Using the Theory of Planned Behavior for Explaining Mobile Phone Recycling: The Role of Subjective Norms

Abstract

This study examines the extent to which standard Theory of Planned Behavior (TPB) predictors (Attitude; Subjective Norms; and Perceived Behavioral Control) explain the intention to recycle an end-of-use mobile phone. Our data originate from empirical research on Greek citizens conducted in the summer of 2022 (N = 258). Through ordinal regression modeling, we found that all the TPB predictors positively influence one’s intention to recycle. In contrast to most previous studies, we found that Subjective Norms is the most influential predictor. However, its influence is dependent on the type of norm (descriptive vs. injunctive), the type of social referent (family vs. close friends), and, most importantly, one’s level of identification with the social referent.

1. Introduction

According to the International Telecommunication Union, in 2023, 78% of the world’s population owned a mobile phone [1]. This translates into over 6.27 billion mobile phones currently in use. Unlike other technological products, mobile phones have a remarkably high replacement rate. A recent review found that the in-use lifespan of mobile phones was less than 3 years in most countries [2]. However, it seems that their replacement rate is accelerating, further fuelling the proliferation of end-of-use mobile phones; for example, a study in China, the world’s largest smartphone market, reported that the average life expectancy of mobile phones decreased from 2.9 years in 2011 to 2.21 years in 2018 [3], whereas it is projected that the smartphone replacement cycle in Western Europe will decrease to 33 months in 2025, compared with 40 months in 2020 [4].

The continuously growing global demand for mobile phones poses a significant sustainability challenge. Toxic chemicals are used for the extraction of lithium, the rare chemical element found in mobile phone batteries. These chemicals, alongside traces of lithium itself, may find their way into ‘waste storage ponds, tailings piles, processed waters, evaporate basins, and transported products; [and they] have biophysical consequences that could adversely impact human metabolism, neuronal communication, soil ecology, and aquatic life’ [5]. A substantial share of the metals that make smartphones operate the way they do (such as ‘tungsten [which] allows our phone to vibrate, cobalt and rare earth elements, which give our phone its crisp sound, indium [which is used in the] touch screen, and gold’ [6]) come through Artisanal and Small-Scale Mining (ASM) [7]. Since ASM is practiced by unskilled laborers via basic technologies, it poses a significant threat to the health and safety of its practitioners, as well as to the soil, crop production, and water reserves of neighboring communities (op.cit.). Smartphones also have a significant carbon footprint. The average life cycle greenhouse gas (GHG) emissions of smartphones is approximately 73 kg CO₂eq (kilograms of carbon dioxide equivalent), with almost two-thirds of these emissions occurring during the extraction of the rare metals used in mobile phones and the phone manufacturing phases [8].

The effects of mobile phones on sustainability may be substantially curtailed through recycling, a practice that makes both environmental and economic sense. Thus, a recent study comparing four different scenarios for mobile phone recycling in China concluded that ‘there is an over 70% chance that the recycling of waste cellphones has zero or negative emissions to the environment’ [9]. Similarly, ref. [10] found that Fairphone smartphones—a device produced using recycled, fair trade, and conflict-free materials—emit much fewer GHG emissions during its lifecycle than did mobile phones constructed from virgin materials. Furthermore, smartphones, owing to their gold and palladium contents, are the most profitable category of e-waste recyclables [11] (see also [12] for a similar conclusion).

Despite the obvious positive consequences of mobile phone recycling, their actual recycling rates are disappointingly low. In China, the world’s largest smartphone market, their recycling rate was just 4.2% (2021 data) [13], whereas in the European Union, which is a global leader in mitigating climate change and protecting the environment, only 10% of smartphones were recycled in 2022 [14]. To improve these rates, detailed policy interventions are needed, and to design these interventions, it is necessary to establish which factors influence one’s decision to recycle their mobile phone.

