'always'. The market research firm conducted a pilot study of these questions to ensure clarity and comprehension. The pilot data were not included in the final sample. As explained above, we used the revised New Ecological Paradigm (NEP) (Dunlap et al 2000) to measure environment-related beliefs, which is composed of 15 agree-disagree questions (appendix C contains all questions). The NEP questionnaire has been used in various regions and emerging countries such as Latin  (Gooch 1995). All these antecedents ensure the applicability of the questionnaire in the Colombian context. Furthermore, NEP questions were included in the pilot study to reinforce their appropriateness. To keep the questionnaire simple and short, we asked 15 questions in two levels (agree/disagree), instead of using five levels. This simplification does not reduce the reliability of the scale (Alwin and Krosnick 1991) and allows us to capture the main direction of beliefs.
Our design takes advantage of the timing of the two surveys, so we are certain that all participants were affected by lockdowns similarly. There is no way to have a control group, not confined, to calculate treatment effects. Therefore, to control for potential confounds we also included several sociodemographic covariates that may affect PEBS, such as age, sex, education level, household income, household size (number of people living in the same house), and home size (in square meters). For the 2020 survey, we added a measure of the severity of an individual's COVID-19 experience. We used four 7-point questions that inquired about economic impact, family loss, personal physical health, and personal mental health, ranging from 'has negatively affected you' to 'has positively affected you' (see appendix D). Later, in the discussion section we address the limitations arising from attributing the results to the lockdown situation without a comparison to a control group.

Descriptive
We first calculated the scores for the three categories of pro-environmental behaviors that we measured: energy and water savings, recycling, and sustainable purchasing. We averaged the items of each category, which yields a score within the 1-7 scale where 7 expresses the most sustainable behavior, that is, the individual 'always' performs all the behaviors in the category. We also calculated a combined measure of PEBS by averaging the scores of the three categories. Thus, the overall pro-environmental behavior score may also take values between 1 and 7. Table 2 contains the mean values of the three categories and the aggregated values for the entire sample (i.e., T1 and T2). In the following section, we analyze the changes before and after post lockdown, controlling for relevant factors, to test Hypotheses 1, 2, and 3.
The aggregation of different PEBS into categories may be subject to debate. Can specific behaviors constitute latent aggregate behavioral categories? Similar behaviors may be observed, such as the separation of organic waste and recycling of plastic bottles. Furthermore, some seemingly less related behaviors, (e.g. as using reusable bags and recycling) may even generate spillover effects (Arias and Trujillo 2020). However, other behaviors that fall into the same category (e.g., sustainable purchasing), such as fuel consumption when buying a car and avoiding excess packaging, may be uncorrelated. We conducted a Confirmatory Factor Analysis to test this notion and our choice of PEBS categories to reveal the behavioral patterns in our data. (See the detailed results in appendix E). We found that 18 of the 16 measured behaviors were significantly loaded within the expected category. There were two exceptions: setting the water temperature low when doing laundry was not correlated with other energy/water-saving behaviors and reducing the use of detergents did not correlate with other sustainable purchasing behaviors. We retained these questions in the analysis because they still add to the total PEBS score.
Regarding the severity of the individual experience of the crisis in the 2020 sample, our 4-item measure proved to be reliable (alpha = 0.84). The average effect was 3.85 (S.D = 1.42, min = 1, max = 7). We also analyzed ecocentric and anthropocentric beliefs using new ecological paradigm questions. Table 3 shows the percentage of positive responses (agree/disagree) item-by-item, comparing T1 and T2. Significant changes were observed in 10 of the 15 items. The most salient ones were a reduction from 60% to 40% in the belief that humans were destined to rule over nature, followed by a 54% to 41% decrease in the belief that nature´s equilibrium was strong enough to face the impact of modern industrialized nations. In addition, there was a 54% to 41% reduction in the percentage of people agreeing that the current ecological crisis had been exaggerated.
