This study aims to examine food waste behaviours among university students in Indonesia by adapting the theory of planned behaviour (TPB) alongside the theory of domains of knowledge.
A total of 518 students at five Indonesian universities participated in an online survey, answering questions about their food waste behaviours based on the TPB and the domains of knowledge frameworks. Among them, 250 students had taken a course that covered the food waste topic, while 268 students had not taken the course. The data were later analysed by using partial least squares-structural equation modelling and measurement invariance across the composite models.
The findings reveal that environmental awareness, attitude towards food waste behaviours, subjective norms and different types of knowledge which are procedural knowledge, declarative knowledge, personal and social knowledge significantly influence food waste behaviour. Students who attended a course related to the food waste topic were more likely to align their intentions to reduce food waste with their reduction behaviour.
While previous research has highlighted the importance of the TPB in understanding food waste intentions and behaviour, studies also show that students’ knowledge about food waste plays a crucial role in shaping these intentions. This study builds on these insights by combining the TPB with the domains of knowledge theory, specifically looking at how participating in food waste-related courses influences food waste behaviours among Indonesian students.
Introduction
In its report for the UN Decade of Education for Sustainable Development, UNESCO highlighted the vital role of education for sustainable development (ESD) in equipping people from all walks of life to confront and solve pressing issues that threaten the sustainability of planet Earth (UNESCO, 2005). Within the framework of Agenda 2030, UNESCO emphasised that ESD is a crucial component of the sustainable development goals (SDGs) on quality education, acting as a key for all other SDGs. It empowers individuals to acquire the knowledge, skills, values and attitudes needed for fostering sustainable development (UNESCO, 2019).
Among these goals, SDG (12.3) specifically targets responsible consumption and production with a global objective to halve food loss and waste (FLW) by 2030. Presently, approximately one-third of the food produced for human consumption is lost or wasted globally (Food and Agriculture Organization, 2011). The geographical context of the current study is Indonesia. A recent research report published by the Indonesian Ministry of National Development Planning reveals that food loss and waste during the period of 2000–2009 reached an alarming 115–184 kg per capita annually, resulting in economic losses equivalent to 4%–5% of Indonesia’s gross domestic product in terms of economic loss (Indonesian Ministry of National Development Planning, 2021). Furthermore, the total FLW-associated emissions during those periods were estimated at 1.702,9 Mt CO2 eq. Two key contributors to this FLW problem named in the Ministry report are lack of information/education for food workers and consumers and food waste-related behaviour.
The purpose of this study is to examine food waste behaviours among university students in Indonesia. By exploring the attitudes, knowledge and behaviours of Indonesian university students’, it is possible to assess the impacts of ESD on students and to pinpoint the issues that need to be addressed to develop more effective ESD programmes that encourage, and support behaviour change in Indonesia’s food waste context. The data for the study come from a project where courses at five Indonesian universities were updated and modernised to include content related to food waste management.
Several theoretical approaches have been taken to analyse FLW-related behaviours, with the TPB being one of the most prominent. The TPB posits that intentions precede and explain behaviour shaped by factors such as subjective norms, attitude towards behaviour and perceived behavioural control (Ajzen, 1991). This theory has also been applied in the Indonesian context. For instance, Ariyani and Ririh (2020) suggested that the model should consider environmental awareness, and education plays an important role in doing so. The domains of knowledge approach in sustainability education research (Kaiser and Fuhrer, 2003; Redman and Redman, 2014) identifies key areas that can be influenced through education. This theory has not, however, been empirically tested in Indonesia before. In addition, combining the two theoretical approaches (the TPB and the domains of knowledge) is a fruitful path yet to be explored.
Literature review
Previous studies on sustainable behaviours from a theory of planned behaviour perspective
The TPB proposes that intentions reflect the motivational factors that influence a behaviour. In addition, perceived behavioural control directly predicts behavioural achievement or outcomes (Ajzen, 1991). Since the original publication of the theory, numerous studies have used it to predict human behaviour. In the context of food waste, the theory has been adapted by incorporating factors like food purchasing routine and religious traditions (Aktas et al., 2018). Environmental awareness has also emerged as a key predictor of behaviours such as green product purchases (Sharma and Foropon, 2019) and eco-conscious consumerism (Hameed et al., 2019). In Pakistan, the theory has been applied to investigate the organic food purchasing intention (Al-Swidi et al., 2014).
In Indonesia, the TPB has been expanded to predict food waste behaviour by including environmental awareness, government intervention and purchasing habits (Ariyani and Ririh, 2020). Findings revealed that environmental awareness significantly influenced behaviour, subjective norms and perceived behavioural control. In addition, personality traits such as extraversion and agreeableness have been shown to affect behavioural intention related to food waste (Jamaludin et al., 2020). A more recent study on students in the Philippines extended the TPB by incorporating food waste knowledge, which was found to predict intention to reduce food waste (Valentin et al., 2024). Table 1 provides a brief overview of previous studies conducted in the Asian developing countries that used the TPB to explore sustainable behaviour.