The Theory of Planned Behavior (TPB) has been extensively used to explain pro-environmental behaviors, including recycling. It is a psychological theory that ‘can and has been applied to situations in which individuals are faced with a choice among alternative behavioral options… even if the only alternative to performing a given behavior is not to perform it’ [15].

Despite its prominence, only a handful of studies have employed the TPB to examine mobile phone recycling, and all of them have been conducted in Asian countries. Additionally, almost all these studies have employed the TPB in tandem with other explanatory variables, thus making it impossible to establish the unique relevance/potential of the TPB in explaining mobile phone recycling. Furthermore, the only study exclusively using the TPB framework (see [16]) was conducted in a single Chinese province; thus, its findings may not be representative of the situation in other countries. Accordingly, our research, by assessing the predictive power of the TPB for explaining mobile phone recycling in the context of a European (Union) country, not only provides policy-relevant information but also contributes to a theoretically under researched topic.

2. Related Works

In the TPB framework, an individual’s intention to perform a particular behavior is assumed to be influenced by three prior factors: one’s ‘Attitude’, ‘Subjective Norms’ and ‘Perceived Behavioral, PBC’. ‘Attitude refers to the degree to which a person has a favorable or unfavorable evaluation or appraisal of the behavior in question… Subjective Norm [refer]… to the perceived social pressure to perform or not to perform the behavior […and…] Perceived Behavioral Control [refers]…to the perceived ease or difficulty of performing the behavior’ [17]. Accordingly, in this study, we test the following hypothesis:

Hypothesis 1 (H1).

‘Attitude’, ‘Subjective Norms’, and ‘Perceived Behavioral Control’ positively impact one’s intention to recycle their mobile phone.

While it has long been claimed that ‘the relative importance of attitude, subjective norms, and perceived behavioral control in the prediction of intention is expected to vary across behaviors and situations’ [17], a number of meta-analyses have shown that the Subjective Norms (SN) predictor has consistently been less satisfactory than the Attitude and ‘Perceived Behavioral Control’ predictors. For example, an analysis of ‘185 independent studies’ that used the TPB to explain various behaviors revealed that ‘the subjective norm–intention correlation is significantly weaker than […] the attitude […and…] PBC–intention correlation[s]’ [18]. Similarly, a meta-analysis of 43 research articles that employed the TPB for studying household and/or e-waste recycling behaviors concluded that ‘subjective norms showed poor efficacy in explaining recycling behavior’ [19]. Additionally, the handful of studies that employed the TPB to study mobile phone recycling specifically returned mixed results. Thus, study [20] reported SN to be the strongest predictor of mobile phone recycling, whereas [16,21,22,23] reported a much weaker (yet statistically significant) effect compared to Attitude and PBC. On the other hand, in the studies of [24,25], the Subjective Norms predictor was not statistically significant.

Previous research has argued that SNs’ weak explanatory power may be due to its suboptimal measurement. Thus, study [5] claimed that ‘the most likely explanation for the poor performance of the subjective norm component lies in its measurement: many authors use single item measures, as opposed to more reliable multi-item scales’, and their meta-analysis results revealed that ‘multiple measures of subjective norms are more strongly related to [recycling] intentions [than single ones]’ (op.cit). Similarly, study [19] also opined in favor of employing multiple measures of SN because this would minimize the risk of using an inappropriate/irrelevant personally significant referent while measuring the respondent’s normative beliefs. They argue that ‘many case studies merely treat family members or friends as the normative source group’ without actually checking whether the respondent actually identifies with these groups. This shortcoming is particularly striking since the conceiving father of the TPB, Icek Ajzen, has long stressed the significance of selecting ‘important’ (i.e., participant-relevant) social referents while examining the influence of SN on behavior(al intention) [17]. Thus, study [19] advised that ‘the measurement of subjective norms should be set as multiple groups identified by the participants, such as the community, municipality, colleagues, or learning partners (not only family members and friends)’ (p. 14, our emphasis).

