An Integrated Model of Consumer Recycling Behavior: Insights from Protection Motivation, Construal Level, and Regulatory Focus Theories

Background and problem statement

In response to the worsening waste problem, the Malaysian government has focused on promoting 3R initiatives (i.e., reduce, reuse, and recycle) as part of its national strategy toward zero-waste development (see⁷). Although consumer participation is a critical success factor—given that 65% of municipal solid waste (MSW) originates from Malaysian households³—motivating households to engage in 3R initiatives remains a significant challenge. This is evident from the failure of two nationwide 3R promotion campaigns conducted in 1993 and 2000,⁸ which manifests in low household participation in recycling programs.³

Disillusioned with the results of voluntary initiatives, the government shifted towards mandating household recycling through legislative enforcement (see⁷). This shift corresponds to what ölander, ThØgersen⁹ describe as the “applied behavioral approach” to consumer recycling behavior (CRB), where the emphasis is placed on facilitating recycling through incentives, enabling conditions, and sanctions. Under this paradigm, it is assumed that behavior shapes attitudes and motivations, rather than attitudes driving behavior.

However, the mandatory approach has been criticized for its limited effectiveness and efficiency (see¹⁰), and for generating contempt and backlash among socioeconomic stakeholders (see^9,11–13). As Zikmund, Stanton¹⁴ aptly observed, such policies address “only a symptom of the problem,” whereas the root cause lies in consumers’ lack of motivation for cooperation. From this perspective, motivation emerges as the central driver of behavior.

In contrast to the applied behavioral model, research in social psychology emphasizes psychological, contextual, and demographic antecedents of recycling motivation and behavior.⁹ Various studies have outlined comprehensive taxonomies of factors influencing recycling behavior (see^15–24). These factors are typically explained using theoretical frameworks such as the Theory of Planned Behavior (TPB²⁵), the Norm Activation Model (NAM²⁶), the Theory of Reasoned Action (TRA²⁷), and the Value-Belief-Norm Theory (VBN²⁸).^29,17

A primary, if not ultimate, objective of promoting CRB is to mitigate environmental degradation caused by MSW pollution and the overexploitation of natural resources. Environmental deterioration becomes a concern only when it poses a “threat” to human well-being.³⁰ Similarly, Rainear, Christensen³¹ argue that the perception of anthropogenic environmental degradation as a threat—hereafter referred to as perceived environmental threat—is a significant predictor of motivation for pro-environmental consumer behavior (PECB). Although the motivational role of environmental threat perception has been extensively investigated within the pro-environmental behavior (PEB) literature (for a review see^32,33), it remains surprisingly underexplored in the context of CRB,^17,34 with only a limited number of studies addressing the link between CRB motivation and environmental risk perception (e.g.^34–36).

Communicating environmental threats effectively remains a challenge.³⁷ This study argues that a major barrier is the detachment of environmental threats from the self. Specifically, the perceived lack of self-relevance of environmental threats can be attributed in part to the psychological distance typically associated with such threats risks.^38–40 As Carmi, Kimhi⁴¹ note, “in most cases, environmental threats are related to events that are not necessarily immediate, are often subject to controversy, and do not necessarily have a direct negative impact on any single person alone” (p. 2242). Research shows that most individuals lack a “tangible and concrete”⁴² or “vivid [and] personally relevant”⁴³ mental image of environmental threats. Consequently, a lack of motivation for PECB is often attributed to the perception of environmental threats as psychologically distant events.⁴⁴

Furthermore, this study maintains that the omission of perceived vulnerability from mainstream conceptualizations of environmental risk perception^45,46 also contributes to the failure of risk communication to establish self-relevance. Most studies conceptualize environmental risk perception in terms of perceived severity and probability (e.g.^47–55).

The notion of self-relevance is crucial for successful environmental risk communication for several reasons. First, individuals often believe that others are more vulnerable to harm than themselves, a phenomenon rooted in optimistic bias and self-enhancement bias.^56–59 This tendency is particularly pronounced regarding environmental threats (see^45,60). Second, according to appraisal theory,⁶¹ while perceived severity and probability provide individuals with the “knowledge” of the harm, this knowledge alone is insufficient to motivate action. Motivation requires the “heat” generated through primary appraisal, the perception that the threat is relevant to personal well-being (also see⁶²).

Appraisal theory further posits that primary appraisal is necessary but not sufficient; individuals must also engage in secondary appraisal—an evaluation of their ability to cope with the threat given personal and situational constraints. When individuals perceive a threat as beyond their control, they are likely to engage in psychological defense mechanisms such as denial or avoidance. This dynamic helps explain why social marketers often hesitate to use fear appeals to promote PECB, as negative emotions can backfire, leading to defensive denial or disengagement from the issue.⁶³

Research purpose

Drawing on protection motivation theory (PMT), construal level theory (CLT), and regulatory focus theory (RFT), this study proposes and tests an integrated framework (Figure 1) to answer the following research questions (RQ):

Figure 1. The conceptual framework.

Figure 2. The Structural model.