Overall, the pattern of changes in beliefs measured by the NEP indicates a decrease in anthropocentric items. Based on the recommendations made by Dunlap et al (2000) in the development of the revised scale of NEP, which we used in this study, it is advisable to determine the sample-specific factor structure of the scale, which should reveal anthropocentric and ecocentric dimensions, , accordingly, we conducted a factor analysis of the NEP for both samples and found a clear two-dimensional solution. The two underlying unrotated factors closely match the two main emphases discussed by Dunlap et al (2000): an anthropocentric, human-dominant, and  ecocentric perspective on the relationship between humans and nature (See appendix F for details of these factor analyses) 3 . Consequently, for all subsequent examinations of our data, which focus on the effects of NEP on PEBS, we used two constructed variables: anthropocentrism and ecocentrism. Subsequently, we tested the hypothesized aggregated changes in ecocentrism and anthropocentrism before and after lockdowns. Note that based on the factor analyses and the theory behind the construction of the scale, we assume that these dimensions are orthogonal. Anthropocentrism and ecocentrism should not be understood as extremes of a continuum.

Effect of lockdown on PEBS and beliefs
The first step in the analysis is to investigate the effects of lockdowns on PEBS and Beliefs. For this purpose, we conducted two sets of linear regressions. In the first group, we estimated one model for each category of PEBS and one for the aggregated as dependent variables. In the second group, we estimated one regression each for  anthropocentrism and ecocentrism. In all models, we included a dummy variable to capture the effect of pre lockdown and post lockdown observations on the dependent variables (i.e., T1 and T2) and controlled for demographic variables. Table 4 contains the regressions for PEBS, and table 5 the regression for beliefs. There were significant changes in PEBS from T1 to T2, which were positive for recycling but negative for energy and water savings and sustainable purchasing. This mutual cancellation of changes in the PEBS categories produced a non-significant change in the aggregated PEBS scores. These results support Hypotheses 1, 2 and 3, regarding disruption affecting PEBS. Table 4 shows the regression results.
Regarding the effect of lockdown on beliefs, we calculated an aggregate score for ecocentrism (EC) and anthropocentrism (AC) following the item classification shown in table 3, which was also consistent with the factor analysis performed on the answers to the NEP questionnaire. Six items capture anthropocentrism and nine items capture ecocentrism. The average scores were 51% for AC and 83% for EC, indicating that our sample was much more ecocentric. As expected, the Pearson correlation coefficient between EC and AC was only 8.1%. We conducted linear regressions on AC and EC as dependent variables, using the dummy variable for T1 and T2 and the same controls. We found no significant effects of T1/T2 on AC or EC. This means that there were changes in some of the items, as shown in table 3, however, in the aggregate score, the magnitude of EC and AC beliefs remained stable. These results partially support our expectation that disruptions alter absolute beliefs. The changes were very subtle and item-specific, as previously shown.

Effect of beliefs on PEBS
As already explained, this study also seeks to explore potential changes in the way beliefs and PEBS are related before and after consumption restrictions. As in the previous analyses, we looked at each category of PEBS separately and then at the combined measure of PEBS. We used pooled data from before and after lockdowns to estimate the main effects of both EC and AC on PEBS using linear regressions. Based on these results, we estimated conditional linear regressions to assess and probe the moderating effect of a dummy variable for T1 (pre) and T2 (post) on the effect of beliefs on PEBS. Such moderation analysis allowed us to test whether the link between beliefs and PEBS changed with the experience of lockdowns. We first estimated four main effect models: One for each PEB (energy and water savings, recycling, and sustainable purchasing), and one for the aggregated PEBS score. AC and EC were included simultaneously as Independent Variables. We also controlled for age, education, gender, household income, house size, and the number of people living in each household. The results are presented in table 5. We found that EC was positively related to recycling, sustainable purchasing, and aggregated PEBS, but surprisingly, it was negatively related to energy and water savings. In contrast, AC was negatively correlated with recycling and aggregated PEBS. These results are expected, according to extant knowledge, but we want to highlight that EC seems to have a larger influence on PEBS.