We tentatively argue that there may be a further reason for the poor performance of the SN predictor: its partial measurement. In the TPB, the normative beliefs encapsulated in SN are of two kinds: ‘injunctive’ and ‘descriptive’. In particular, ‘injunctive normative belief is the expectation or subjective probability that a personally significant referent individual or group (e.g., friends, family, spouse, coworkers, one’s physician, or supervisor) approves or disapproves of performing the behavior under investigation. Descriptive normative beliefs, on the other hand, are beliefs about whether significant others themselves perform the behavior’ [15]. Employing both of them in an explanatory framework is necessary not only because they represent qualitatively different aspects of the normative beliefs (and, thus, different sources of motivation, see [26]) but also because each of them may invoke different ‘personally significant referents’: as [15] succinctly noted, in regard ‘to going on a weight-loss diet, injunctive normative beliefs may include the perceived expectation of a physician, but beliefs about the physician’s own behavior (descriptive normative beliefs) may be irrelevant and thus not come readily to mind’. Thus, SNs’ influence needs to be considered proportional to the sum of ‘each accessible normative belief…with respect to a given social referent, whether injunctive or descriptive, …in interaction with the referent’s importance or significance to the individual’ (op.cit.).

On the basis of the previous discussion, we test the following hypotheses:

Hypothesis 2 (H2).

Both descriptive and injunctive SNs positively affect recycling intention.

Hypothesis 3 (H3).

The effects of descriptive and injunctive SNs on recycling intention differ depending on the type of social referent.

Hypothesis 4 (H4).

The effect of a respondent’s identification with the referent is moderated by their descriptive SN.

Hypothesis 5 (H5).

The effect of a respondent’s motivation to comply with the referent is moderated by their injunctive SN.

3. Materials and Methods

3.1. Context and Sample

In the summer of 2022, we posted an online questionnaire asking participants to express, under conditions of anonymity, their views concerning the recycling of the mobile phones they no longer use. The questionnaire was communicated electronically through the academic email databases of the authors’ host university as well as through the university’s social media (Facebook and LinkedIn) pages, requesting that the recipients/readers not only participate in the research but also forward and share the questionnaire’s link to their personal and social networks. The online questionnaire remained available between 1 June and 30 June 2022, and 358 individual responses were collected. The fact that our sample was self-selected may have introduced bias, with more environmentally concerned individuals being more likely to take the time to fill in the questionnaire and thus being overrepresented in our sample (see [27]). Furthermore, our sample is not representative of the Greek population, with female, younger and highly educated individuals constituting the bulk of the respondents (see

Table 1. Demographic characteristics of the sample.

Gender	Percentages
Male	37.4
Female	61.7
Generation
Silent Generation (born 1925–1945)	0.3
Baby Boomers (1946–1964)	10.9
Generation X (1965–1979)	22.1
Millennials (1980–1994)	37.2
Generation Z (1995–2012)	28.8
Educational Attainment (achieved)
High School or lower	18.7
University degree/Vocational training	34.9
M.Sc. degree	33.5
Ph.D. degree	12.8

). These factors are expected to restrict variability and result in weakened correlations. We will return to these potential limitations in the concluding section of the paper.

3.2. Variables Used

3.2.1. Dependent Variable

The dependent variable for our analysis is one’s intention to recycle (RI) their mobile phone. In particular, the participants in this study were asked, ‘When the time comes to replace the mobile phone you are currently using, how likely is it that you will dispose of it at a certified point for e-waste recycling?’. The possible answers were measured on a 7-point scale ranging from ‘1’ (very unlikely) to ‘7’ (very likely).

