Switching from LPG to electricity in Indonesia: a study in the Java Bali grid

Abstract

Technology innovations shift toward using cleaner and more convenient prime energy and energy converter appliances. It has driven some lower-middle-income countries to switch from fossil fuels to electricity. However, Indonesians don’t feel the urgency to switch from liquified petroleum gas (LPG)-fueled cookware to electricity. This study aims to: (1) explore usefulness (USE), ease of use (EOU), and time-oriented convenience (TOC) as antecedents of functional value (FUNC), recognition value (RECV), emotional value (EMOV), and epistemic value (EPIV); (2) examine the moderating effect of conditional value (CODV) on the intention to switch cooking energy (INT). A cross-sectional study gathered 290 valid questionnaires from households in Jakarta and other provincial capitals in Java.

The PLS-SEM approach was applied to examine hypotheses and analyze gender, age, and region segmentation. The results reveal the impact of both USE and TOC on FUNC, RECV, and EMOV. Moreover, EOU and TOC positively influence EPIV. The linear relationship between USE and EPIV and between RECV and INT is insignificant. However, a quadratic relationship exists. This study shows that conditional value moderates the relationship between emotional value and switching intention. Additionally, gender has an impact on the influence of EOU and TOC on EPIV, indicating the presence of heterogeneity. These findings contribute to conceptual advancement in behavior science by integrating TAM, TCV, and VAM. Furthermore, it illuminates relationships between USE, EOU, TOC, consumption values, and intention. In addition, it gives professionals in the consumer electronics industry greater confidence in developing product appeal, segmenting products, and creating persuasive promotions.

IMPACT statement

In recent decades, electricity has become increasingly important as a driving force for societal economic growth and the advancement of prosperity through the enjoyment of life. Switching to electric kitchen appliances is an enormous challenge for the Indonesian government, as most people rely on LPG for cooking energy. We investigate the attractiveness of induction stoves to people using LPG with a power capacity of 1300 VA or higher in the Java-Bali grid. Specifically, we examine how usefulness, ease of use, and time-oriented convenience influence multidimensional consumption values (functional, recognition, emotional, and epistemic) and switching intentions controlled by conditional value. Practitioners are encouraged to focus more on product appeal characteristics and gender concerns when designing products and advertising strategies to increase market switching intention.

Keywords:

• switching intention
• consumption value
• household energy
• VAM

SUBJECTS:

• Power & Energy
• Asian Studies
• Consumer Psychology
• Statistics for Social Science
• Sociology
• Urban Studies
• Urban Cultures

REVIEWING EDITOR:

• Kaouther Kooli, Bournemouth University, United Kingdom

1. Introduction

It is crucial to win customers’ loyalty by comprehending consumer behavior at every decision-making stage—before, during, and after consumption. The migration theory is frequently To analyze switching behavior in marketing phenomena (Cheng et al., 2019), (Chang & Chen, 2014) is frequently applied. The theory suggests that moving to a new place is prompted by discomfort in the place of origin as a push factor and the attractiveness of the destination as a pull factor (Zeng et al., 2021). Implementing the energy transition, particularly household energy, poses challenges for developing countries. The government implements various economic strategies and incentives to lower switching barriers. Reducing the current energy supply has been proven effective in driving the conversion process (Budya & Yasir Arofat, 2011) (Figure 1).

LPG-fueled stoves are extensively used in Indonesian households for daily cooking (Budya & Yasir Arofat, 2011). Indonesia is one of the developing countries that has unbalanced domestic LPG supplies. According to Indonesia’s Ministry of Energy and Mineral Resources (2020), LPG imports are expected to increase by over 80% to meet rising domestic demand, as illustrated in Figure 2. Nevertheless, there has been a rise in household customer demand for increased installed power capacity since 2017 (Hilmiyati-Mas’adah et al., 2023). The program to convert one million LPG stoves to induction stoves is expected to lower LPG imports by 136.8 thousand tons annually, cut LPG subsidies by 778.7 billion IDR annually, and save US$59.9 million per year (PLN, 2022).

