The theory of planned behaviour is often used to explain the green consumption behaviour of consumers (Khan et al., 2019; Hameed et al., 2019b). Researchers have suggested that the TPB can be broadened by involving external variables that comprehensively explain phenomena (Ajzen, 1991; Perugini and Bagozzi, 2001). To strengthen the underpinning of the theory involving products with recycled packaging, this study integrates significant predictors (environmental concern, knowledge, and value) related to antecedents of attitude (Onel and Mukherjee, 2015b). Moreover, recent studies have observed the significant impact of religiosity on ecological behaviour (Ghazali et al., 2018) because preservations of environment and natural resources are found in guidance book of every religion (Ali et al., 2019a).

The TBP further suggests that individual behaviour significantly relies on attitudes and the social influence of society. These behavioural beliefs stimulate consumption behaviour towards eco-friendly products. Religion is an important construct linked with attitude, and it determines consumer intentions and behaviour. Therefore, it would be interesting to know whether such relationships are contingent on consumer religiosity (Cyril De Run et al., 2010).

Environmental value (EV) has been defined as values that are actions oriented inclined towards the environment (Sony and Ferguson, 2017) and considered as determinants of individual’s attitudes and behaviours. Earlier literature primarily focused on three distinct aspects of individual’s environmental beliefs, i.e.; values, attitudes, and concerns; where Values has been used to define as ‘internal standards’ that exceeds in certain circumstances, and further assist to formulate certain attitudes and behaviours (Sony and Ferguson, 2017). Values initially affect a person’s attitude, which subsequently influences his or her behaviour; this pattern is called the value-attitude-behaviour hierarchy (Homer and Kahle, 1988; Sihombing, 2007).

Altruistic values and egoistic values influence environmental concerns (Schultz and Zelezny, 1998). Counterintuitively, such perspective seems more influential to generate emotional appeals and facilitate marketers to tailor better offerings for green products. So, the environmental value is only created if the person has positive attitude with the environment. Furthermore, this suggests that people with stronger environmental values have positive attitudes towards eco-friendly products (Suki, 2016). Therefore, a consumer who values the man-nature orientation will exhibit an attitude that supports green product consumption. In the light of the TPB framework and the above literature, we proposed the following hypothesis:

The environmental knowledge (EK) has been defined as individual’s ‘cognitive ability’ for comprehension of the environmental sustainability issues related to soil and water pollution, waste generation and recycling problems, and further their impact on physical environment specifically on the society at large (Saeed and Azmi, 2018). Earlier literature has primarily emphasised that environmental knowledge [ENK] is an important predictor of green consumerism (Wu et al., 2017). Researchers have argued that ENK directly affects consumers’ attitudes and purchase intentions towards ecological products; thus, higher levels of perceived environmental knowledge lead to increased purchase behaviour related to eco-friendly products (Yadav and Pathak, 2016).

Cerri et al. (2018) argued that ecologically conscious individuals are more sensitive to the environment than others; consequently, ENK influences an individual’s purchase intentions towards green products. Therefore, it is plausible that perceived environmental knowledge could be linked with attitude and intentions for the consumptions of the green products. Considering the above discussion, this study seeks to analyse the fundamental association between environmental knowledge and attitudes towards eco-friendly products. We proposed the following hypothesis:

Generally, environmental concerns (EC) influence consumers’ intentions to resolve environmental issues (Prakash and Pathak, 2017). Furthermore, they improve consumers’ sense of responsibility and passion regarding the protection of the environment at the individual level (Dagher and Itani, 2014). Hence, the consumer’s individual level environmental concern provoke the more alertness at the personal level that give way to the collective orientation towards pertaining environmental issues and the readiness to resolve them, ranging from waste recycling behaviour (Zhao et al., 2014) to green buying behaviour (Paul et al., 2016).