3.2.2. Predictor Variables

An individual’s ATT toward mobile phone recycling was measured through the following 6 items, adopted in modified form from [28]: ‘Would you say that recycling the mobile phone you no longer use is:…very bad/very good; totally useless/very useful; displeases me/satisfies me; irresponsible/responsible; does not make sense/sensible; and, ‘a waste of time/very necessary’. Each item was measured on a scale from ‘1’ (the ‘negative’ extreme assessment of recycling) to ‘7’ (the ‘positive’ extreme assessment of recycling), with ‘4’ serving as the middle/neutral point.

An individual’s ‘Perceived Behavioral Control (PBC)’ toward recycling was measured through 4 items, adopted in slightly modified form from [16]: ‘I know where I can recycle my old mobile phone’; ‘I have plenty of opportunities to recycle my old mobile phone’; ‘If I wish to recycle my mobile phone, I can do it without difficulty’; and ‘Participating in mobile phone recycling depends solely on me’. The answers were coded on a 5-point scale ranging from ‘1’ (strongly disagree) to ‘5’ (strongly agree), with ‘3’ serving as the neutral point.

The variables measuring an individual’s Subjective Norms (SN), both their perception of the socially prescribed mode of conduct regarding mobile phone recycling and their willingness to conform to these social expectations, were formulated on the basis of [29] as follows:

Descriptive Subjective Norm (SN_descriptive): ‘The members of my family recycle the mobile phones they no longer use’; and ‘My closest friends recycle the mobile phones they no longer use’.

Injunctive Subjective Norm (SN_injunctive): ‘The members of my family think that I ought to recycle the mobile phones I no longer use’; and ‘My closest friends think that I ought to recycle the mobile phones I no longer use’.

Identification with the referent: ‘When it comes to mobile phone recycling, I want to behave as the members of my family behave’; and ‘When it comes to mobile phone recycling, I want to behave as my closest friends behave’.

Motivation to comply: ‘Regarding mobile phone recycling, I want to do what the members of my family think I should do’; and ‘Regarding mobile phone recycling, I want to do what my closest friends think I should do’.

The answers to these questions were also coded on a 5-point scale ranging from ‘1’ (strongly disagree) to ‘5’ (strongly agree), with ‘3’ as the neutral point.

The correlations between the norms variables are shown in

Table 2. Correlation scores for the norms variables (Kendall’s tau-b, 2-tailed). All reported scores are statistically significant at the 0.001 level.

	SNdescript Family	SNdescript Friends	SNinjunct Family	SNinjunct Friends	Identify with Family	Identify with Friends	Motivated by Family	Motivated by Friends
SNdescriptive
Family recycles	1.000
Friends recycle	0.624	1.000
SNinjuctive
Family thinks I should recycle	0.610	0.569	1.000
Friends think I should recycle	0.446	0.583	0.604	1.000
Identification with…
Family	0.355	0.341	0.438	0.300	1.000
Friends	0.202	0.254	0.209	0.306	0.528	1.000
Motivated to comply with…
Family	0.292	0.296	0.303	0.280	0.609	0.641	1.000
Friends	0.216	0.236	0.151	0.215	0.435	0.615	0.660	1.000

.

3.3. Methods

To test our hypotheses, we fitted four ordinal regression models to the data. The first model (Model 1) included only the six ATT and four PBC predictor variables. In Models 2A and 2B, we added the descriptive and injunctive SNs predictors, respectively. A comparison of the results of these three models will allow us to test Hypothesis 1 (i.e., ATT, PBC, and SN are statistically important predictors of recycling intention) and Hypothesis 2 (i.e., both descriptive and injunctive SNs positively affect one’s recycling intention). Furthermore, comparing Models 2A and 2B will allow us to check whether Hypothesis 3 holds (‘the effect of the descriptive and injunctive SNs on recycling intention will differ depending on the type of social referent’). Finally, in Model 3, we added the predictor variables ‘identification with referent’ and ‘motivation to comply’. By comparing Model 3 with the previous models, we can test Hypotheses 4 and 5, that is, to what extent the respondent’s identification (and motivation to comply) with a social referent is moderated by the respondents’ descriptive (and injunctive) SN.