Previous studies using the TPB on the issue of sustainable behaviour in developing countries in Asia
| Authors/ year | Country | Analysis methods | Key results |
|---|---|---|---|
| Al-Swidi et al., 2014 | Pakistan | CFA-SEM | Subjective norms moderate the effect of perceived behavioural control on organic food purchasing intention |
| Aktas et al., 2018 | Qatar | PLS-SEM | Factors such as Ramadhan, food choice and financial attitude influence food waste behaviours, in addition to the intention to reduce food waste |
| Hameed et al., 2019 | Pakistan | PLS-SEM | Intrinsic religious orientation and green trusts significantly influence attitudes towards green products although these attitudes do not necessarily translate into eco-conscious behaviour |
| Sharma and Foropon, 2019 | India | Multiple regression analysis; analysis of variance | Green purchasing behaviours are predominantly affected by product attributes. Education had no significant impact on either conditional or unconditional green purchase |
| Ariyani and Ririh, 2020 | Indonesia | PLS-SEM | Environmental awareness alongside attitude towards behaviour, subjective norms and perceived behaviour control, significantly impacts food management intentions |
| Jamaludin et al., 2020 | Malaysia | PLS-SEM | Extraversion and agreeableness influence behavioural intention for food waste management in addition to attitude, subjective norms and perceived behavioural control |
| Valentin et al., 2024 | Philippines | Multiple regression analysis; Simple linear regression analysis | Food waste knowledge, attitude towards food waste reduction, subjective norms, and perceived behavioural control predict intentions to reduce food waste. These intentions, in turn, predict actual food waste reduction behaviour |
PLS-SEM = Partial least squares-structural equation modelling;
CFA-SEM = Confirmatory factor analysis-structural equation modelling
Domains of knowledge in sustainability education
To inspire sustainable practices around food and waste, it is vital to understand how knowledge is framed within educational systems. Beyond the traditional fact-based approach to learning (declarative and procedural), subjective knowledge such as effectiveness and social knowledge, needs to be emphasised, as they evolve with local and cultural differences in perceptions, beliefs and desires (Redman and Redman, 2014). Declarative knowledge, often ecological in focus, involves understanding how ecosystems function and the ways in which human activities affect the environment. Procedural knowledge, on the other hand, refers to “how to” information, equipping individuals with the skills to take action. The effectiveness knowledge reflects people’s perceptions of whether a certain behaviour is desirable or effective, while social knowledge concerns people’s understanding of others’ motives and expectations (Redman, 2013). While conceptually promising, these types of knowledge lack strong empirical backing, particularly in developing countries. Studies that have explored these knowledge domains suggest their collective impact is crucial in changing behaviour. High levels of declarative knowledge alone do not consistently lead sustainable behaviours (Redman and Redman, 2014). Instead, subjective and procedural knowledge are particularly important for changing behaviour (Redman and Redman, 2016). More research is still needed to further explore the relationships between the knowledge domains and sustainable behaviours.
Theoretical framework and hypotheses
The proposed model draws from the TPB and the theory of domains of knowledge as seen in Figure 1. The behaviour of interest is defined as sustainable food waste management behaviours over the course of a year, such as composting, sorting waste and engaging in the community discussions. This behaviour is influenced by an individual’s intention to act, which is shaped by factors such as environmental awareness, attitude towards behaviours, subjective norms and perceived behaviour control.
Environmental awareness (AWR) is defined as an individual’s recognition of the food waste phenomenon and its broader impacts, encompassing economic, social and environmental concerns (Grob, 1995; Rasool et al., 2021; Zimmer et al., 1994). Attitude towards behaviour (ATB), on the other hand, refers to how favourably or unfavourably individuals perceive a particular action (Ajzen, 1991). This attitude serves as a strong indicator of an individual’s willingness to perform a certain pro-environmental behaviour. ATB can be assessed through three key dimensions: economic impacts, health issues and safety issues (van der Werf et al., 2021; Visschers et al., 2016).
Subjective norms (SNO) encompass external influences such as familial, parental, cultural, religious norms and societal norms during parties and holidays. These norms shape how individuals perceive the influence of their surroundings to perform particular behaviour (Botetzagias et al., 2015; Greaves et al., 2013; Russell et al., 2017). The TPB suggests that perceived behavioural control, which refers to an individual’s belief in their ability to perform a particular action (Ajzen, 1991), aligns closely with the concept of procedural knowledge in knowledge theory. Procedural knowledge (PROC), defined as the information that enables an individual to take action (Kaiser and Fuhrer, 2003), is closely intertwined with perceived behavioural control. As such, these two constructs have been merged in this study to represent an individual’s capacity to act, particularly in relation to composting, repurposing leftovers and reducing food waste by purchasing and cooking habits (Ariyani and Ririh, 2020). The development of effective control over one’s behaviour requires both knowledge and the acquisition of knowledge and skills, which can only be attained through considerable effort (Bandura, 2001). Hence, perceived behavioural control is shaped by an individual’s procedural knowledge, and conversely, perceived behavioural control (self-efficacy) may enhance the acquisition of procedural knowledge (Schunk and Pajares, 2002). Given their reciprocal nature, both have been consolidated into a single construct.