Previous studies have emphasized that environmental concerns are significantly related to attitudes towards eco-friendly packaging products that strengthen behaviours related to this issue (Yadav and Pathak, 2016). In addition, they are an efficient driver of environmentally friendly behaviours (Pagiaslis and Krontalis, 2014; Kautish et al., 2021). Therefore, environmental concerns can be viewed as a critical factor that affects the attitudes of customers towards eco-friendly packaging products. Thus, we proposed the following hypothesis:

In the sphere of marketing, attitude is a process that guides consumers’ decisions regarding whether they will buy a specific product (Bhuian et al., 2018). Consumers’ attitudes towards intentions and purchasing decisions involving green products represents their awareness of environmental sustainability, which is different from their general environmental attitudes (Jaiswal and Singh, 2018). Several scholars have linked consumers’ awareness of environmental sustainability with green purchase intentions (Lai and Cheng, 2016) and purchase behaviour (Yadav and Pathak, 2016). Previous studies suggest that individuals with strong green attitudes are more inclined to purchase eco-friendly products (Joshi and Rahman, 2015). Therefore, if a consumer’s attitude towards a product is positive, his or her purchase intentions will be affected. Hence, consumers’ positive attitudes towards green products stimulate their readiness to purchase them (Cheah and Phau, 2011; Khan et al., 2021). Therefore, these positive attitudes are considered significant factors in stimulating green purchase intentions. Thus, we proposed the following hypothesis:

Subjective norms (SN) refer to the perceived social pressure placed on individuals to behave in a certain way due to the influence of their families, relatives, and close friends (Ajzen, 1991). Subjective norms have been found to be a significant construct in pro-environmental consumer behaviour (Yadav and Pathak, 2016). Herbes et al. (2018) found that the influence of subjective norms is culturally specific, and collective societies are more inclined to behave according to the influence of relevant reference groups. Hence, according to the consumer behaviour literature, an individual under the influence of subjective norms behaves in a positive way to create a positive social image among his peers, friends and family (Kumar et al., 2017). Consequently, subjective norms play an influential role in predicting consumers’ intentions to purchase specific products (Noble et al., 2009).

Furthermore, subjective norms have a significant relationship with attitudes and purchase intentions towards green products (Kim and Karpova, 2010), and are linked with consumer intentions (Paul et al., 2016) towards products with recycled packaging. Thus, based on the assumption of TPB, while purchasing recycled packaging products, consumers are influenced by opinion of the surroundings influential group of people. So, the above discussed literature proposed the following hypothesis:

Perceived behaviour control (PBC) refers to how individuals’ opinions influence certain behaviours (Ajzen, 2002). Yeon Kim and Chung (2011) found that PBC is associated with green purchase intentions. Hassan et al. (2016) stated, if the significant others are not inclined towards purchase of a specific product it will pose a psychological barrier for the customer to purchase it and ultimately negatively effect on individuals PBC. Therefore, we proposed that a high level of PBC will positively affect consumers’ green purchase intentions. Based on the TPB and the literature discussed above, the following hypothesis is formulated:

Behavioural intentions (BI) represent the readiness of an individual to engage in certain behaviour. A high level of intention towards a specific behaviour is likely to stimulate that behaviour (Ajzen, 1991; Husnain and Toor, 2017). In consumer psychology, behavioural intentions are a key factor in predicting consumers’ actual behaviour. Hence, incorporating green purchase intentions into pro-environmental studies is an important way to gauge actual eco-friendly behaviour (Yadav and Pathak, 2016; Khan et al., 2021). Kim et al. (2020) argued that intentions are greatly linked with attitudes to perform in a particular manner. So, the dedicated attitude will enhance the intentions of the consumers to make an actual purchase decision related to the behaviour.

Similarly, green products create environmental awareness among consumers, as purchasing is an important element of behavioural changes. Therefore, the literature reveals that purchase intentions are an essential element of determining behaviour (Lai and Cheng, 2016). Therefore, we proposed the following hypothesis:

Since last few decades, religiosity as a research area has been extensively studied in western context. Research studies appeared in this domain were mainly referred and considered to Christianity faith (Raggiotto et al., 2018; Mortimer et al., 2020; Wang et al., 2020a). In literature, religiosity is differentiated from religion and can be defined as pattern of living that reflect in attitude, knowledge and values of the societal members (Akhtar et al., 2020b; Tsoraya et al., 2022). In turn, such cognitive beliefs and mannerism forms behaviours and practices of individuals and society (Samad et al., 2021; Tsoraya et al., 2022). Consequently, individual high on Islamic religiosity can dampen to low rate of deviant behaviour such as recycled packaged goods consumption. Considering the undeniable importance of Islamic religiosity in the context of pro-environmental behaviour, we examined the role of Islamic religiosity as a moderator in the relationship between the TPB constructs and purchase intention. In this research, we viewed Islamic religiosity as a moderator for the following reasons. First, the idea of viewing Islamic religiosity as a moderator between the TPB constructs and the intention to buy recycled products was taken from contingency theory, which proposes that the relationship between two variables is dependent or contingent on a third variable. Therefore, the involvement of Islamic religiosity as a moderator in the examined relationship leads to a better understanding of the related phenomena and helps avoid misleading ideas regarding contingency relationships among the TPB constructs. Similar to the way that contingency theory (Venkatraman, 1989) contributes to the understanding of inconsistent attitudes, SN and PBC contribute to the understanding of intentions. Second, a recent studies on pro-environmental behaviour (Bhuian et al., 2018; Soomro, 2019) suggested that the moderating role of Islamic religiosity should be considered in the context of the relationships between the antecedents of TPB and consumers’ purchase intentions towards eco-friendly products. Moreover, researchers in the fields of marketing and consumer behaviour agree that the concept of Islamic religiosity is significant in thoroughly explicating pro-environmental consumer behaviour (Setyawati et al., 2020; Nickerson et al., 2022). However, very little research has been specifically concerned with demonstrating the role of religiosity in non-western contexts, i.e., Malaysia and Pakistan (Wang et al., 2020a). Finally, in the context of green IT, PBC has been viewed as a significant factor influencing green consumption (Dezdar, 2017) in Malaysia.

The studies that focus on investigating the correlation between religion and the intention to buy have validated the notion that religion significantly affects an individual’s preferences (Graafland, 2017). However, the extent to which religion affects an individual varies from person to person. Ultimately, it is not only a person’s religion but also the passion and intensity with which he or she holds his or her religious beliefs and affiliation (religiosity) that matter in the context of his or her intention to buy. Therefore, religiosity as a whole affects an individual’s thinking, lifestyle, and buying decisions. Similarly, Stern (2000) suggested that consumers with a high level of religiosity exhibit strong pro-environmental attitudes that consequently affect their intentions to buy.

Islamic Religiosity could be a significant construct to analyse in the settings of Islamic societies such as Pakistan and Malaysia, particularly in the context of consumer attitudes towards products with recycled packaging. Figure 1 shows the theoretical framework of our study. Based on the TPB framework and the literature discussed above, the following hypotheses were proposed:

A time-lagged approach was used to collect the data at three points in time. Temporal research designs view the causes as preceding the effects in time; this makes temporal precedence a necessary provision for causation in a model (Cook et al., 1979). It is very convenient in the social sciences to determine time lags and researchers opts such design to minimize social desirability bias (Cole and Maxwell, 2003; Husnain et al., 2021, 2022). The constructs of pro-environmental attitudes, subjective norms and perceived behaviour control were measured at Time 1, those of religiosity and the intention to buy recycled packaging products were measured at Time 2 (2 weeks after Time 1), and that of consumer purchase behaviour was measured at Time 3 (2 weeks after Time 2). The data were collected between April and June 2019 in three rounds. For time 1, data on all the independent and moderating variables were collected during the first 3 weeks of April; 2 weeks after Time 1, data were collected from the same respondents for the mediating variables; and 2 weeks after Time 2, data were collected for all the dependent variables. A non-random purposive sampling technique was used. The units of analysis examined were students who are currently enrolled in different semesters of under graduation and post-graduation programs, i.e., environmental science, in different universities of Pakistan and Malaysia. The foremost reason for employing this students sample was that young people have high need for uniqueness with more inclination towards adoption of new ideas than others (Ottman et al., 2006) and have more concern on recyclability issues (Nuojua et al., 2022). The respondents that were recruited were offered modest compensation to ensure their participation in the study. A personally administered questionnaire was devised to collect the statistical data required for the study. For the Pakistani sample, initially, 500 questionnaires were distributed at time 1, and 418 of these were received back; 372 questionnaires were returned at time 2, and 340 were returned at time 3. Of these, 324 were found to be complete in all aspects and useable for the study. Likewise, in Malaysia, 300 questionnaires were given to the respondents at time 1, and 290 were returned; 286 questionnaires were returned at time 2, and 278 were returned at time 3. Of these, 266 were considered useable. A total of 590 questionnaires formed the sample used for the study.