Based on these results, we tested the moderation effect of disruption on the effects of EC and AC on PEBS. We estimate conditional linear regressions using the PROCESS macro in SPSS for instances where we found a significant main effect of either EC or AC on PEBS. Therefore, we conducted six regressions (four using EC as Independent Variable, and two using AC). Lockdown is included as an Independent Variable and  operationalized once again as a dummy variable that captures T1/T2 data. We also included interaction terms between T1/T2 and EC or AC. We used the same control variables in this study. We found that the negative effect of AC on both recycling and aggregated PEBS increased slightly from T1 to T2. Meanwhile, the positive effect of EC on sustainable purchasing and aggregate PEBS became non-significant in T2. Virtually no change was observed in the effect of EC on recycling, and the puzzling negative effect of EC on energy and water savings increased from T1 to T2. Appendix G contains the regression results and figure 1 shows the probing of the moderations. We probed all the analyzed moderation regardless of the significance of the interaction terms in the regressions to show a complete analysis of all the relationships. These results suggest that the attitudebehavior gap increases after consumption disruption (COVID-19). We explore this finding in detail in the Discussion section. Given that at T2, EC only had a surprising negative effect on energy and water savings, and AC showed a strong negative effect on total PEBS, mostly because of its effects on recycling, we explored whether the perceived severity of the disruption moderated these two effects. We then continued the test using only the data from T2. To this end, we used the SPSS Process Plug (Hayes 2017) to run a new set of conditional regressions and probe the moderation effects.
We found clear moderating effects of the severity of experience. The negative effect of ecocentrism on energy and water saving is contingent on the experience of being moderate or highly severe. The negative effect of anthropocentrism on recycling was contingent on the severity being low or moderate, whereas severity did not moderate the positive effect of ecocentrism on recycling. Finally, we estimated the moderation model for the negative effect of anthropocentrism on combined PEBS and found that it was contingent on the experienced severity to be low or moderate. For high severity, there was no effect, whichsoever. Tables 6 and 7 contain the regressions with the moderation analysis, and figure 2 shows the moderating effect of experienced severity on the effect of anthropocentrism on combined PEBS.

Discussion
Our first aim was to test whether the consumption disruption during Covid-19 lockdowns affected Proenvironmental behaviors (PEBs) in a predictable way. We found significant changes in PEBS from T1 (prelockdown) to T2 (post-lockdown) in the way we hypothesized. Our results support the idea that context change leads to new choices because individuals seem to be more attentive and deliberate in adapting to the new context characteristics (Verplanken and Wood 2006). Based on the discontinuity theory, we hypothesized that both increase and decrease PEBS. Recycling improved, but the use of resources (i.e., energy and water) increased, and sustainable shopping decreased. As Verplanken et al (2008) stated, when the context changes, behavior must be renegotiated. We interpret those lockdowns caused renegotiation to be affected by two aspects. First, the saliency of place attachment and the fact of being locked down favors better recycling but unsustainable shopping. Second, a strong motivation towards self-protection and augmented risk avoidance may lead to unsustainable behaviors, such as overusing energy and water. This is consistent with Protection Motivation Theory (PMT) (Tchetchik et al 2021). Furthermore, habit discontinuation may cause behavioral effects beyond the lockdown period.
We also explored the changes in eco-centric and anthropocentric beliefs. Although we expected significant changes in the strengths of both beliefs, the results did not show such changes. We found subtle and itemspecific reductions in a few anthropocentric items, but not in the aggregated scores. We found that people felt disempowered by nature and were more aware of the ecological crisis. This finding implies that disruption may engender a lower perception of human dominance over nature. People seemed to have developed a humble view of humankind in front of nature. Nonetheless, the lack of a significant change in aggregated scores is consistent with literature that highlights the inherent stability of empowerment-related beliefs (Antonetti and Maklan 2014). In our case, general views of the relationship between humans and nature remain.