Effectiveness knowledge (EFFE) further influences behaviours by shaping how individuals perceive the impact of their actions on the environment or others (Redman and Redman, 2014). Declarative knowledge (DECL), which focuses on understanding how environmental systems operate, adds another layer to this relationship (Kaiser and Fuhrer, 2003; Redman and Redman, 2014). Social knowledge (SOC), derived from observing the motives and behaviours of others, also plays a role in shaping individual actions (Kaiser and Fuhrer, 2003; Redman and Redman, 2014).
As illustrated in Figure 1, environmental awareness plays a pivotal role in fostering intention. Awareness is not only the understanding and being knowledgeable regarding the environmental issues but also participating actively in environmental organisations. It serves as a catalyst for developing positive attitudes and engagement in environmentally friendly actions (Karatekin, 2014). Individuals with a higher level of environmental awareness are more likely to engage in intentional pro-environment behaviours, as they understand the causal relationships and potential impacts of their actions (Mei et al., 2016). Therefore, assessing the level of awareness and concern towards environmental issues is crucial.
From an economic perspective, managing organic waste holds significant value, as it contains substantial levels of energy and nutrients, making it highly recoverable and economically beneficial. Many countries encourage individuals to manage or at least separate the organic and inorganic waste due to its economic advantages (de Sadeleer et al., 2020). By separating and managing waste either individually or communally, individuals may reduce waste collection fees, save money and repurpose organic waste into compost for personal use. Thus, economic benefits can act as a driving force behind the intention to manage food waste.
As awareness of the environmental and social issues caused by food waste grows within communities, it is crucial to examine whether the subjective norm that has been formed within the community also influences the intentions to manage organic waste. Subjective norms are defined as the perceived social influences or pressures that encourage or discourage individuals from engaging in a specific behaviour (Ajzen, 1991). These norms reflect individuals’ beliefs about how their actions might be viewed by their reference groups (Al-Swidi et al., 2014). Research has shown a significant impact of subjective norms on individual’s intention (Tarkiainen and Sundqvist, 2005). Furthermore, procedural knowledge – understood as series of step-by-step instructions to complete a task – plays a pivotal role in shaping behaviour (Anderson et al., 2001). This type of knowledge is demonstrated through actions rather than through conscious recollection, such as the ability to drive a car. Mastery of procedural knowledge is often associated with a stronger intention to perform a certain action (Makransky and Petersen, 2021).
In addition to factors such as environmental awareness, attitude towards behaviour, subjective norms and procedural knowledge on behaviour, intention is expected to mediate the relationships among variables in the TPB. The mediation effect is significant because it can uncover the underlying the various social processes that further clarify the relationship within such models like the TPB (Povey et al., 2000; Wang et al., 2020). The study aims to investigate the mediation effect of intention, effectiveness knowledge, declarative knowledge and social knowledge on behaviour, with consideration given to environmental awareness, attitude, subjective norms and procedural knowledge. To explore individuals’ intention and behaviours regarding food waste reduction and to promote food waste management in their everyday routines, the following hypotheses were developed:
Environmental awareness positively and significantly influences intention to manage food waste.
Attitude towards behaviour positively and significantly influences intention to manage food waste.
Subjective norms positively and significantly influence intention to manage food waste.
Procedural knowledge positively and significantly influences intention to manage food waste.
Intention positively and significantly influences behaviour to manage food waste.
Procedural knowledge positively and significantly influences behaviour to manage food waste.
Declarative knowledge positively and significantly influences behaviour to manage food waste.
Effectiveness knowledge positively and significantly influences behaviour to manage food waste.
Social knowledge positively and significantly influences behaviour to manage food waste.
Data collection and analysis
Methods and design
As part of capacity-building initiatives for higher education institutions under the CBHE Erasmus+ framework, an interdisciplinary consortium IN2FOOD (Interdisciplinary Approach Towards Fostering Collaborative Innovation in Food Waste Management) was established in 2021 attempting to foster collaborative innovations in food waste management. One of the goals of this project was to modernise 30 existing courses taught in Indonesian universities to address issues related to food waste management. The expertise provided by EU partners in areas such as user innovation, food waste management and sustainable food systems enabled the Indonesian partners to creatively design their courses to equip future generations with skills needed to tackle societal challenges, particularly those related to food waste issues. The survey was conducted in March and April 2022 as a part of the modernised courses, and the respondents were students on those courses. In addition, a control group consisting of students who had not been exposed to these modernised courses was also surveyed during the same period, this control group provided a comparative basis for the educational intervention.
This study used a survey based on the TPB and the theory of domains of knowledge. The survey was co-designed by the authors drawing on previous research (Aktas et al., 2018; Ariyani and Ririh, 2020; Redman and Redman, 2014). Administered in the Indonesian Language, the survey reached university students aged 18–23 and was implemented in March and April 2022. The respondents were Indonesian students from all consortium members, including those who has taken the modernised courses and those who had not participated in any environmental awareness courses. Most of these students resided in urban and populated areas such as the Jakarta Metropolitan, Bandung Agglomeration and Malang Areas. At the time of the survey, most students lived with their parents and families because of the pandemic situation.