Moreover, due to the overwhelming presence of subsidized competing products in the market, it becomes challenging to gauge the impact of induction stoves’ attractiveness on perceived value and subsequent consumer behavior. An induction stove is mostly selected after comparing its suitability, quality, price, and type because consumers are hesitant and concerned about the quality and performance of white goods purchasing decisions. Therefore, the availability of references and information on the usage and benefits of induction stoves becomes a crucial factor for consumers in their decision-making process (Noverita & Mussry, 2020). To be chosen as an alternative energy converter, it may be more effective to highlight the induction stove’s similarity to the current product, LPG stoves (Arens & Hamilton, 2018). This study would assist in an empirical analysis of consumers’ perceptions and intentions regarding an induction stove as a substitute in Jakarta and three major cities in the Java-Bali grid.

Recently, marketers realized that selling products or services by pursuing potential customers is about building perception in customers’ minds. Perceived value is often used to predict a consumer’s decision to act. It is a determinant factor of adoption intention in the Value-based Adoption Model (VAM) framework (Hsu & Lin, 2018; Talawanich & Au, 2020; Vishwakarma et al., 2020), satisfaction (Diniso & Duh, 2021; Mosavi, 2018) or loyalty (Diniso & Duh, 2021). The theory of consumption value (TCV) explains consumption choices from a value standpoint (Gonçalves et al., 2016). It has been the subject of several studies, including digital marketing (Talwar et al., 2020), green products (Zailani et al., 2019), hospitality (Wei et al., 2017), and higher education (Rivera et al., 2018). In addition, recent studies have examined specific values (Talwar et al., 2020; Yoo et al., 2013) or all dimensions of TCV (Awuni & Du, 2016; Candan & Yıldırım, 2013) to align with the research context. Some studies provide inconsistent evidence regarding the impact of consumption values on intention, such as functional value (Yoo et al., 2013), (Awuni & Du, 2016), recognition value (Yoo et al., 2013), (Zailani et al., 2019), (Kuswandi, 2023), emotional value (Kuswandi, 2023), and epistemic value (Awuni & Du, 2016).

Two main beliefs of the Technology Acceptance Model (TAM) regarding new product acceptance, usefulness (Chu & Lu, 2007; Hsu & Lin, 2018; Sohn & Kwon, 2020), and ease of use (Vishwakarma et al., 2020) significantly impact perceived value. People living in urban areas perceive time as a scarce resource compared to those in rural areas. This leads to a change in the standard of convenience. Consequently, time-oriented convenience affects product value perceptions (Shahijan et al., 2018; Wong, 2021). The significance, however, is not consistent in the contexts of e-commerce usefulness (Chen et al., 2018), ease of use of online music purchasing (Chu & Lu, 2007), airline mobile apps (Talawanich & Au, 2020), service convenience (Wong, 2021), and time convenience in a long queue (Kremer & Debo, 2015). Assessing these research gaps empirically is challenging. Despite the importance of consumption values in understanding consumer behavior, we are aware of only a few empirical studies examining household cooking energy. This study proposes a framework for predicting LPG consumers’ intentions to switch to electricity as their cooking energy. First, this study combines TCV and VAM by exploring four dimensions of consumption value, namely functional, recognition, emotional, and epistemic, in constructing perceived value. Although identifying consumption values is important for understanding consumer behavior, there are few empirical studies examining household cooking energy. Second, VAM is enhanced by integrating time-oriented convenience and TAM’s key beliefs—usefulness and ease of use—into the VAM framework that influences functional, recognition, emotional, and epistemic value. Third, Later, we evaluate the impact of the four consumption values on the intention to change cooking energy. Additionally, the study investigates the impact of conditional value on moderating the intention to switch. The findings will provide some insight into understanding the influential and extrinsic motives that customers take into consideration. The study contributes to business stakeholders by allowing electronic household manufacturers to develop segmented products and marketing strategies that enhance product acceptance.