Despite the stable nature of AC and EC beliefs during the pandemic, we found that the disruption generated by the COVID-19 lockdowns affected how these beliefs influence PEBS, mostly by ameliorating the positive effect of eco-centrism. This is the bottom line of the contribution of this study. As expected, AC beliefs were negatively correlated with recycling and aggregated PEBS. This finding is consistent with the proposition that human dominance over nature could encourage a 'do nothing' approach to the environment (Zhao et al 2018) with AC being a maladaptive belief concerning pro-environmental behavior (Milfont et al 2013). This effect remained after the disruption and even slightly increased. Thus, AC beliefs drive people to engage in inaction and self-serving behaviors. Conversely, as mentioned above, the effects of EC beliefs on PEBs were significantly altered by the lockdowns. The moderation models show that at pre-lockdown, ecocentrism has a positive effect on most of the sustainable behaviors under study, which is consistent with the literature (Vining and Ebreo 1992, Scott and Willits 1994, Stern et al 1995, Schultz and Oskamp 1996, Blake et al 1997, Roberts and Bacon, 1997, Tarrant and Cordell 1997, Schultz and Zelezny 1998, Ebreo et al 1999, O'Connor et al 1999. However, after post lockdown, we found that the positive effect of EC on sustainable purchasing and aggregated PEBS disappears. These results indicate that the ecological worldview somehow loses power over PEBS even when there is a slight increase in EC beliefs.
One of the most significant challenges in promoting sustainable lifestyles and consumption is the reduction in the attitude-behavior gap. In other words, people do not seem to act upon positive attitudes and beliefs about the environment (Peattie 2010, Ceglia et al 2015, Hanss et al 2016, Nguyen et al 2019. Our results imply that the attitude-behavior gap increased after lockdowns, rendering the effects of EC statistically insignificant. This could be the result of other factors at play, such as sticky negative emotions, perceptions of threat, and coping appraisals, all of which are triggered by the health crisis. In other words, the effects of ecocentrism may have been offset by short-term concerns and fears generated around the virus, which persisted even after the lockdowns were lifted. This implies that severe consumption disruptions may trigger short-term adaptive consumption behaviors that override belief-based actions. Engendered habit discontinuation may be a determining factor in the extension of such behaviors beyond lockdowns. Another aspect that contributes to shaping behavioral patterns is emotional. The negative emotions experienced during confinement may trigger emotional self-regulatory effects aimed at reducing such emotions through immediate gratification (Trujillo 2022). Therefore, a reorder in consumption priorities during and after the crisis could emerge, considering short-term satisfaction rather than sustainability criteria. This short-term satisfaction may explain why we found unsustainable behaviors related to the overuse of resources and shopping during the pandemic, as hypothesized. Our results suggest that such behavioral drivers persist once the main crisis is over. The duration of such behavioral patterns remains an open question.
Besides satisfaction, self-protection and risk avoidance could also prevail during lockdowns, leading to more safety, convenience, hygiene-related behaviors, and less sustainable behaviors. For instance, our findings show that although recycling increased, sustainable purchasing and energy-and water-saving behaviors decreased. This may be consistent with Protection Motivation Theory (PMT), which describes behavioral changes that stem from the need to protect oneself in light of a threatening situation (Tchetchik et al 2021). Under this theory, people´s behavior is the outcome of balancing different risks and potential benefits (Janmaimool 2017, Tchetchik et al 2021. Thus, although environmental issues such as climate change are a threat to people, during the COVID-19 disruption and several months afterwards, the pandemic effects posed a more significant and latent threat (i.e., threat appraisal). Consequently, individuals' decisions to engage in certain behaviors may be made based on their motivation to protect themselves, disengage eco-centric beliefs, and prioritize riskmitigating behaviors. In fact, our results show that the negative (and puzzling) positive effect of EC beliefs on the use of water and energy was moderated by the severity of the crisis, such that it was greater for people who experienced moderate and high levels of difficulties during COVID-19. Thus, people genuinely threatened by the disrupted event could somehow ignore their EC beliefs, leaving room for threat and coping appraisals to guide their behaviors. Following PMT, such appraisal activates beliefs regarding self-efficacy and collective response efficacy to face threats (Tchetchik et al 2021). In addition, consumption range was radically restricted during lockdowns, which may engender a sense of disempowerment and lack of control, which may impede pro-environmental behaviors (Trujillo 2022). The decrease in some items related to anthropocentric beliefs may be a cue for such loss of empowerment (e.g., a reduction of 20% in the belief that humans were destined to rule over nature). That is, people may feel the power of nature, but this leads to inaction and self-serving behaviors rather than pro-environmental behaviors.