To run the models, we utilized the statistical language R v. 4.4.1 [30], which was implemented in the integrated development environment RStudio v. 1.4 [31]. The R package ordinal v. 12-4.1 [32] was used to run the ordinal regression models, while the analysis of the results was performed with the package tidymodels v. 1.2.0 [33]. The tidymodels package allows us to identify statistically significant results. In addition, the R package rcompanion v. 2.4.36 was used to check the models’ fit by computing the pseudo R² [34]. The probability scores shown in Figure 1 were calculated via the ggplot2 package v. 3.5.1 [35].

4. Results and Discussion

Most of our respondents (90.8%) had only one mobile phone at the time of the research, whereas 83.2% of them claimed that they knew that mobile phones can be recycled. Our respondents tended to replace their mobile phones less often than what has been reported in other countries: while available research suggests that the average lifespan of a mobile phone is (considerably) less than 3 years, 58.1% of our respondents claimed that they replaced their mobile phones ‘every 3 years or more’. The predominant reason for replacing one’s mobile phone was claimed to be its functional failure (‘it was ruined/did not work’, 51.7%), followed by its obsoleteness (‘the device was old’ (22.1%) or ‘it was obsolete’ (19%)). Similar to other studies, the majority of our respondents stored their out-of-use mobile phones: 86.6% of them claimed to have mobile phones stored in their homes, whereas 63.7% had stored their last mobile phone. Concerning the reasons for storing these devices, on a scale from ‘1’ (not at all important) to ‘5’ (very important), the most important reason was found to be ‘reusing [the phone] if needed’, followed by the wish ‘to protect my personal data’ (median values of 4.00 and 2.50, respectively; multiple answers were allowed for this question). Recycling rates were found to be very low, with only 8.9% of the respondents claiming that they had recycled their mobile phones through a designated/certified e-waste recycling scheme.

With respect to the possible predictors of one’s intention to recycle mobile phones,

Table 3. Ordinal regression coefficients (odds ratios (ORs) and confidence intervals [95% CIs]) for mobile recycling intentions.

	Model 1	Model 2A	Model 2B	Model 3
ATTITUDE
Recycling a mobile phone I no longer use, is… … Good	n.s.	n.s.	n.s.	n.s.
…Useful	n.s.	n.s.	n.s.	n.s.
…Satisfying	1.51 *** [1.22, 1.87]	1.50 *** [1.22, 1.86]	1.51 *** [1.22, 1.87]	1.58 *** [1.27, 1.98]
…Responsible	1.32 * [1.02, 1.72]	n.s.	n.s.	1.31 * [1.00, 1.71]
…Makes sense	n.s.	n.s.	n.s.	n.s.
…Necessary	1.36 *** [1.08, 1.72]	1.31 * [1.04, 1.66]	1.31 * [1.03, 1.66]	1.30 * [1.02, 1.66]
PBC
I know where to recycle	1.41 ** [1.11, 1.74]	1.32 ** [1.07, 1.64]	1.35 ** [1.09, 1.67]	1.33 * [1.07, 1.67]
Have plenty of opportunities to recycle	n.s.	n.s.	n.s.	n.s.
I can recycle without difficulty	n.s.	n.s.	n.s.	n.s.
Recycling depends solely on me	n.s.	n.s.	n.s.	0.77 * [0.611, 0.959]
Subjective Normsdescriptive
Family		1.34 ** [1.09, 1.50]	n.s.	n.s.
Friends		n.s.	n.s.	2.44 ** [1.42, 4.24]
Subjective Normsinjuctive
Family			n.s.	n.s.
Friends			n.s.	n.s.
Identification with…
…Family				n.s.
…Friends				3.90 *** [1.9, 8.11]
Motivated to comply with…
…Family				n.s.
…Friends				n.s.
SNdescriptive*Identification
Family				n.s.
Friends				0.70 ** [0.566, 0.874]
SNinjuctive*Motivated
Family				n.s.
Friends				n.s.
Model fit indices
Nagelkerke R2	0.256	0.294	0.303	0.340
AIC	1252.1	1238.2	1237.6	1234.8
LRT (Chi square)	102.98 ***	120.86 ***	125.44 ***	144.26 ***
Δχ2 from previous model	---	17.882 ***	n.s.	18.821 *

Statistical significance, p: *** < 0.001; ** < 0.01; * < 0.05; n.s.: not statistically significant.

presents the results of our regression models.