The survey measured variables using reflective items on 1–5 scales as seen in Appendix 1. Some items were in the reversed scales ensuring that value 1 means the most unfavourable and value 5 means the most favourable in terms of food waste-related items. The data collected from 518 complete responses (85% completion rate) were analysed to assess the impact of the modernised courses on students’ attitudes, knowledge and behaviours. Of these responses, 250 students were part of the ESD1 group, while 268 students were part of the control group ESD0. Demographic statistics for these groups are summarised in Table 2.
Instrument validity and reliability
The initial phase of the analysis focused on ensuring the validity and reliability of the instrument. Using the partial least squares (PLS) algorithm in the SMART-PLS software, items with low outer loadings (less than 0.5) were literary removed. To confirm the discriminant validity, the cross-loading matrix and HTMT ratio were examined. A high correlation between the constructs EFFE and SOC indicated that both might measure the same factor. Consequently, these two constructs were combined into one renamed Personal and Social Knowledge (PERSOC), which encompasses two dimensions, i.e. perceived personal effectiveness and their social ability to influence others. In addition, items BEHA1 and PROC2 were eliminated due to strong loading (above 0.5) on unintended constructs. The final items included in the new construct are detailed in Table 3. As a result, H8 and H9 were consolidated into H8*.
Personal and social knowledge
| Item | Statement | Five-point scale (1–5) |
|---|---|---|
| EFFE1 | How would you rate your ability to reduce the amount of waste you produce through your food purchasing decisions? | Poor − excellent |
| EFFE2 | How would you rate your ability to reduce the amount of food waste your household produces? | Poor – excellent |
| EFFE4 | I believe that the actions of an individual are central for achieving sustainability | Strongly disagree – strongly agree |
| SOC1 | How would you rate your ability to influence members of the household/family to take action towards reducing food waste? | Poor – excellent |
| SOC2 | I admire people who are conscious about their food waste decisions | Strongly disagree – strongly agree |
| SOC3 | Reducing food waste makes me feel good | Strongly disagree – strongly agree |
H8*: Personal and social knowledge positively and significantly influences behaviour in managing food waste.
After rerunning the PLS algorithm, the revised discriminant validity showed an improved HTMT ratio, with all values below 0.85 as shown in Table 4. The cross-loading matrix also demonstrated improvement with no items correlating with unintended constructs (see Table 5), confirming adequate discriminant validity for all constructs.
Discriminant validity (HTMT ratio)
| Construct | ATB | AWR | BEHA | DECL | INT | PERSOC | PROC | SNO |
|---|---|---|---|---|---|---|---|---|
| ATB | ||||||||
| AWR | 0.34 | |||||||
| BEHA | 0.167 | 0.367 | ||||||
| DECL | 0.231 | 0.679 | 0.533 | |||||
| INT | 0.441 | 0.676 | 0.42 | 0.636 | ||||
| PERSOC | 0.415 | 0.588 | 0.348 | 0.566 | 0.735 | |||
| PROC | 0.144 | 0.431 | 0.649 | 0.772 | 0.725 | 0.643 | ||
| SNO | 0.47 | 0.632 | 0.295 | 0.543 | 0.837 | 0.725 | 0.602 |
Cross loading
| Item | ATB | AWR | BEHA | DECL | INT | PERSOC* | PROC | SNO |
|---|---|---|---|---|---|---|---|---|
| ATB1R | 0.844 | 0.184 | −0.099 | 0.125 | 0.263 | 0.236 | 0.012 | 0.189 |
| ATB2R | 0.87 | 0.239 | −0.076 | 0.143 | 0.272 | 0.27 | 0.009 | 0.271 |
| ATB3R | 0.712 | 0.165 | −0.109 | 0.109 | 0.196 | 0.183 | −0.033 | 0.192 |