Therefore, the main objectives of this study are: (1) to identify the product’s benefits (usefulness, ease of use, and time-oriented convenience) that influence functional, emotional, recognition, and epistemic values; (2) to investigate the impact of determinant values on intention to switch household cooking energy; and (3) to examine the moderating effect of conditional value on intention to switch household cooking energy. Furthermore, to increase confidence in the model’s robustness, the proposed model was also evaluated on non-linear correlation and observed segmentation. Accordingly, this research is organized as follows: This current section briefly introduces the research problem and objectives. The second section discusses the literature review and hypotheses development. The third section describes the research methodology, and the fourth section presents the results related to the objectives; the fifth section provides a discussion of the results and their implications and limitations to be considered for future research; and finally, the last section offers conclusions.

2. Literature review and hypotheses

2.1. Product’s benefit and consumption values

Value as experience is the result of the interpretation of one’s own and others’ experiences and future dreams, which is influenced by the interactive exchange of information about intrapersonal and interpersonal values (Helkkula et al., 2012). Assessing value is subjective and constantly changing, so the value is unique and influenced by beneficiaries (Vargo & Lusch, 2008). In the context of electricity consumption, an electronic user is the main actor in the value co-creation process to integrate the value proposed by the manufacturers of electrical household appliances and the power-generating companies. Generating companies deliver value in terms of reliability and availability. Energy converter manufacturers offer comfort, less effort, or convenience in supporting daily activities, as well as other benefits related to electric household appliances.

The Value-based Adoption Model framework measures perceived value as a trade-off of benefit (usefulness, enjoyment) and sacrifice (technicality, perceived fee) to accept alternative products, which played a significant role in the decision to use personal goods (Kim & Choi, 2017). Two main beliefs of the Technology Acceptance Model (TAM) regarding new product acceptance—perceived usefulness and perceived ease of use—are considered relevant benefits for this topic. The term ‘usefulness’ refers to the degree of acceptance based on a subjective assessment (Davis, 1989) of the induction stove that would improve performance, including speed, results, productivity, effectiveness, and smoothness of activities, and reduce post-processing activities such as washing utensils, pans, and so on. It is an extrinsic motivation that pursues the customer’s decision process to obtain benefits (Lee et al., 2005) and significantly affects perceived value (Liao et al., 2022). Perceived usefulness shows different effects by gender and age (Venkatesh et al., 2003). Compared to perceived ease of use, perceived usefulness is more powerful to influence intention directly (Racero et al., 2020) or indirectly (Lee et al., 2005; Liao et al., 2022) across usage periods (Venkatesh et al., 2003). In the context of using an induction stove as a substitute for an LPG-fueled stove, electric cookware manufacturers offer usefulness as a benefit, including improved performance, speeded-up performance, increased work results, productivity, and effectiveness, and reduced cleaning work (Davis, 1989).

Ease of use is the degree to which the individual perceives the induction stove’s operating procedures and interactions as easy to learn and control (Davis, 1989). It shares similarities with variables in other theories, namely effort expectancy in UTAUT (Venkatesh et al., 2003), the opposite meaning of complexity in MPCU (Thompson et al., 1991), or technicality in VAM (Kim & Choi, 2017). Ease of use acts as a gateway for innovation’s acceptance. For instance, ease of use was a determining factor for Indonesians in adopting gas stoves in 2007 (Budya & Yasir Arofat, 2011) and has consistently been an influential factor in converting to induction stove usage (Noverita & Mussry, 2020).

Emerson defines "quality of life" as "the fulfillment of values, goals, and individual needs through the actualization of abilities or lifestyles" (Felce & Perry, 1995). Flanagan (1978) asserts that convenience, both physical and material well-being, is one of the material comforts impacting one’s quality of life. "Use dimension" is a concept of convenience derived from the practical side of use and influenced by technological advancement (Brown, 1990). Convenience emphasizes time-saving for consumers as a benefit of using products or services (Shahijan et al., 2018) while ease of use refers to the time duration spent studying or mastering new technologies or systems (Abbad, 2021). Innovation can assist in lowering the time and effort that users spend on specific activities (Hoyer et al., 2012:420). As a result, this changing behavior establishes a new standard of comfort. Due to lifestyle changes in urban areas, there are higher expectations for the fulfillment of comfort from products and services. Thus, marketing strategies should consider these changing comfort standards. For instance, promotional advertising content illustrates advantages in saving time (Gross & Sheth, 1989) or offers new convenience values (Shahijan et al., 2018). In research on energy conversion processes for households from kerosene to LPG in Indonesia, one of the reasons users are willing to convert is because of the superiority of LPG in terms of cooking duration (Budya & Yasir Arofat, 2011). Time-oriented convenience in this study refers to six categories of Gross and Sheth (1989): fast to use, save time, busy situation, specified time amount, double use of time, and escape a time-oriented lifestyle.