Our results confirmed Hypothesis 1. Comparing Model 1 (which includes only the Attitude and PBC predictors) to Model 2A (in which descriptive SN were added) shows that the latter fits the data better, as evidenced by the higher Nagelkerke R², the lower Akaike Information Criterion (AIC), and the statistically significant Δχ² change. This was also the case when we compared Model 1 to Model 2B (which includes both injunctive and descriptive SN). The relevance of the SN predictors was further corroborated by the results of Model 3, as shown in Table 3. If one’s friends practice phone recycling, then that person is more than 2.4 times more likely (odds ratio) to recycle than not.

While we established that descriptive SN influences one’s intention to recycle, starting from Model 2A (when it was first introduced) to Model 3, this was not the case for injunctive SN, which turned out to be statistically nonsignificant. Thus, Hypothesis 2 was partially confirmed. We also found that while the impact of descriptive SNs on the intention to recycle was contingent on the social referent, this did not turn out to be the case for injunctive SNs; thus, Hypothesis 3 was also partially confirmed.

As follows from Figure 1, where only the two extreme data series are depicted in color for the sake of making the differences more clearly visible, for those who strongly agree that their friends recycle their out-of-use mobile phones, wanting to behave as their friends increases the probability of recycling from 0.12 to 0.60. This is a much more pronounced increase than that experienced by those who strongly disagree that their friends recycle (from 0.15 to 0.50) (confirming Hypothesis 4). On the other hand, the interaction between one’s injunctive SN and their ‘motivation to comply’ with friends and family was statistically nonsignificant. To put it simply, the influence of family/friends’ opinions on one’s intention to recycle was the same regardless of their willingness to behave the way their family/friends want them to (rejecting Hypothesis 5).

5. Conclusions

In this work, we employed the Theory of Planned Behavior to account for an individual’s intention to engage in mobile phone recycling. Although this is one of the most widely used theories for explaining pro-environmental behaviors, it has rarely been used in the context of mobile phone recycling, and existing research has yielded conflicting results. Furthermore, since previous research has found Subjective Norms to be the most inconsistent predictor of recycling, we wished to examine why this may have been the case. In particular, we hypothesized that this was not due to the irrelevance of the Subjective Norms, as suggested by some researchers, but rather to their suboptimal measurement. This problem may occur in two ways: on the one hand, by employing single measurements, and on the other hand, by using nonrelevant social referents. To counter the first issue, in this study, we employed four SN’s predictors to measure an individual’s descriptive and injunctive normative beliefs relative to their family members and closest friends. To address the second issue, we considered the referent’s importance to the individual by measuring the extent to which the respondent wants to behave as (or to do what) the referent (does).