| AWR1 | 0.178 | 0.764 | 0.233 | 0.331 | 0.348 | 0.302 | 0.188 | 0.295 |
| AWR2 | 0.147 | 0.729 | 0.213 | 0.379 | 0.33 | 0.286 | 0.169 | 0.301 |
| AWR3 | 0.231 | 0.689 | 0.136 | 0.42 | 0.322 | 0.311 | 0.17 | 0.265 |
| AWR4 | 0.142 | 0.681 | 0.191 | 0.35 | 0.302 | 0.297 | 0.185 | 0.152 |
| BEHA2 | 0.002 | 0.257 | 0.724 | 0.377 | 0.246 | 0.336 | 0.334 | 0.128 |
| BEHA3 | −0.036 | 0.261 | 0.715 | 0.345 | 0.273 | 0.297 | 0.36 | 0.175 |
| BEHA4 | −0.184 | 0.147 | 0.835 | 0.315 | 0.208 | 0.111 | 0.38 | 0.099 |
| BEHA5 | −0.145 | 0.162 | 0.817 | 0.339 | 0.222 | 0.177 | 0.355 | 0.14 |
| DECL1 | 0.163 | 0.219 | 0.255 | 0.655 | 0.294 | 0.243 | 0.362 | 0.21 |
| DECL3 | 0.11 | 0.294 | 0.255 | 0.663 | 0.31 | 0.243 | 0.285 | 0.231 |
| DECL4 | 0.207 | 0.347 | 0.188 | 0.666 | 0.36 | 0.336 | 0.298 | 0.296 |
| DECL5 | 0.087 | 0.46 | 0.448 | 0.84 | 0.392 | 0.427 | 0.495 | 0.241 |
| DECL6 | 0.071 | 0.471 | 0.363 | 0.758 | 0.299 | 0.359 | 0.417 | 0.203 |
| EFFE1 | 0.204 | 0.309 | 0.19 | 0.337 | 0.374 | 0.732 | 0.312 | 0.311 |
| EFFE2 | 0.166 | 0.325 | 0.253 | 0.401 | 0.351 | 0.789 | 0.363 | 0.285 |
| EFFE4 | 0.216 | 0.262 | 0.106 | 0.272 | 0.362 | 0.559 | 0.183 | 0.288 |
| INT1 | 0.285 | 0.258 | 0.077 | 0.28 | 0.644 | 0.399 | 0.148 | 0.349 |
| INT2 | 0.181 | 0.336 | 0.308 | 0.354 | 0.725 | 0.299 | 0.452 | 0.279 |
| INT3 | 0.054 | 0.231 | 0.395 | 0.258 | 0.58 | 0.167 | 0.353 | 0.218 |
| INT4 | 0.245 | 0.349 | 0.085 | 0.303 | 0.668 | 0.36 | 0.205 | 0.33 |
| INT5 | 0.232 | 0.299 | 0.114 | 0.274 | 0.627 | 0.397 | 0.167 | 0.313 |
| PROC1 | −0.048 | 0.189 | 0.442 | 0.476 | 0.246 | 0.209 | 0.838 | 0.173 |
| PROC3 | 0.131 | 0.204 | 0.164 | 0.315 | 0.267 | 0.407 | 0.517 | 0.245 |
| PROC4 | −0.005 | 0.167 | 0.297 | 0.323 | 0.432 | 0.324 | 0.725 | 0.225 |
| SNO1 | 0.284 | 0.287 | 0.132 | 0.28 | 0.318 | 0.362 | 0.176 | 0.715 |
| SNO2 | 0.233 | 0.258 | 0.129 | 0.227 | 0.351 | 0.301 | 0.225 | 0.762 |
| SNO3 | 0.016 | 0.193 | 0.104 | 0.137 | 0.273 | 0.209 | 0.164 | 0.592 |
| SOC1 | 0.128 | 0.249 | 0.302 | 0.322 | 0.268 | 0.785 | 0.302 | 0.286 |
| SOC2 | 0.313 | 0.332 | 0.121 | 0.287 | 0.391 | 0.588 | 0.143 | 0.318 |
| SOC3 | 0.309 | 0.326 | 0.173 | 0.27 | 0.448 | 0.635 | 0.186 | 0.341 |
Note: *EFFE and SOC are now in the same construct (PERSOC)
Internal reliability was evaluated using composite reliability (CR) and average variance extracted (AVE). Cronbach’s alpha was not used due to the reflective nature of the measurement (Fornell and Larcker, 1981). Table 6 shows that the composite reliability (CR) for all constructs exceeded 0.7, with AVE values above 0.4, and some constructs surpassing 0.5. This indicated acceptable reliability, as composite reliability is above 0.7, and outer loading of more than 0.5 for all items.
Construct reliability and validity
| Construct | Cronbach’s alpha | rho_A | Composite reliability | Average variance extracted (AVE) |
|---|---|---|---|---|
| ATB | 0.74 | 0.764 | 0.852 | 0.658 |
| AWR | 0.683 | 0.686 | 0.808 | 0.514 |
| BEHA | 0.775 | 0.773 | 0.857 | 0.6 |
| DECL | 0.777 | 0.821 | 0.842 | 0.519 |
| INT | 0.657 | 0.662 | 0.785 | 0.423 |
| PERSOC | 0.789 | 0.833 | 0.841 | 0.473 |
| PROC | 0.533 | 0.6 | 0.742 | 0.499 |
| SNO | 0.455 | 0.469 | 0.733 | 0.481 |
The model’s goodness-of-fit was measured by the standardised root mean square residual which equals to 0.083, slightly exceeding the 0.08 threshold, yet still indicating an acceptable fit. In addition, the model avoided overfitting, suggesting a parsimonious structure. This model fit suggests that the data aligns well with the model. The resulting measurement model, including outer loadings for each indicator and the composite reliability for each construct, is illustrated in Figure 2.