Consumer Recycling Behavior (CRB)

CRB is a specific form of PECB. The concept of PECB broadly refers to “the purchase, use, and disposal of personal and household products that have [positive] environmental impact”.⁶⁴ From a marketing perspective, CRB is categorized as an environmentally responsible disposal behavior.⁶⁵ Disposal behavior pertains to the stage of the consumption process wherein consumers discard household wastes “before, during, or after product use”.⁶⁶ Within the context of this study, CRB refers to consumers engaging in source separation—sorting and segregating their household waste at the point of disposal.⁶⁵

Several studies (e.g.,^34,67–69) have investigated CRB as part of the broader construct of Consumer Solid Waste Management Behavior (CSWMB), which includes the 3R activities (reduce, reuse, and recycle). However, this study deliberately focuses solely on CRB rather than aggregating it within the wider CSWMB framework for several reasons. First, as Phulwani, Kumar, Goyal²² argue, different forms of CSWMB represent distinct behaviors with varying antecedents. Second, applying Ajzen and Fishbein’s⁷⁰ principle of correspondence—which posits that predictors should be operationalized in direct relation to the specific behavior under investigation (see also^9,71,72)—this study seeks to ensure conceptual alignment between variables and observed behaviors. Investigating multiple facets of CSWMB would require lengthy questionnaires to establish the principal of corresponding, which risks reducing response quality due to participant fatigue (Burchell & Marsh, 1992; Herzog & Bachman, 1981). Thus, narrowing the scope to CRB serves to maintain both the focus and reliability of the study.

Motivation

Motivation is a central construct in the social psychological explanation of human behavior.⁷³ It is defined as “the forces acting on or within an organism to initiate and direct behavior”⁷⁴. Motivation determines not only which actions individuals choose to undertake, but also the extent of their determination and commitment toward those actions.

From a social psychological perspective, motivation arises from two primary categories of variables: internal and external forces.^75,76 Internal sources include needs, cognitions, and emotions. Needs refer to physiological and psychological requirements essential for survival, growth, and well-being. Cognitions encompass the unique mental processes—such as perceptions, beliefs, and expectations—that guide individuals’ understanding and actions. Emotions are characterized as automatic, rapid, and intense reactions to salient stimuli. In contrast, external sources pertain to the social and environmental factors that either incentivize or discourage particular behaviors.

Within the CRB literature, various determinants of motivation and behavior have been explored. These include behavioral (e.g.,^77–80), neuropsychological factors (e.g.,^81–83), personal values (e.g.,^77,84), self-protective motives (e.g.,^77,84), and the influence of persuasive messaging (e.g.,^80,85–88).

Protection motivation theory

Protection Motivation Theory (PMT), developed by Rogers,^89,90 is rooted in both appraisal theory and the expectancy-value tradition.⁹¹ Initially formulated to explain health-related behaviors, PMT has since been applied across various domains, including PECB. Although the earliest application of PMT to environmental behavior can be traced to Kantola, Syme, and Nesdale’s⁹² study on water conservation motivation, it was not until the second half of the 2010s that PMT gained broader traction within the PECB literature (for a review see³²).

In line with Lazarus’ appraisal theory, PMT postulates that protection motivation is a function of two sets of appraisals; the threat appraisal which corresponds to the Lazarus’ concept of primary appraisal, and the coping appraisal which reflects the notion of secondary appraisal.

Threat appraisal

A key advantage of the PMT over other risk perception models is that it treats perception of severity and vulnerability in their own rights. Perceived severity- a value element; refers to “appraised severity of [a] depicted event” given rise by perceived “magnitude of noxiousness of [the] depicted event”.⁹³ In the context of environmental threats, perceived severity is conceived as subjective evaluation of magnitude or seriousness of the harms caused by anthropogenic environmental deteriorations.⁹⁴

Perceived vulnerability-an expectancy component, is conceptualized as the “expectancy of exposure to the event” formed by “the conditional probability that the event will occur provided that no adaptive activity is performed” (Rogers & Mewborn, 1976, p. 55). In the context of this study, it refers to individuals’ subjective evaluation of the likelihood of being personally affected by the harm caused by deterioration of environment (Fernando, Sivakumaran, & Suganthi, 2016; Rainear & Christensen, 2017).

Coping appraisal

Consistent with appraisal theory, PMT posits that without sufficient coping appraisal, the mere perception of high severity and vulnerability to a threat can result in a boomerang effect, diminishing rather than enhancing motivation.⁹⁰ Within the PMT framework, coping appraisal comprises three key constructs: (1) perceived response efficacy—the expectation that the recommended behavior (recycling within the context of this study) would actually mitigate the threat, (2) Bandura’s⁹⁵ notion of self-efficacy—the expectation that “one is or is not capable of performing a behavior”, and (3) perceived cost of the recommended action—the subjective assessment of the “inconvenience, expense, unpleasantness, difficulty, complexity, side effects, disruption of daily life, and overcoming habit strength”.⁹⁰ PMT postulates that protection motivation is the positive function of perceived self-efficacy and perceived response efficacy, while perceived cost undermines motivation.