Our results show that all of the predictors are relevant in regard to phone recycling. ‘Attitude’, ‘Perceived Behavioral Control’ and, most importantly (as it is our focus), ‘Subjective Norms’ were found to have a positive and statistically significant impact on recycling intention, although, as we mentioned in the ‘Context and Sample’ section, the self-selective nature of our sample could have weakened any existing correlations. However, we would like to point to two limitations of the present study. First, owing to the self-selective nature of our sample, the descriptive statistics presented in the Results section are not generalizable to the Greek population. Thus, future studies interested in identifying the true magnitude of mobile phone recycling in Greece—and the reasons it does (not) occur—should employ random sampling. Second, while our models explained a considerable proportion of the variation in the dependent variable (i.e., mobile phone intention), as evidenced by Nagelkerke’s R scores (ranging from 25.6% for Model 1 to 34% for Model 3), these very same results suggest that more explanatory variables are needed to predict recycling behavior more accurately. Thus, on the one hand, we recommend that future research use more proxy variables to tap into latent TPB constructs; for example, one’s concern over data security may be used as an extra proxy variable for the ’Attitude‘ construct. On the other hand, researchers may obtain more accurate results by coupling the psychological constructs of the TPB with demographic variables such as age, economic status, educational attainment, and possibly gender. The demographic variables may plausibly impact one’s intention to recycle either directly or indirectly, through the standard TPB predictors-(see, for example, [36]).

Our findings also suggest two study avenues for future research. On the one hand, the results highlight the importance of measuring an individual’s endorsement of societal standards in tandem with their Subjective Norms to obtain better and more accurate predictions of behavioral intentions. In the original formulation of the TPB [17], the relevance of Subjective Norms in explaining a behavior rested on the idea that an individual behaves in a certain way because of their endorsement of a socially expected mode of conduct. However, the stress here is not on the ‘socially expected’ but on the ‘endorsement’. Simply asking respondents what any of their social referents (wish them to) do misses this point: an individual may be well aware of what these referents (expect them to) do and, at the same time, may be indifferent to these examples/expectations. A case to the point is the overall low correlation coefficients between the descriptive Subjective Norms and the ‘Identification with’ referent variables (as well as between injunctive Subjective Norms and ‘motivation to comply with’ the referent), which we found in the present study (see Table 2). As shown by our results, ignoring one’s endorsement of socially prescribed practices may result in suboptimal, or even biased, results.

On the other hand, our results suggest that future research should examine in more detail the interplay and relevance of descriptive and injunctive Subjective Norms in explaining mobile phone recycling. This suggestion may come as a surprise given that in our study, we found, contrary to our expectations, that only descriptive, and not injunctive, Subjective Norms are a statistically significant predictor of recycling intentions. Given that previous research has established that descriptive and injunctive norms’ effects differ depending on the particular setting/behavior (e.g., [37]), one may assume that our results prove that injunctive SNs are irrelevant in regard to mobile phone recycling. While this may be the case, the substantial correlations between descriptive and injective Subjective Norms we observed in our sample (see Table 2), as well as their changing statistical significance as we moved from Model 1 to Model 3 (see Table 3), support an alternative explanation: as far as mobile phone recycling in concerned, descriptive and injunctive SNs are congruent. Therefore, it is important to conduct more studies in different national settings to determine which of the two is the case.

Our findings also have practical implications for all those interested in promoting mobile phone recycling, such as policy-makers, mobile phone companies, and environmental NGOs. Our results indicate that the strongest influence on one’s intention to recycle their mobile phone comes from one’s social surroundings (see Table 3, Model 3). Overall, if people think that their social milieu is engaged in recycling, they are more likely to engage in this behavior themselves. This suggests that if we wish to enhance mobile phone recycling, then informational, educational, and/or advertising campaigns by state actors and environmental NGOs should present mobile phone recycling as a casual behavior performed by ordinary people from all societal backgrounds and sectors. In a similar vein, certified mobile phone recycling outlets could relocate their recycling bins to outside their premises so that onlookers may witness their fellow citizens engaging in this behavior. This latter point is also related to the next influential contextual factor we have identified in our research: the knowledge of where to recycle one’s obsolete mobile phone is lacking. Making such recycling facilities more prominent and better communicating their location and existence will have a positive effect on recycling intentions. Finally, public awareness campaigns should focus on and stress those attitudinal predictors we found to positively influence recycling intentions: presenting mobile phone recycling as a ‘necessary’ step to protect the environment, as a ‘responsible’ behavior, and as a ‘satisfying’ action can affect how people actually engage in recycling.