Path analysis
Upon establishing validity and reliability, basic statistics at the construct level can be seen in Table 7. These statistics explore students’ attitude towards food waste (ATB), awareness of the food waste problem (AWR), intention to reduce food waste (INT) and engagement in waste reduction behaviours such as composting, sorting waste or participating in social activities related to it (BEHA). The lowest mean value was found in BEHA (2.412), while ATB (4.051) and AWR (4.103) showed relatively high means. This suggests that although students are aware of and have a favourable attitude towards food waste reduction, they have yet to frequently engage in such behaviours. Social norms (SNO) supporting these behaviours also scored relatively high (3.932), indicating a supportive environment. Knowledge domains revealed the lowest mean is for procedural knowledge (PROC) at 3.059, while personal and social knowledge had the highest mean at 3.868.
Statistical values for the different construct variables
| Construct | Mean | Median | SD | Excess kurtosis | Skewness |
|---|---|---|---|---|---|
| ATB | 4.051 | 4 | 0.713 | 1.581 | −0.898 |
| AWR | 4.103 | 4.022 | 0.546 | 0.608 | −0.509 |
| BEHA | 2.412 | 2.262 | 0.851 | −0.018 | 0.514 |
| DECL | 3.628 | 3.612 | 0.629 | −0.259 | 0.024 |
| INT | 3.816 | 3.816 | 0.534 | −0.12 | 0.138 |
| PERSOC | 3.868 | 3.888 | 0.575 | 0.006 | −0.051 |
| PROC | 3.059 | 3 | 0.707 | 0.201 | 0.3 |
| SNO | 3.932 | 4 | 0.612 | −0.215 | −0.182 |
The coefficient of determination for behaviours was 0.266, and for intention, it was 0.414, indicating that variations in these factors can be moderately predicted by the independent variables. Using bootstrapping technique, the significance of path coefficients is detailed in Table 8.
Significance of path (first path model)
| Path | Original sample (O) | Sample mean (M) | Standard deviation (STDEV) | T-statistics (|O/STDEV|) | P-values |
|---|---|---|---|---|---|
| ATB → INT | 0.169 | 0.174 | 0.04 | 4.235 | 0.000 |
| AWR → INT | 0.251 | 0.251 | 0.043 | 5.873 | 0.000 |
| DECL → BEHA | 0.266 | 0.267 | 0.049 | 5.376 | 0.000 |
| INT → BEHA | 0.068 | 0.066 | 0.051 | 1.325 | 0.093 |
| PERSOC → BEHA | 0.048 | 0.051 | 0.043 | 1.093 | 0.137 |
| PROC → BEHA | 0.251 | 0.254 | 0.049 | 5.136 | 0.000 |
| PROC → INT | 0.335 | 0.336 | 0.041 | 8.18 | 0.000 |
| SNO → INT | 0.216 | 0.215 | 0.044 | 4.942 | 0.000 |
All hypotheses, except H5 and H8*, were supported with p-values below 0.05. Notably, the newly formed construct from the theory of knowledge domains does not predict behaviours (H8*). The TPB was also not confirmed in this study (H5). Following the TPB, it is posited that this new construct may influence behaviour through the mediation of intentions (INT). Furthermore, all knowledge from the domains of knowledge, including declarative knowledge, may effect behaviour through intention (Liu et al., 2022). Path significance is provided in Table 9 and the total effects are in Table 10.
Path significance (second path model)
| Path | Original sample (O) | Sample mean (M) | Standard deviation (STDEV) | T-statistics (|O/STDEV|) | P-values |
|---|---|---|---|---|---|
| ATB → INT | 0.126 | 0.125 | 0.039 | 3.216 | 0.001 |
| AWR → INT | 0.169 | 0.167 | 0.045 | 3.762 | 0.000 |
| DECL → INT | 0.087 | 0.087 | 0.05 | 1.751 | 0.040 |
| INT → BEHA | 0.142 | 0.136 | 0.052 | 2.747 | 0.003 |
| PERSOC → INT | 0.215 | 0.218 | 0.055 | 3.877 | 0.000 |
| PROC → BEHA | 0.374 | 0.377 | 0.044 | 8.521 | 0.000 |
| PROC → INT | 0.232 | 0.231 | 0.047 | 4.919 | 0.000 |
| SNO → INT | 0.17 | 0.171 | 0.047 | 3.634 | 0.000 |
Total effects
| Path | Original sample (O) | Sample mean (M) | Standard deviation (STDEV) | T-statistics (|O/STDEV|) | P-values |
|---|---|---|---|---|---|
| ATB → BEHA | 0.018 | 0.017 | 0.008 | 2.214 | 0.014 |
| ATB → INT | 0.126 | 0.125 | 0.039 | 3.216 | 0.001 |
| AWR → BEHA | 0.024 | 0.023 | 0.012 | 2.043 | 0.021 |
| AWR → INT | 0.169 | 0.167 | 0.045 | 3.762 | 0.000 |
| DECL → BEHA | 0.012 | 0.012 | 0.009 | 1.311 | 0.095 |
| DECL → INT | 0.087 | 0.087 | 0.05 | 1.751 | 0.040 |
| INT → BEHA | 0.142 | 0.136 | 0.052 | 2.747 | 0.003 |
| PERSOC → BEHA | 0.03 | 0.03 | 0.014 | 2.24 | 0.013 |
| PERSOC → INT | 0.215 | 0.218 | 0.055 | 3.877 | 0.000 |
| PROC → BEHA | 0.407 | 0.409 | 0.041 | 9.982 | 0.000 |
| PROC → INT | 0.232 | 0.231 | 0.047 | 4.919 | 0.000 |
| SNO → BEHA | 0.024 | 0.023 | 0.01 | 2.43 | 0.008 |
| SNO → INT | 0.17 | 0.171 | 0.047 | 3.634 | 0.000 |
The study confirms both the TPB and theory of knowledge domains in the model, with the exception of declarative knowledge (DECL). Procedural knowledge strongly influences intentions to reduce food waste (H4), as well as behaviours (H6).