Construal level theory

This study draws on Construal Level Theory (CLT) of psychological distance^96,97 to explain the self-detachment of environmental threats, and the consequent demotivation for CRB. CLT has been applied across a wide range of research topics (see^98–100). The central premise of CLT is that individuals’ perceived psychological distance from an event influences their responses to it.^96,97,101

Psychological distance is defined as the extent to which “any target of thought is removed from the egocentric here and now on dimensions of time, space, social distance, or hypotheticality”.¹⁰² Temporal distance concerns the perceived immediacy of an event, ranging from the present moment to the distant future. Spatial distance is defined in terms of the geographical proximity of the event, referring to whether it occurs nearby or in a distant location. Social distance pertains to the perceived similarity or kinship with the people affected by the event. Hypothetical distance refers to the degree of certainty regarding the occurrence of the event.⁹⁶

According to CLT, psychological distance is closely linked to the construal level of thought. Construal level denotes the extent to which the object of thought is represented in abstract or concrete terms. A high construal level reflects an abstract mode of thinking, characterized by “schematic, decontextualized representations […] [that] consist of a few superordinate, core features of events”.⁹⁷ In contrast, a low construal level reflects a concrete mode of thinking, producing “unstructured, contextualized representations that include subordinate and incidental features of events”.⁹⁷

CLT posits that as psychological distance increases, individuals are more likely to construe events in abstract terms; conversely, events perceived as psychologically close are construed more concretely. Furthermore, the four dimensions of psychological distance are interrelated, such that a perceived increase or decrease in one dimension often transfers to changes in the others (Trope & Liberman, 2010).

Regulatory focus theory

Although it is conventionally assumed that psychological distance and environmental risk perception are negatively correlated, several studies have not found support for this hypothesis (e.g.,^103,104) or have even reported contrary evidence (e.g.,^105–107). For instance, Spence, Pidgeon¹⁰⁵ demonstrated that distal framing of climate change impacts induced greater perceptions of fear and severity of the threat. The authors attribute this result to individuals’ biased optimism regarding their own risk of threat. Drawing on the notion of congruency, Brügger, Morton, Dessai¹⁰⁶ argue that distal perception of threat is a better predictor of mitigation behavior because it aligns more closely with the construal level associated with such behavior. In another study, Brügger, Morton, Dessai¹⁰⁷ found that the proximate psychological distance of climate change predicts perceived risk and behavioral intention only when it interacts with fear. Collectively, these findings suggest that the relationship between psychological distance and threat appraisal may depend on additional moderating factors.

The inconsistent findings regarding the association between psychological distance and environmental risk perception highlight the need to identify potential moderators influencing this relationship. Prior research has called for further investigation into moderators of threat perception both generally (e.g.,^108,109) and specifically within the frameworks of CLT and PECB (e.g.,^98,110). In response to this need, the present study proposes that regulatory focus theory offers a valuable lens to identify conditional factors affecting the link between psychological distance and environmental risk perception. Existing studies have demonstrated that individuals’ regulatory focus (RF) is associated with their threat perceptions (e.g., Awagu & Basil, 2016; C. Keller et al., 2012; A. Y. Lee & Aaker, 2004; Y.-C. Lin et al., 2012), as well as with their perceptions of and responses to psychological distance and construal level (e.g.,^111–113).

Drawing a distinction between the needs for ‘nourishment’ and ‘protection’, Higgins¹¹⁴ introduced two motivational systems, or regulatory foci: promotion and prevention. Driven by the desire to fulfill ideals and attentive to achievement needs, individuals with a promotion focus tend to “see the glass half full”, promotion-focused individuals strive to approach opportunities so as to avoid “errors of omission”—i.e., missing an opportunity. Due to this eager goal pursuit strategy, promotion-focused individuals tend to explore many alternatives to ensure they do not miss any chances for advancement.

In contrast, those with a prevention focus are driven by the need for security, making them vigilant against negative outcomes. Prevention-focused individuals’ concern with fulfilling ‘oughts’ (i.e., responsibilities) compels them to avoid “errors of commission”—i.e., failing to stay out of harm’s way. Due to this vigilant goal pursuit strategy, prevention-focused individuals prefer to pursue as few alternatives as possible in order to minimize the probability of making wrong choices and incurring losses.¹¹⁵ As Higgins, Silberman¹¹⁶ state, promotion-focused individuals are attentive to the absence or presence of positive outcomes, whereas prevention-focused individuals are concerned with negative consequences.

Later, Higgins¹¹⁷ introduced the ‘regulatory fit’ principle, which maintains that individuals evaluate a goal-oriented activity more favorably when it aligns with their dominant regulatory focus. Under conditions of regulatory fit, motivation is stronger and choices are perceived as more satisfactory,¹¹⁷ perceived value is higher,¹¹⁸ brand attitudes are enhanced,^{115,119–121} and, in short, “people feel right about what they are doing”.¹²⁰

Research hypotheses

Association between threat appraisal and protection motivation

Previous applications of the theory in PECB context are mixed. While some studies report positive association between perceived severity of, and/or perceived vulnerability to the threat and intention to recycle,^34,35 organic food consumption,¹²² intention to purchase green products,^123,124 and other facets of pro-environmental behavior (e.g.^125,126); others (e.g.^{35,122,127,128}) do not find support for the postulated associations between perceive severity of, and/or perceived vulnerability to threat and protection motivation in PECB context. In line with the PMT postulation, this study posits following hypotheses (H):

Association between coping appraisal and protection motivation

The PECB literature reports mixed results on the association between the facets of coping appraisal and protection motivation. While majority of the results indicate PECB is a positive function of both perceives response efficacy and perceived self-efficacy,^{31,122–124,128–130} others don not support the postulated relationships for perceived response efficacy,³⁴ and perceived self-efficacy.^35,127 In regard with perceived cost, except for Janmaimool³⁴ and Kim, Jeong, Hwang,³⁵ all of the above mentioned studies find support for the negative association between the construct and protection motivation. In line with the PMT, this study posits following hypotheses:

Association between psychological distance and protection motivation mediated by threat appraisal