This study explores the multidimensionality of consumption values of TCV in constructing the perceived value of the VAM framework (Ghazali & Mutum, 2020). The theory of consumption value distinguishes functional, social, emotional, epistemic, and conditional values that are perceived by individuals to drive consumption decisions (Gonçalves et al., 2016; Jamrozy & Lawonk, 2017).

Subjectivity colors the composition of consumption values when the decision to consume is made. The functional value appears on the attribute profile of an induction stove and is derived from prominent characteristics or performance. Product reliability and durability are considered to provide quality guarantees and economical operational costs equal to the benefits received (Zailani et al., 2019).

Emotional value is measured by the condition of the feelings and affection generated. This value is obtained from the ability to arouse, trigger, or perpetuate associated feelings, such as emotions of joy, comfort, cheerfulness, sadness, boredom, calmness, nervousness, and fear (Mutum et al., 2021). For example, support for energy-saving campaigns, environmental preservation, and national patriotism. These actions evoke feelings of moral pride.

The term "recognition value" refers to the social recognition defined by the personal image constructed from the responses of various social groups. Opinions are formed because of positive or negative stereotypes held by a specific demographic, social status, culture, or ethnic group (Zailani et al., 2019). When compared to LPG-fueled stoves, using an induction stove improves the user’s self-image and recognition in the social community.

Epistemic value is based on the ability to arouse curiosity, create novelty, or try something new. The latest technological innovation embedded in a new product arouses someone’s curiosity to try it (Mutum et al., 2021). Cooking with no real flame to monitor is still uncommon for Indonesians, and it might be significant in influencing switching intentions.

VAM-based research makes it evident that antecedent factors categorized as benefits have a positive impact on perceived value (Hsu & Lin, 2018; Liao et al., 2022). Determinant benefits that have an impact on perceived value are usefulness (Lian, 2020; Sohn & Kwon, 2020; Vishwakarma et al., 2020) and ease of use, which is the opposite meaning of technicality (Vishwakarma et al., 2020). Meanwhile, recent research focused on service convenience as a benefit that influences perceived value (Pham et al., 2018; Wong, 2021). This study is one of few that examines the impact of usefulness, ease of use, and time-oriented convenience on each dimension of consumption value. Consequently, this study tests the hypotheses as follows:

H1a. Usefulness has a significant effect on functional value.

H1b. Usefulness has a significant effect on recognition value.

H1c. Usefulness has a significant effect on emotional value.

H1d. Usefulness has a significant effect on epistemic value.

H2a. Ease of use has a significant effect on functional value.

H2b. Ease of use has a significant effect on recognition value.

H2c. Ease of use has a significant effect on emotional value.

H2d. Ease of use has a significant effect on epistemic value.

H3a. Time-oriented convenience has a significant effect on functional value.

H3b. Time-oriented convenience has a significant effect on recognition value.

H3c. Time-oriented convenience has a significant effect on emotional value.

H3d. Time-oriented convenience has a significant effect on epistemic value.

2.2. Consumption values and intention to switch

In this study, the intention to switch from LPG to electricity as cooking energy by using an induction stove is applied to measure the impact of its attractiveness on consumption value. Switching to alternative cooking energy could be a complete or partial conversion; thus, intentions are determined by decreasing current product usage, increasing alternative product consumption, and committing to switching action (Chang & Chen, 2014; Wu et al., 2017).