Present study has employed combination of several scales that were adapted and modified with respect to cross-cultural context of the research (See Appendix 1). Such adaption to construct items is desirable to reinforce the respondent understanding, whose intrinsic meanings can manifest differently across settings (Rudran and Kumar, 2017). Meanwhile adapted constructs scale and items were evaluated by known expert researchers to ensure face validity. All the constructs were measured on a 5-point Likert scale; 1 denoted “strongly disagree” and 5 denoted “strongly agree.” Environmental concern was measured using a 4-item scale adapted from the work of Sivek (2002). Environmental knowledge was measured with an 8-item scale proposed by Haron et al. (2005). Environmental values were measured with 4 items adopted from the work of De Groot et al. (2007). A measurement scale adapted from the work of Huang et al. (2004), which contained 5 items, was employed to measure attitudes. Subjective norms were measured using 4 items adopted from the work of Ajzen (1991). To measure perceived behaviour control, a 3-item scale taken from the study of Martinho et al. (2015) was used. To conceptualize religiosity, a 13-item scale was adopted from the work of Eid and El-Gohary (2015). The behavioural intention to buy recycled packaging products was measured using 5 items adapted from the work of Zeithaml et al. (1996). Actual environmental consumer behaviour was measured using 4 items adapted from the work of Schlegelmilch et al. (1996).

The sample of this study was composed of 453 respondents from Pakistan (N = 310) and Malaysia (N = 143). The demographic details of the respondents are shared in Table 1. To mitigate potential systematic errors, gender and education were included in the structural model as control variables.

To verify the constructs’ independence and to validate the discriminant validity of the study variables, a confirmatory factor analysis was performed. Appendix 1 shows that the significant factor loadings >0.50 (Hair et al., 2014) and that the magnitude of the average variance extracted (AVE) > 0.5 as recommended by Bagozzi and Yi (2012). While this confirmatory factor analysis (CFA) was performed, to obtain better model fit indices, several items, i.e., one item regarding environmental concern (EC), two items regarding subjective norms (SN), one item regarding religiosity (R), and three items regarding behavioural intentions (BI), were excluded due to poor factor loadings. The discriminant validity of all the constructs in Table 2 shows that the square root of the correlations was less than the AVE estimates (Fornell and Larcker, 1981). Table 3 shows the results regarding both countries and demonstrates that the hypothesized model has a good fit (χ²/Df = 1.72, p < 0.01; root mean square residual [RMR] = 0.046; goodness fit index [GFI] = 0.91; comparative fit index [CFI] = 0.88; and root mean square error of approximation [RMSEA] = 0.051).

In this study, several diagnostic analyses and procedures were conducted to control for common method bias (CMB). To control potential effects of self-reporting bias, survey contains neutrally worded questions to ensure answer options are not leading (Podsakoff et al., 2003; Brannick et al., 2010). To reduce the potential for CMB, the present study used a time-lagged design.

Two statistical techniques, Harman (1967) and marker variable were used to examine the potential effects of CMB. Applying Harman’s one factor test, all the factors were constrained into a single factor and results showed that sole factor can only explain 23 and 29% of total variance for Pakistani and Malaysian sample, respectively, indicating that there is no problem of CMB. Additionally marker variable approach, in which common variance was calculated by getting square root of each path to common factor. Results indicate that estimated common variance was below 50% for both samples (35% in the Pakistan, 28% in the Malaysia) further providing assurance that there is no issue of common method bias in this study.

The discriminant validity of the variables was established by contrasting the hypothesized model with several three-factor models and a single-factor model. The comparison results corresponding to the models (Table 3) of both countries, i.e., Pakistan and Malaysia indicated that the additional models examined, including a three-factor model combining attitude, subjective norms, perceived behaviour control and religiosity [χ²/Df = 14.95; RMR = 0.14; GFI = 0.53; CFI = 0.59; and RMSEA = 0.02], a three-factor model combining intention towards recycled products and behaviour [χ²/Df = 24.15; RMR = 0.23; GFI = 0.45; CFI = 0.61; and RMSEA = 0.13] and a single-factor model [χ²/Df = 34.23; RMR = 0.13; GFI = 0.22; CFI = 0.33; and RMSEA = 0.37], did not fit the data as well as the hypothesized model. Thus, the discriminant validity of the hypothesized variables was supported.