Lee, Sung, Wu, Ho, Chiou⁴⁹ report positive impact of proximate psychological distance on pro-environmental behavior (PEB), mediated by perceived environmental risk. Similarly, Carmi, Kimhi⁴¹ report a negative correlation between the perceived psychological distance of environmental threats and the perceived severity of their consequences. Psychological distance was a strong predictor of participants’ emotional engagement with environmental issues and their willingness to take action to mitigate them. Drawing on Weber’s¹³¹ work, the authors suggest that when environmental threats are perceived as psychologically closer, individuals form more vivid mental images of the associated harms, heightening the perceived severity of the consequences. Likewise, Spence, Poortinga, Pidgeon¹³² show that a closer perception of climate change risks predicts a greater intention to reduce energy consumption, both directly and indirectly via increased environmental concern. Notably, this study is among the few to simultaneously examine all facets of psychological distance. Furthermore, experimental results from Scannell, Gifford¹³³ reveal that perceiving climate change impacts as spatially closer enhances individuals’ engagement with the issue, with this effect being strengthened by their attachment to place.

Building on the notion of psychological distance as an egocentric concept—defined as the “subjective experience that something is close or far away from the self”⁹⁶—this study argues that perceiving environmental threats as proximal makes them feel personally relevant, thereby heightening the sense of vulnerability. Specifically:

According to CLT, a low construal level involves forming a detailed and concrete representation of an event. In this context, when an environmental threat is construed concretely, individuals are more likely to associate it with vivid dangers such as “cancer” or “famine” rather than broader, more abstract concepts like “disease” or “drought.” As Rogers⁹⁰ suggests, such vivid portrayals are likely to enhance the perceived severity of the threat. In other words:

According to, Florence, Fleischman, Mulcahy, Wynder,¹³⁴ the literature has yet to establish a definitive understanding of whether, and in what ways, construal level relates to PECB. While some studies report no significant relationship between the two constructs (e.g.,¹⁰⁴), others identify both positive (e.g.,^135,136) and negative (e.g.,^{80,104,137–140}) associations. Reflecting on these mixed findings, , Keller, Marsh, Richardson, Ball¹⁴¹ suggest that the relationship between psychological distance and PECB is likely shaped by complex mediation processes. Following Brügger,¹⁴² this study proposes that psychological distance does not influence PECB directly; rather, the link between the two is better understood through mediating factors such as perceived threat from environmental degradation. Specifically:

Moderated mediation of threat appraisal by regulatory focus

Several studies have investigated the fit between regulatory focus and construal level (e.g.,^{143–146,115,147,148}), as well as with psychological distance (e.g.,^149–155), and their impacts on various dependent variables, including PECB (e.g.,^88,156–158). These studies typically operationalize regulatory fit in terms of the interaction between regulatory focus and psychological distance.

The interaction between regulatory focus and construal level is often explained through the regulatory fit identified between prevention (promotion) focus and low (high) construal mindsets, as articulated by Förster, Higgins.¹⁴⁴ They argue that the vigilant goal pursuit strategy characteristic of prevention focus requires the precise determination of every single step in advance to minimize errors. This strategy is exclusive, in the sense that it involves eliminating unnecessary details to focus on specific, precise information needed to maintain security. Such exclusive information acquisition and processing necessitates attention to concrete details in the immediate environment, aligning with a low construal mindset.

Conversely, the eager goal pursuit strategy entailed in promotion focus demands an inclusive, broad, and general consideration of all available opportunities to avoid missing an opportunity. This exploratory cognitive style, which depicts a wider and more holistic picture, is better aligned with a high construal mindset.

Building on this reasoning, the present study posits the following: since prevention (promotion) focus fits with low (high) construal mindsets, and given that proximate (distant) psychological distance is congruent with low (high) construal mindsets, a regulatory fit is expected between prevention (promotion) focus and proximate (distant) psychological distance. Applying this to the context of current study, prevention (promotion) focus is expected to undermine (intensify) the proposed negative association between psychological distance and threat appraisal. In other words, the tendency to discount environmental risks due to their psychological distance will be more pronounced among promotion-focused individuals than among prevention-focused individuals. From a regulatory fit theoretical perspective, the (mis)fit between (prevention) promotion focus and the facets of psychological distance is expected to manifest itself in the motivation for recycling. That is, prevention (promotion) focus is expected to undermine (intensify) the indirect negative impact of the facets of psychological distance on recycling intention. Therefore:

The Sampling and data collection

This study employed a survey method rather than an experimental design, because experimental designs do not guarantee valid causal inferences when models involve mediation and moderation analyses, particularly when latent constructs are included (see^159–161).

Following the recommendation of Hair, Black, Babin, Anderson,¹⁶² the required sample size was determined using power analysis. For this purpose, Soper¹⁶³ has developed an online, freely accessible software tool, which has been utilized in several prior studies (e.g.,^164–166). Based on the parameters for the proposed model—desired effect size of 0.2, 14 latent variables, 59 observed items, a significance level of 0.05, and a desired statistical power of 0.8—the recommended minimum sample size was 521 respondents.

Data collection took place between May 2024 and February 2025, using a convenience sample of university students in Selangor, Malaysia. Respondents were approached in person and invited to participate by scanning a QR code with their own devices, which directed them to an online questionnaire. A total of 508 valid responses were obtained for the final analysis.