Recent research indicates various compositions of TCV as significant driving factors of behavioral intention. Functional and social values of mobile instant messaging apps influence usage intention (Dhir et al., 2020). Functional, emotional, and social values were found to have significant influences on the intention to use the SOLOMO service (Yang & Lin, 2017). Emotional and epistemic values were confirmed as important factors for the purchase intention of the bamboo product (Yoo et al., 2013). Adoption intention for green products was affected by all values of TCV (Ali et al., 2019). Conditional value, together with the functional, emotional, and epistemic value of biofuels, were significant values to encourage a willingness to pay a higher price (Zailani et al., 2019). On the contrary, the impact of each value of TCV was inconsistent in other empirical studies, for instance, the functional and social values on intention to purchase (Yoo et al., 2013), the social value of willingness to pay (Zailani et al., 2019), functional, conditional, and epistemic values of young customers on green purchase intention (Awuni & Du, 2016). As a result, the researchers put forward 4 hypotheses.

H4. Functional value has a significant effect on the intention to switch.

H5. Recognition value has a significant effect on the intention to switch.

H6. Emotional value has a significant effect on the intention to switch.

H7. Epistemic value has a significant effect on the intention to switch.

2.3. The moderating role of conditional value

Consumers are confronted with conditional value in times of urgency, emergency, special occasions, or stimulus. In the uncertainty of the incumbent product’s availability, the benefits of an alternative product can encourage functional value, social value, and even emotional value under normal circumstances (Zailani et al., 2019). When LPG is scarce, for example, the functional, social, and emotional value of an induction stove has a greater impact on switching intentions than when LPG is widely available. Thus, conditional value is predicted to have a moderating effect on the impact of functional, recognition, and emotional values on behavioral outcomes (Sweeney & Soutar, 2001). This study, therefore, proposed hypotheses that:

H8a. Conditional value moderates the influence of functional value on the intention to switch.

H8b. Conditional value moderates the influence of recognition value on the intention to switch.

H8c. Conditional value moderates the influence of emotional value on the intention to switch.

3. Methodology

Based on the literature reviews, this study proposes a conceptual research framework to demonstrate the effect of an induction stove’s attractiveness (usefulness, ease of use, and time-oriented convenience) and consumption values (functional, recognition, emotional, and epistemic values) on switching intention, with conditional value as a moderator.

3.1. Measures

The proposed hypotheses were verified by an empirical study. We considered the measurement of previous literature and adapted it to suit the topic. To measure usefulness and ease of use, we used the scale of the Technology Acceptance Model (Davis et al., 1989). The convenience of an induction stove was inspired by time-oriented advertising content categories (Gross & Sheth, 1989) and transformed into six indicators (Hilmiyati-Mas’adah et al., 2023). To measure functional value, recognition value, and conditional value, we adapted items previously used by Zailani et al. (2019). We applied items for measuring emotional value and epistemic value proposed by Mutum et al. (2021). Questions related to the intention to switch were adapted from Wu et al. (2017) and Chang and Chen (2014) and modified to fit the topic. All 38 indicators were reflectively measured using a five-point Likert scale (Appendix A). The questionnaire was developed for the Indonesian version, including information about energy consumption characteristics and respondents’ socio-demographics.

3.2. Sampling and data collection

In this research, a pilot test was conducted to validate the survey instrument after three users of the induction stove reviewed the questionnaire items on the pre-test. The primary survey was rolled out from January to March 2022 after revising the wording and format of the questionnaire, gathering a total of 290 valid responses, which is above the minimum sample size of 60 of the 10-times rule method (Hair et al., 2017, 2011). A cross-sectional survey was carried out among households that use LPG for their daily cooking energy and utilize electricity at a minimum of 1300 VA. It applied a purposive and non-probability sampling method targeting people in and outside of Jakarta (i.e., Bandung, Yogyakarta, and Surabaya). Selected respondents were interviewed to get further information about the items of the questionnaire. Descriptive information about the respondents is shown in Appendix B, Table B1.