Multigroup analysis
To examine measurement invariance across the composite models, a three-step procedure was used to analyse the difference between students exposed to sustainability courses and those who were not (Hair et al., 2017; Henseler et al., 2016). The first step confirmed that all groups had configural invariance, meaning that the composite was specified equally for all groups. The second step, assessing compositional invariance, revealed no significant differences between the groups. The results show that all original correlations are greater than 0.05 and their p-values show insignificant differences. Hence, the composite is not formed differently across the two groups which support the compositional invariance. The third step, using a permutation algorithm, indicated some invariance to test if the data groups have statistically significant differences in their parameter estimates. Permutation p-values for both mean and variance show that there is some invariance, therefore a multigroup analysis is then suitable for further analysis, whose bootstrapping results can be seen in Table 11.
Bootstrapping results for multigroup analysis
| Path | Path coefficients mean (ESD0) | Path coefficients mean (ESD1) | p-value (ESD0) | p-value (ESD1) | Invariance |
|---|---|---|---|---|---|
| ATB → INT | 0.135 | 0.117 | 0.013 | 0.012 | No |
| AWR → INT | 0.171 | 0.133 | 0.004 | 0.017 | No |
| DECL → INT | 0.085 | 0.072 | 0.114 | 0.184 | No |
| INT → BEHA | 0.036 | 0.228 | 0.352 | 0.001 | Yes |
| PERSOC → INT | 0.209 | 0.266 | 0.011 | 0.000 | No |
| PROC → BEHA | 0.398 | 0.344 | 0.000 | 0.000 | No |
| PROC → INT | 0.188 | 0.253 | 0.013 | 0.000 | No |
| SNO → INT | 0.19 | 0.168 | 0.003 | 0.002 | No |
In Table 11, it can be seen that there is a significant difference in the path coefficients for students exposed to modernised courses with food waste-related content (ESD1) versus those students who were not exposed to those courses (ESD0). The group’s path from intention to behaviours was significant only for students who took the modernised courses (p-value = 0.001). This suggests that their intention to reduce food waste aligns with their behaviours, while the students who did not take such courses showed divergence (p-value = 0.352). Other path coefficients did not differ between the groups, indicating that factors such as environmental awareness, attitude towards behaviours, subjective norms, procedural knowledge, personal and social knowledge significantly affect intentions and behaviours for food waste reduction. Declarative knowledge, however, did not have meaningful effects.
Discussion
This study integrates the TPB with the theory of domains of knowledge, exploring how food waste-related course participation affects food waste reduction behaviours among the Indonesian students. The findings indicate that attitudes towards behaviours, subjective norms and perceived behaviour control (procedural knowledge) are reliable predictors of individual intentions and food waste reduction behaviours, which is aligned with the TPB. However, the theory of domains of knowledge does not directly affect behaviours but operates through the mediation of intentions, which is not elaborated by previous research (Redman and Redman, 2014). Declarative knowledge, which reflects familiarity with the environmental and food waste issues, did not influence behaviours for food waste reduction. It is worth noting that procedural knowledge which corresponds to perceived behaviour control has the strongest effect to intentions and behaviours. In general, Indonesian students continue to rely on procedural knowledge that demonstrates the ability to perform the desired food waste behaviours.
It is important to note that in this study, effectiveness knowledge and social knowledge have been combined into a single construct. Effectiveness knowledge is primarily related to the perceived individual ability to perform an action, while social knowledge pertains to interpersonal relationships. In the context of Indonesian students, this combination may be explained by Indonesia’s cultural emphasis of collectivism where personal and social aspects of life are closely intertwined (Hofstede, 1983).
The findings of the study clearly demonstrate that ESD has a significant role to play in reaching the SDGs, particularly in reducing food waste (Redman and Redman, 2014). Students who were exposed to modernised course content on food waste reduction exhibited positive differences in their intentions to engage in behaviours that reduce food waste. As their understanding of the issues increased, encompassing procedural, personal and social knowledge, their future behaviour is likely to be further influenced in a positive direction.
This research highlights the importance of integrating the food waste issue into the education system, particularly through curricula that foster sustainability competences, which include different types of knowledge. Previous research has highlighted that focusing solely on students’ declarative knowledge is insufficient (Redman and Redman, 2014). Educational content should also incorporate assignments to present holistic solutions and practical case studies that address various types of knowledge simultaneously.