Prior to data collection, an ethic approval letter (Reference No. RMC/REC/EA/083/2024) was obtained from the Research Management Center (RMC) at Multimedia University (MMU).

The online form constitutes of an introductory section for obtaining participants implied consent by informing them that, by completing the questionnaire, they voluntarily agree to take part in the study. Furthermore, participation was conditional for being 18 years and above by including a screening question.

Pilot study

Following Bryman,¹⁶⁷ a pilot study was conducted to identify potential issues with the reliability and validity of the measurement instrument. The first 40 responses were used for the pilot analysis. Consistent with Hair, Black, Babin, Anderson,¹⁶² the results suggested that reliability was unlikely to be a concern: all Cronbach’s alpha values were well above 0.6, no item-to-total correlations fell below 0.5, and no inter-item correlations were less than 0.3.

Since there were no methodological differences between the pilot and main studies, and because the pilot study did not lead to any modifications of the measurement scale or study design, the pilot data were incorporated into the main analysis (see^168,169).

Data refinement

There were no missing or out-of-range data due to the data collection method employed in this study. The online questionnaire was designed such that successful submission was conditional upon answering all questions; respondents were required to complete every item to submit their responses. Furthermore, given the format of the online questionnaire, it was not possible for respondents to select more than one choice per question. Additionally, since no open-ended questions were included in the instrument, there was no risk of receiving out-of-range values.

Assumption of covariance-based structural equation modeling (CB-SEM)

The data met all assumptions required for CB-SEM as outlined by Hair, Black, Babin, Anderson.¹⁶² No instances of skewness beyond ±3 or kurtosis beyond ±10 were observed, indicating univariate normality of distribution, which further suggests multivariate normality.^162,170 Multicollinearity was not a concern, as all variance inflation factors (VIFs) were below the threshold of 10 (see¹⁷⁰). Univariate outliers were unlikely, given that no normalized value exceeded the ±3.29 threshold. The Mahalanobis D² test revealed no multivariate outliers, as all D²/df values were less than 2.5 (see¹⁶²). Scatter plots of Bartlett factor scores obtained from the confirmatory factor analysis (CFA) indicated linear relationships among all pairs of hypothesized dependent and independent variables (see^171,172). Furthermore, examination of residual plots from simple linear regressions for all hypothesized related variable pairs confirmed the assumption of homoscedasticity (see^162,173).

Common method variance (CMV)

Following Rodríguez-Ardura, Meseguer-Artola,¹⁷⁴ several a priori strategies were implemented to minimize the risk of common method variance (CMV). Except for demographic items, which were placed at the end of the questionnaire, all other items were presented in random order to avoid systematic response bias.¹⁷⁵ Respondents were assured of their anonymity and were informed that there were no right or wrong answers, thereby reducing evaluation apprehension. Special care was taken to ensure clarity and simplicity of the scale items, and detailed instructions were provided to guide respondents in completing the questionnaire and minimizing confusion.

In addition to these preventive measures, a post hoc test was conducted using the unmeasured latent method construct (ULMC) approach, following the procedure outlined by Collier.¹⁷⁶ Specifically, a ULMC model was generated by adding a common method factor to the CFA model, establishing direct paths from this factor to all observed indicators. The underlying rational is that the common method factor captures the variances across constructs caused by the potential method bias. The comparison of chi-square values between the CFA model and the ULMC model revealed no statistically significant difference at p≤0.05, indicating that CMV is unlikely to pose a concern in this study.

Measurement

All constructs in this study were measured using semantic differential and Likert-type scales, all either adapted or adopted from previously validated instruments. Following Clark-Carter and Clark-Carter, Clark-Carter,¹⁷⁷ all scales were anchored on 7-point response formats to facilitate the application of parametric analyses to ordinal data.

The four facets of psychological distance were measured using items adapted from Loy, Spence.¹⁷⁸ Except for perceived cost of behavior, which was measured with items adapted from Rainear, Christensen,³¹ and intention to recycle, measured with items adapted from Wang, Mangmeechai,¹⁷⁹ all other PMT constructs were measured using items from Witte, Cameron, McKeon, Berkowitz.¹⁸⁰ Finally, regulatory focus was assessed using items adapted from Haws, Dholakia, Bearden.¹⁸¹ Socially desirability responding was measured by the items adopted from.¹⁸²

The measurement model

Reliability

Following the recommendations of Hair, Black, Babin, Anderson,¹⁶² three criteria were used to assess the reliability of the measurement instrument. First, all Cronbach’s alpha values exceeded 0.7, and no instance of item-to-total correlation below 0.5 or inter-item correlation below 0.3 was observed. Second, all composite reliability values were above the 0.7 threshold, indicating satisfactory latent construct reliability. Third, standardized factor loadings from the CFA were above 0.7 for all items except five (i.e., hyp1, imp4, prev1, self3, and socl3, see Table 1). These items were retained for several reasons. First, in line with Hulland, Baumgartner, Smith,¹⁸⁴ items with loadings above 0.5 may be retained if necessary to preserve the construct’s integrity. Eliminating these items would have resulted in some constructs being measured by fewer than three items, which is discouraged in SEM due to concerns about model stability.¹⁶² Second, following Collier,¹⁷⁶ the retention of these items was justified as they did not negatively affect the average variance extracted (AVE) values. Third, the removal of these items did not lead to improvements in the measurement model’s goodness-of-fit (GOF) indices.