Lockdowns have also altered the link between anthropocentrism and PEBS. We found a negative relationship between AC beliefs and PEBS only for those who perceived and experienced low levels of COVID severity. It seems that when the disruption did not bring severe negative consequences, AC beliefs prevented individuals from behaving sustainably, as noted in the previous literature (Zhao et al 2018, López andCuervo-Arango 2008) . Conversely, when people faced severe negative consequences due to the pandemic, other beliefs, perceptions, and appraisals became important in the consumption behavior patterns, as we have already explained.
Regarding the limitations of our study, we want to highlight two salient aspects. First, we are using observational data, using a sample people who were fully affected by lockdowns. It was not possible to have a control group. This precludes the possibility to test our research questions in a more robust way (i.e. calculating treatment effects). Nonetheless, to ameliorate this problem, we included in our measures and estimates, several potential confounding factors that could also affect beliefs and PEBs and controlled for them statistically. To the best of our knowledge, there was no other situation that could have influenced our dependent variables. Second, we chose a limited set of PEBs to conduct our analysis. Care should be taken when generalizing our results to a wider range of PEBs, in particular those that are not habitual and are unrelated to everyday life, such as traveling or buying an electric car. It is not clear how habit discontinuation and place attachment could influence them, hence our hypothesis are not directly applicable. In addition, we measured a range of behaviors whose direct impact on GHE is hard to determine (For some estimations, see Ivanova et al 2020) and we acknowledge the importance of research on impactful sustainable behaviors (Geiger et al 2018). Following their classification, we included behaviors in the three states of consumption, namely, acquisition, usage and disposal. Moreover, PBEs may be connected through spillover effects (Arias and Trujillo 2020), therefore, the influence on some of the PEBs we studied in this paper may favoring the adoption of others trough spillover, enhancing the aggregated impact.

Conclusions, implications, and future research
Our results suggest that contextual factors determine the relationship between beliefs and the PEBS. However, extant knowledge implies that the influence of beliefs on behavior is rather stable, therefore, our study opens a new pathway of inquiry into the role of contextual factors in the belief-behavior connection. This research used the Covid-19 severe disruption in the consumption context to explore this question, but its results extend to any other situation where consumption contexts are severely altered by external forces. (e.g., natural disasters, war displacement, health-related events). Environmentally caused disruptions will be of special interest in generalizing our results within the sustainable consumption framework. We are limited to the types of PEBs and beliefs measured, but the range of behaviors and beliefs related to sustainable lifestyles is obviously much wider. Future research may extend this study's results by using other combinations of beliefs and behaviors. Another limitation is the use of cross-sectional data captured using self-reported frequency questions that depend on recall and memory. Additional longitudinal and experimental studies are needed to confirm our main results.
contextual events can promote their adoption more effectively. Simultaneously, it is important to determine which behaviors do not fit with contextual demands to focus on efforts and resources. For example, focusing attention on promoting water conservation during a pandemic could contrast with the current safety demands of individuals. However, promoting awareness of other PEBS at home during lockdowns is a better way to encourage sustainability.
Our results also highlight that it is useful to recognize the disruption process. The discontinuity hypothesis asserts that a contextual change allows renegotiating some behaviors, but that there is a window of opportunity to make those changes. However, the identification of such windows is difficult. Our results highlight that the adaptive capacity of consumption behaviors follows the process of disruption and how people experience it. Policy interventions that promote sustainability may benefit from understanding such a process to capture when subjective experience favors the adoption of PEBs.
In addition to the ideas already mentioned, encouraging sustainability to face contextual challenges may be a path to future strategies. COVID-19 caused a disruption that permeated life in several domains, but the focus of mitigation at the individual level was mostly on health-related issues and income. Our findings showed that lockdowns led people to feel disempowered, which was driven by a decrease in some aspects of anthropocentric beliefs. Promoting the right PEBs may harness these feelings. For example, in the case of economic disruption, using PEBS oriented toward reusing, reselling, and repairing may contribute to saving and coping with economic and social issues. Hence, public policy and private action may be much more effective in promoting PEBs if they identify ways to help people cope with the threats faced in a disrupted context, as well as how to empower individuals through desirable practices and sustainable lifestyles.