The present study uses structural equation modelling technique to test the hypotheses using maximum likelihood (ML) estimation method. This method is capable to obtain estimation for parameters form multivariate normal distribution as well as non-normally distributed data. For the structural model, the model fit indices for the Pakistani data (χ²/Df = 1.830; RMSEA =0.047; CFI = 0.914; GFI = 0.902; TLI = 0.894; and AGFI = 0.79) and the Malaysian data (χ²/Df = 1.836; RMSEA = 0.051; CFI = 0.87; TLI = 0.76; GFI = 0.81; and AGFI = 0.90) were examined. The RMSEA is considered one of the most informative criteria for examining how well a hypothesized model fits observed data (Bagozzi and Yi, 2012); hence, a good error of approximation indicates a good fit. Thus, the results indicate that the data fit the hypothesized model well.

Table 4 shows the explanatory power (R²) of the model’s estimates of the endogenous constructs. For the Pakistani sample, it is evident that the proposed model exhibits a high level of explanatory power for environmental attitude (R² = 0.41), while it exhibits a relatively low level of explanatory power for behavioural intentions (R² = 0.26) and buying behaviour towards recycled products (R² = 0.62). Similarly, for the Malaysian sample, the proposed model shows a high level of explanatory power for environmental attitude (R² = 0.39) and buying behaviour towards recycled products (R² = 0.46) but a lower level of explanatory power for behavioural intentions (0.21). Hair et al. (2011) proposed that the R² values of 0.75, 0.50, or 0.25 can be employed to indicate substantial, moderate, or weak explanatory power, respectively. Therefore, per these standards, one value corresponds to substantial explanatory power, five values to moderate explanatory power and two to weak explanatory power. Moreover, regarding the hypothesis tests for both samples, Table 4 shows the respective significance levels of the path coefficients.

First, considering the Pakistani sample, the findings with respect to environmental values, (β = 0.17 and p < 0.01), knowledge (β = 0.44 and p < 0.01), and concern (β = 0.31 and p < 0.05) provide good support for H1a, H1b and H1c, respectively. Environmental attitudes and subjective norms were found to have significant positive effects on intentions (β = 0.59 and p < 0.01; β = 0.28 and p < 0.05), confirming H2 and H3, while perceived behavioural control was found to be insignificant at p > 0.05 (H4 was not supported).

In the Malaysian sample, similar results were found in terms of confirming the predictors of environmental attitude as environmental values (β = 0.11 and p < 0.05), environmental knowledge (β = 0.26 and p < 0.01), and environmental concern (β = 0.17 and p < 0.01). Similarly, environmental attitude and subjective norms had significant positive effects on intentions (β = 0.55 and p < 0.01; β = 0.22 and p < 0.05), supporting H2 and H3. Similar to the Pakistani sample, perceived behavioural control did not have an effect on intentions; thus, H4 was not supported in the Malaysian sample.

A moderated regression analysis was used to test the moderating role of religiosity. To examine the moderating effects suggestion of Gaskin (2012b) was followed, the interaction term of each relationship was produced by multiplying the standardized value of the moderator, i.e., religiosity, with each predictor’s standardized value (environmental values, environmental knowledge, and environmental concern). As indicated in Table 4, the interaction between environmental attitude and religiosity (REL_x_ATT) was positively related to intentions in the Pakistani and Malaysian samples (β = 0.22^* and p < 0.05; β = 0.14^* and p < 0.05). These findings are in line with H6. Regarding H7, the interaction between social norms and religiosity (REL_x_SN) was found to be significant in both the Pakistani and Malaysian samples (β = 0.19^** and p < 0.01; β = 0.10^* and p < 0.05). Finally, as shown in Table 4, the interaction between perceived behavioural control and religiosity (REL_x_PBC) did not moderate the relationship between perceived behavioural control and intentions in either country’s sample (β = 0.08 and n.s; β = 0.01 and n.s). Hence, H8 was not supported.

Gaskin (2012a) suggestion was followed in interpreting the moderating effects. As Figures 2A,B show, it was concluded that the patterns of religiosity’s interactive effects were consistent with the prediction of H6; in particular, environmental attitude (shown on the horizontal axis) had a stronger positive effect on the intention to use recycled products (shown on the vertical axis) when the examined consumers showed high levels of religiosity (shown on the straight line) than when the consumers exhibited low levels of religiosity (shown on the dotted line). Thus, H6 was supported for both samples. Similarly, Figures 2C,D show that the pattern of interaction was consistent with the prediction of H7; that is, subjective norms (shown on the horizontal axis) had a stronger positive effect on the intention to use recycled products (shown on the vertical axis) when the consumers showed high levels of religiosity (shown on the straight line) than when they exhibited low levels of religiosity (shown on the dotted line). Hence, H7 was supported.