Validity

Convergent validity of the constructs was supported, as all average variance extracted (AVE) values exceeded the 0.5 threshold (Collier, 2020). Discriminant validity was assessed using the Fornell-Larcker criterion.¹⁷⁶ The results showed that, for each construct, the square root of its AVE was greater than its correlations with any other construct. Additionally, the heterotrait-monotrait (HTMT) ratio of correlations analysis¹⁷⁶ provided further evidence of discriminant validity, with no HTMT values exceeding the recommended threshold of 0.85.

Measurement model’s fit estimation

A CFA using the maximum likelihood estimation (MLE) approach was conducted in SmartPLS4 to evaluate the goodness-of-fit (GOF) of the measurement model. Based on the thresholds recommended by Hair, Black, Babin, Anderson,¹⁶² the model demonstrated a satisfactory fit to the data (χ2 = 1941.877, df = 1339, p = 0, χ2/df = 1.45, RMSEA = 0.03, SRMR = 0.039, NFI = 0.899, TLI = 0.962, CFI = 0.966). Modification indices suggested no valid changes with χ² shifts larger than 10 (see¹⁷⁶).

Structural model’s fit estimation

The structural model (Figure 2) was specified to include the hypothesized main effects while controlling for socially desirable responding, as well as age, gender, and education. In line with Schoemann, Jorgensen,¹⁸⁵ the main effects of moderating variables were also controlled for in the model. The inclusion of control variables followed the procedure outlined by Collier.¹⁷⁶

A ‘Basic CB-SEM algorithm’ was employed in SmartPLS4 to assess the goodness-of-fit (GOF) of the structural model. Except for χ² and the NFI, all other indices indicated a satisfactory fit (χ² = 2186.238, df = 1489, p = 0, χ²/df = 1.468, RMSEA = 0.03, SRMR = 0.046, NFI = 0.892, TLI = 0.959, CFI = 0.963). The lower value of the NFI was a particular concern. Given the structural model’s complexity, it was expected that the NFI would be inflated.¹⁶² Although the marginal deviation of NFI in the measurement model (a difference of just 0.001) was not problematic, it became more significant in the structural model. To address this, the item with the lowest factor loading (i.e., HYP1) was removed, leading to an improvement in the NFI (χ² = 1988.452, df = 1433, p = 0, χ²/df = 1.381, RMSEA = 0.028, SRMR = 0.042, NFI = 0.9, TLI = 0.966, CFI = 0.97). Modification indices did not suggest any further valid alterations with a chi-square shift of at least 10. An inspection of the standardized residual covariance matrix did not reveal any model fit issues. Reliability and validity checks confirmed no problems, as all Cronbach’s alphas exceeded 0.7, no composite reliabilities fell below 0.7, and no average variance extracted (AVE) values were less than 0.5. Additionally, all HTMT values were below 0.8, further supporting discriminant validity.

Hypothesis testing

Analysis of the direct relationships

A bootstrapping procedure was employed using SmartPLS4 to examine the hypothesized direct and mediation effects. Specifically, a one-tailed, bias-corrected and accelerated (BCa) approach with 10,000 resamples was applied. The results provided support for all hypothesized direct and mediated relationships ( Tables 2 and 3). Consistent with H1, a statistically significant positive association was observed between perceived severity and the intention to CRB (Unstandardized path coefficient (B) = 0.2, t = 4.750, p < 0.001). In line with H2, perceived vulnerability was also positively associated with CRB intention (B = 0.180, t = 5.618, p < 0.001). Both efficacy-related constructs were significant predictors of CRB intention, lending support to H3 and H4. Specifically, perceived response efficacy showed a positive association with CRB intention (B = 0.082, t = 2.147, p < 0.05), as did perceived self-efficacy (B = 0.075, t = 2.377, p < 0.05). Consistent with H5, perceived cost of CRB was negatively associated with the behavioral intention, and this relationship was statistically significant (B = -0.112, t = 3.528, p < 0.001).

Furthermore, all four dimensions of psychological distance were found to be significantly and negatively related to threat appraisals. Supporting H6a-d, perceived temporal distance (B =-0.31, t = 5.236, p < 0.001), spatial distance (B = -0.235, t = 4.9, p < 0.001), social distance (B = -0.316, t = 6.925, p < 0.001), and hypothetical distance (B = -0.388, t = 5.812, p < 0.001) were all negatively associated with perceived vulnerability. Similarly, the results supported H7a-d, indicating that perceived temporal (B = -0.174, t = 2.729, p < 0.05), spatial (B = -0.111, t = 2.071, p < 0.05), social (B = -0.177, t = 3.404, p < 0.001), and hypothetical (B = -0.192, t = 2.778, p < 0.055) distances were negatively associated with perceived severity.

Mediation analysis

In line with H8a through H8d, perceived vulnerability fully mediated the relationships between each facet of psychological distance and CRB intention. The mediation effects were significant for temporal distance (B = -0.056, t = 3.738, p < 0.001), spatial distance (B = -0.042, t = 3.95, p < 0.001), social distance (B = -0.057, t = 4.423, p < 0.001), and hypothetical distance (B = -0.070, t = 4.168, p < 0.001).

Similarly, the findings supported H9a through H9d, which posited that perceived severity would fully mediate the effect of psychological distance on recycling intention. Temporal (B = -0.035, t = 2.175, p < 0.05), spatial (B = -0.022, t = 1.827, p < 0.05), social (B = -0.036, t = 2.527, p < 0.05), and hypothetical (B = -0.038, t = 2.174, p < 0.05) distances were each found to influence CRB intention via perceived severity of environmental harm.