The attitude-behavior gap continues to challenge several actors interested in promoting sustainability. Disrupted contexts may increase behavioral gaps, as our research suggests. Thus, policymakers should consider these results when setting strategies and activities to promote PEBs.

Appendix A. The context: the COVID-19 restrictions in Colombia
On March 6, 2020, the first case of COVID −19 disease was officially confirmed and reported in Colombia by the Health & Human Services Authority, Ministerio de la Salud y de la Protección Social. A few days after, on March 11th of 2020, the World Health Organization (WHO) declared COVID-19 as a pandemic, Following the WHO declaration, on March 12th, The Government of Colombia declared a Sanitary Emergency. Restrictions on visits to medical facilities and a ban on public gatherings of more than 50 people were imposed on March 16th, in addition to the closing of schools, including primary, secondary, and higher education.
On March 17, 2020, the government declared economic, social, and ecological emergencies in the country. On March 21st, international airports were closed and international flights were limited to humanitarian flights. Subsequently, the free circulation of people and vehicles, including national and foreign transportation, was limited to economic activities that guaranteed food supply, safety, and healthcare. The use of indoor facemasks, public transportation, and public spaces with a mass influx of people became mandatory on April 6, 2020. Subsequent decrees extended the preventive mandatory lockdown stage from April to June 2020, prohibiting free circulation and establishing exceptions for industries deemed vital for the subsistence of the economy, such as construction (limited to hospitals, transportation, and civil construction), hospitality, manufacturing, and textiles.
On June 1, 2020, the country began a new stage with the objective of gradually recovering its business sector by following control and safety protocols. In the main cities, this was accompanied by sectorized mandatory lockdowns, determined by local dynamics of confirmed cases and the ICU (Intensive Care Units) occupancy.
Main restrictions during lockdown were the following: • Free circulation of private vehicles, restricted for allowed sectors.
• Massive transportation could not operate with occupancy superior to 35%.
• Land and air transportation between municipalities could operate for food provisions only.
• International flights were prohibited.
• Real estate remodeling was prohibited.
• Restaurants could operate only for delivery service.
• Social activities, sports, and recreational events are prohibited in both public and private spaces.
• Public parks, amusing parks, cinemas, and theatres should remain closed.
Restrictions were lifted on September 1, 2020. The government disposed of voluntary confinement with responsible social distancing, which implied the fulfillment of biosafety protocols for every person in the national territory to avoid viral contagion. Local authorities at the municipality level retained the right to restrict activities and impose focalized lockdowns if necessary. During this new stage, only public and private events involving gatherings, bars, clubs, alcohol consumption in public spaces, and business establishments were prohibited. On September 2, 2020, national flights were authorized and from the 21st of September, international flights started operations. Approximately 90% of the economy reopened on September 15th. This study considers September 1st (six months after the beginning of the COVID-19 crisis) as the post-disruption moment for observation and analyses.
Appendix D  Table C1. Revised New Ecological Paradigm (Dunlap et al 2000).
Instructions: Please carefully read the statements listed below and circle the answer to indicate if you agree or disagree with the statement. 1. We are approaching the limit of the number of people the Earth can support Agree Disagree 2. Humans have the right to modify the natural environment to suit their needs Agree Disagree 3. When humans interfere with nature it often produces disastrous consequences Agree Disagree 4. Human ingenuity will ensure that we do NOT make the Earth unlivable Agree Disagree 5. Humans are severely abusing the environment Agree Disagree 6. The Earth has plenty of natural resources if we just learn how to develop them Agree Disagree 7. Plants and animals have as much right as humans to exist Agree Disagree 8. The balance of nature is strong enough to cope with the impacts of modern industrial nations Agree Disagree 9. Despite our special abilities, humans are still subject to the laws of nature Agree Disagree 10. The so-called 'ecological crisis' facing humankind has been greatly exaggerated Agree Disagree 11. The Earth is like a spaceship with very limited room and resources Agree Disagree 12. Humans were meant to rule over the rest of nature Agree Disagree 13. The balance of nature is very delicate and easily upset Agree Disagree 14. Humans will eventually learn enough about how nature works to be able to control it Agree Disagree 15. If things continue their present course, we will soon experience a major ecological catastrophe Agree Disagree