Conclusions
The key determinants of students’ behaviour in food waste reduction include attitude towards behaviours, subjective norms, environmental awareness, procedural knowledge, personal and social knowledge. The implication of the findings confirms both the TPB and the theory of knowledge domains, with the exception of declarative knowledge. Furthermore, it is concluded that knowledge domains influence behaviours indirectly through intentions, reaffirming the TPB.
This research, although exploratory in the context of food waste behaviours in Indonesia, suggests that future studies with larger sample sizes are needed to confirm the complex relationship between independent variables from these two theories. Further inquiries should investigate the gap between intentions and behaviours among students who are not exposed to sustainability courses, as well as the role of teachers’ personal attitudes in fostering and developing pro-environmental behaviours (Elorinne et al., 2020). A pre- and post-measurement of students before and after sustainability courses could offer more robust insights into the impact of such education initiatives.
The authors wish to thank the teachers at five Indonesian universities who modernised their courses during the IN2FOOD project as well as the students who responded to our questionnaire.
Funding: This research was co-funded by the European Commission through the Erasmus+ CBHE project “Resolving A Societal Challenge: Interdisciplinary Approach Towards Fostering Collaborative Innovation in Food Waste Management” (IN2FOOD), project number 618717-EPP-1-2020-1-ID-EPPKA2-CBHE-JP. The European Commission’s support for the production of this publication does not constitute an endorsement of the contents, which reflect the views only of the authors, and the Commission cannot be held responsible for any use which may be made of the information contained therein.
References
Appendix 1. List of questionnaire items
The following questions were evaluated on a five-point scale Strongly disagree (1) – Strongly agree (5):
Food waste creates economic damage to people (AWR1).
Food waste creates social damage among individuals who waste food and those who have no food (AWR2).
I think that phenomenon of food waste is widespread (AWR3).
Food waste is harmful to the environment (AWR4).
Food waste is not a problem for the environment as it is natural and biodegradable (reversed) (AWR5).
I am not worried about the cost of food that I throw away (reversed) (ATB1).
I am not worried about the amount of food thrown away (reversed) (ATB2).
I think it is better to throw away leftovers rather than to risk gaining weight (reversed) (ATB3).
I think it is better to throw away leftovers than to risk eating unsafe food because it is no longer fresh (reversed) (ATB4).
I only throw away food if the expiry date has passed (ATB5).
I know the difference in meaning between the label “use by” and “best before” (ATB6).
My parents or other family members do not like wasting food (SNO1).
Wasting food is considered a sin (SNO2).
Taking home leftover food from parties and restaurants are considered as normal (SNO3).
Having surplus food is more preferable than out of food during parties and events (reversed) (SNO4).
Taking home leftover food is only for cheap people (reversed) (SNO5).
Abundant amount of food is a way of celebrating Idul Fitri or Christmas (Hari Raya) (reversed) (SNO6).
I intend to reduce food waste (INT1).
I intend to recycle my food waste into something useful (INT2).
I intend to eat leftover food (INT3).
I intend to store and reheat the leftover food (INT4).
I intend to make food purchases more mature (INT5).
Food that is thrown away is a very small part of the amount of waste sent to the landfill in Indonesia (Reversed) (DECL8).
Leaving food on my plate is OK with me (Reversed) (EFFE3).
I believe that the actions of an individual are central for achieving sustainability (EFFE4).
It is the responsibility of the government to reduce the problem of food waste. (Reversed) (EFFE5).
I admire people who are conscious about their food waste decisions (SOC2).
Reducing food waste makes me feel good (SOC3).
Respondents were asked to report on a five-point scale Never heard of (1) – Heard of and know a lot about (5), how familiar they are with the following concepts:
Composting (DECL1)
Recycling (DECL2)
Sustainability (DECL3)
Food waste (DECL4)
Respondents were asked to rate their knowledge about the following on a five-point scale of Poor (1) – Excellent (5):
Effects of food waste on the environment (DECL5).
How much food is thrown away in Indonesia (DECL6).
Causes and drivers of food waste in Indonesia (DECL7).
Composting from food waste (PROC1).
Respondents were asked to rate their ability to do the following actions on a five-point scale of Poor (1) – Excellent (5):
…to manage a composting system at home (PROC2)
…to evaluate whether food is edible or not (PROC3)
…to create new meals from leftovers or scraps (PROC4)
…to plan food purchases and/or meals in advance (PROC5)
…to reduce the amount of waste you produce through your food purchasing decisions (EFFE1)
…to reduce the amount of food waste your household produces (EFFE2)
…to influence members of the household/family to take action towards reducing food waste (SOC1)
Respondents were asked on a five-point scale of Never (not at all) (1) – Always (>90% of the time) (5) how often they had made the following actions over the past year:
Composted their food waste at home (BEHA1)
Correctly sorted their garbage (BEHA2)
Discussed the issue of food waste with their friends or family (BEHA3)
Participated in a community initiative related to sustainability (e.g. waste reduction) (BEHA4)
Discussed, posted or shared food waste related content on social media (BEHA5)