Control variables

Among the control variables, facets of socially desirable responding showed statistically significant effects. Impression management was positively associated with intention to CRB (B = 0.109, t = 3.304, p < 0.001), as was self-deceptive enhancement (B = 0.127, t = 3.716, p < 0.001). All other control variables did not show statistically significant associations with the dependent variable.

Moderated mediation analysis

To test the hypothesized moderated mediation effects, a PROCESS analysis was conducted using SmartPLS4, applying a one-tailed, BCa bootstrapping procedure with 10,000 resamples. The results supported H10a-d, which predicted that prevention focus would negatively moderate the indirect effects of psychological distance on recycling intention via perceived severity. The corresponding index of moderated mediation (IMM) for each path was statistically significant ( Table 4). Specifically, prevention focus was found to attenuate the mediated relationships for temporal distance (Unstandardized IMM (B) = -0.039, t = 3.698, SE = 0.010, 95% CI = [-0.058, -0.023]), spatial distance (B = -0.019, t = 2.612, SE = 0.007, 95% CI = [-0.033, -0.009]), social distance (B = -0.029, t = 3.751, SE = 0.008, 95% CI = [-0.043, -0.018]), and hypothetical distance (B = -0.031, t = 3.461, SE = 0.009, 95% CI = [-0.048, -0.018]).

Partial support was found for H11a through H11d. Prevention focus was shown to significantly moderate the indirect effect of perceived temporal distance on recycling intention via perceived vulnerability (B = -0.014, t = 2.123, SE = 0.006, 95% CI = [-0.025, -0.004]]). However, no significant moderation effects were found for the remaining three distance dimensions: spatial (B = -0.005, t = 0.990, SE = 0.005, 95% CI = [-0.015, 0.003]), social (B = -0.007, t = 1.254, SE = 0.005, 95% CI = [-0.016, 0.002]), and hypothetical (B = -0.007, t = 0.973, SE = 0.007, 95% CI = [-0.019, 0.004]).

All predictions under H12a-d were supported. Promotion focus was found to positively moderate the indirect associations between psychological distance and recycling intention through perceived severity. These effects were statistically significant for temporal (B = 0.028, t =2.776, SE = 0.010, 95% CI = [0.014, 0.047]), spatial (B = 0.024, t = 2.863, SE = 0.008, 95% CI = [0.012, 0.039]), social (B = 0.039, t = 3.983, SE = 0.010, 95% CI = [0.025, 0.057]), and hypothetical (B = 0.043, t = 4.027, SE = 0.011, 95% CI = [0.028, 0.064] distances.

By contrast, none of the predictions under H17a-d were supported. Promotion focus did not significantly moderate the indirect effects of psychological distance on intention to recycle through perceived vulnerability. Specifically, no significant moderated mediation effects were observed for temporal (B = 0.005, t = 0.700, SE = 0.008, 95% CI = [-0.007, 0.018]), spatial (B = 0.001, t = 0.122, SE = 0.007, 95% CI = [-0.010, 0.012]), social (B = 0.007, t = 1.112, SE = 1.112, 95% CI = [-0.003, 0.019]), or hypothetical distance (B = 0.003, t = 0.406, SE = 0.008, 95% CI = [-0.010, 0.018]).

Probing moderated mediations

Table 5 presents the results of probing the moderated mediation relationships. The conditional indirect effect (CIE) of perceived temporal distance on intention to recycle—mediated by perceived severity—was statistically significant across all levels of regulatory focus, except when both promotion and prevention foci were low (Confidence Interval (CI) = [-0.044;0.022]) and when prevention focus was average while promotion focus was high (CI = [-0.018;0.034]).

Similarly, the conditional indirect effect of perceived temporal distance on recycling intention—mediated by perceived vulnerability—was significant across all levels of the moderators, except when low prevention focus co-occurred with high promotion focus (CI = [-0.056;0.002]).

Regarding perceived spatial distance, its conditional indirect effect on intention to recycle via perceived severity was significant at most combinations of regulatory focus. However, the effect was not significant when both promotion and prevention foci were high (CI = [-0.06;0.004]), when prevention focus was low and promotion focus was average (CI = [-0.022;0.022]), or when prevention focus was average and promotion focus was high (CI = [-0.020;0.031]).

The indirect effect of perceived social distance—mediated by severity—was likewise significant across most conditions, except when prevention focus was low and promotion focus was average (CI = [-0.036;0.009]), or when prevention focus was average and promotion focus was high (CI = [-0.022;0.021]).

A similar pattern was observed for perceived hypothetical distance. The indirect effect mediated by perceived severity was not significant when prevention focus was low and promotion focus was high (CI = [-0.014;0.031]), or when prevention focus was average and promotion focus was high (CI = [-0.001;0.047]).

Overall, three key patterns emerge from the moderated mediation analysis. First, all conditional indirect effects mediated by perceived severity became non-significant when prevention focus was average and promotion focus was high. Second, with the exception of temporal distance, none of the severity-mediated indirect effects were significant when prevention focus was low and promotion focus was average. Third, in all cases where prevention focus was low and promotion focus was high, the sign of the indirect effect mediated by perceived severity changed from negative to positive (CIE_{Temporal Distance} = 0.067; CIE_{Spatial Distance} = 0.037; CIE_{Social Distance} = 0.044; CIE_{Hypothetical Distance} = 0.062), suggesting a reversal in the direction of the effect under these regulatory focus conditions.

Theoretical implications

The results of the present study provide strong confirmation of PMT, particularly regarding the motivational effects of threat appraisal and coping appraisal. However, in contrast to meta-analytic findings from other contexts (see^186,187), the path coefficients for the facets of coping appraisal in this study were smaller than those for the facets of threat appraisal. We speculate that this weaker predictive power of coping appraisal in the context of PECB, compared to other contexts, could be indicative of a unique characteristic of PECB that interacts with the coping appraisal constructs.

In an unpublished conceptual manuscript, Mir Salehi Herisi, Mahdee J & Fauzi NM¹⁸⁸ argue that PECB, as a collective action, involves individual motivation that is influenced by factors beyond individual agency. This contrasts with the original formulation of PMT, which was designed to explain behaviors where response efficacy is determined primarily by an individual’s own actions. The effectiveness of PECB, however, relies on collective efforts, and as Stern⁶⁴ points out, it is “significant only in aggregate.” Mir Salehi Herisi, Mahdee J & Fauzi NM¹⁸⁸ further discuss the theoretical dynamics between the expectancy factors at the individual action level (i.e., efficacy perceptions) and those at the collective action level (i.e., the expectation of cooperation), exploring how these factors impact PECB motivation.

This study also represents one of the few applications of Construal Level Theory (CLT) that investigates all four facets of psychological distance within the same conceptual framework. The negative associations observed between all facets of psychological distance and the two facets of threat appraisal were consistent with CLT predictions.

The larger association between psychological distance and perceived vulnerability offers additional support for conceptualizing the facets of psychological distance as egocentric constructs. Specifically, the strong connection between psychological distance and vulnerability perception—both tapping into the idea of self-relevance—enhances the explanatory power of psychological distance in relation to vulnerability perception, as compared to perceived severity. The hypothesized association between psychological distance and vulnerability perception is justified by the self-relevance aspect of psychological distance. In contrast, the relationship between psychological distance and perceived severity can be understood through the lens of abstraction. The weaker direct association between psychological distance and severity perception may suggest a missing explanatory element—namely, the construal level of the threat—that could bridge the gap between the explanans (psychological distance) and the explanandum (severity perception). This, however, remains a speculation until it is supported by empirical evidences.

The study’s results also lend support to the theoretical stance of RFT on individual differences in motivational orientation. However, the consistent statistical support found for the moderating impact of regulatory focus specifically on the relationship between psychological distance and severity perception—an effect expected to be driven by the construal level of the threat—suggests that abstraction is the cognitive mechanism through which regulatory focus moderates the cognitive and motivational effects of psychological distance. While this interpretation is plausible, further empirical evidence is needed to confirm the role of abstraction in this moderating process.

Practical implications

The findings of this study highlight that perceived cost is a significant barrier to CRB. Importantly, the concept of cost extends beyond mere financial expenses, encompassing subtler factors such as the time, effort, and disruption to established habits and lifestyles. From this perspective, effective CRB policies must begin with a careful identification of the specific costs associated with recycling behavior and the development of strategies that facilitate and lower these costs for individuals.

The results also suggest that individuals are motivated when they believe their actions are effective in addressing environmental issues. Building on this theoretical insight, CRB policies should include mechanisms for providing regular feedback to the public regarding how consumer participation in sustainable waste management contributes to desired environmental and economic outcomes. Crucially, this feedback must not only communicate the importance of these outcomes but also emphasize their direct relevance to individuals’ well-being. This underscores the necessity of CRB campaigns that are effective in both their design and their implementation.

The importance and relevance of recycling outcomes, as suggested by this study, can be framed through the lens of harm mitigation. Consequently, CRB campaigns should prioritize the effective communication of the threat posed by MSW pollution. The findings indicate that communications should emphasize both the magnitude of the harm and individuals’ vulnerability to it.

Effectively communicating the severity of harm, according to the results, requires framing messages in concrete and relatable terms. For example, stating that “MSW pollution causes cancer” is likely to be more impactful than an abstract assertion such as “MSW pollution is bad for health”. Enhancing the perceived self-relevance of the harm can be achieved by reducing the psychological distance of the threat. Highlighting community-level impacts rather than merely local impacts can simultaneously reduce perceived spatial and social distances, making the threat feel more immediate and personally relevant.

Moreover, the finding that prevention focus reduces the negative effect of perceived psychological distance on vulnerability perception suggests that campaign messages should be framed to activate a prevention-focused mindset. For instance, a message such as “Don’t let waste harm your family—sort today” frames the threat in terms of obligation and protection, aligning with both the prevention orientation and the close psychological distance (i.e., concern for family). In contrast, a message like “Make your life healthier—sort today” could create regulatory incongruence by activating a promotion focus, potentially reducing the message’s effectiveness in a threat-prevention context.

Finally, the study’s evidence that CRB motivation is a function of both threat and coping appraisals suggests that emphasizing the threat of MSW pollution without simultaneously enhancing perceptions of self-efficacy and response efficacy could backfire. Messages that overemphasize threats without empowering individuals to act may induce helplessness, triggering emotion-focused coping rather than the desired problem-focused coping. Therefore, effective communication strategies should combine creative marketing techniques with scientific principles to design messages that both alert individuals to the threat and empower them to take effective action.