4. Analysis and findings

This study applied SmartPLS version 3.2.9 (Ringle et al., 2015) to verify the proposed model and hypotheses via partial least squares structural equation modeling (PLS-SEM). Confirmatory composite analysis was applied (Hair et al., 2020) before inspecting the hypotheses to ensure the convergent validity and discriminant validity of indicators. Advanced analyses were applied, including predictive power and linearity tests. To deeply understand customer preferences, segmentation analysis by gender (female vs. male), location (Jakarta vs. other cities), and age (under 45 vs. 45 or above) enrich the findings.

4.1. Measurement model assessment

The measurement model was assessed by utilizing SmartPLS according to four criteria: (a) indicator reliability; (b) internal consistency reliability; (c) convergent validity; and (d) discriminant validity (Hair et al., 2021); convergent validity, indicator reliability, discriminant validity, and composite reliability (Benitez et al., 2020).

1. Indicator reliability relates to how much variance in each indicator can be explained by the construct, including convergent validity and discriminant validity testing. Convergent validity is measured by the outer loading value of each indicator, and discriminant validity is measured by the cross-loading value between indicators (Hair et al., 2021). SmartPLS estimates that none of the outer loadings are less than 0.708 (loading: 0.745–0.949) and that the cross-loading of indicators is greater than that of other constructs. The results as shown in Appendix C, Table C1, indicate that all 38 indicators are valid.

2. Internal consistency reliability is measured by the parameters mentioned in Appendix A. The internal indicator reliability proposed by Dijkstra-Henseler (ρ_A) lies between the extreme values of Cronbach’s alpha with no indicator’s weight and composite reliability (ρc) which considers the weight of each indicator (Hair et al., 2019). The results show that the composite reliability of three constructs, as shown in Appendix C, Table C2, is above 0.95 but the ρ_A less than 0.95 (intention to switch ρc = 0.951; ρ_A=0.939; ease of use ρc = 0.959; ρ_A=0.947 and emotional values ρc = 0.958; ρ_A=0.934). Thus, redundancy is not an issue in each construct (Hair et al., 2019).

3. Convergent validity was examined with AVE. Table C2 in Appendix C shows that all constructs were valid because their AVE was above the threshold of 0.5.

4. Discriminant validity describes a construct that is unique or statistically different from other constructs. Hair et al. (2019) used the heterotrait-monotrait ratio (HTMT) to detect more effectively. As shown in Appendix C, Table C3, all constructs have discriminant validity. HTMT is between 0.386 and 0.824 (below 0.9).

4.2. Structural model assessment

After confirming that the measurement model is valid and reliable, the next step is to evaluate the structural model results. Collinearity among predictor constructs is unlikely to be a critical issue in the structural model because the variance inflation factor (VIF) of predictor constructs is less than the threshold of 5 (Becker et al., 2015). Although the VIF value of emotional value (3.041) is greater than 3 (Appendix C, Table C4), we can conclude that collinearity among predictor constructs is unlikely to be a critical issue in the structural model.

We examined the direct relationship between two constructs in Figure 3 by excluding the moderator from the model (Hair et al., 2021) and running bootstrapping by default on 10000 subsamples using the BCa confidence interval method, a two-tailed test, at a significant level of 5%. To test the interaction effect of conditional value as a moderator, another bootstrapping by including the moderator was executed, as shown in Table 1.

We find that USE has a strong positive impact on EMOV, FUNC, and RECV (H1c, H1a, and H1b were supported). On the contrary, EOU exerts a much lower impact on these three constructs but strongly impacts EPIV (H2d was supported, but H2a, H2b, and H2c were rejected). TOC has a significant effect on all four consumption values, hence hypotheses H3a, H3b, H3c, and H3d were accepted with the strongest positive impact on FUNC. Furthermore, EMOV, FUNC, and EPIV influence the intention to switch (H4, H6, and H7 were accepted, but H5 was rejected). The conditional value shows a moderating effect on weakening the impact of emotional value on switching intention (H8c accepted, H8a and H8b rejected), therefore, the higher the conditional value, the weaker the relationship between these two constructs. The empirical research model result is shown in Figure 3.

The next step is to consider the explanatory power of the structural model by analyzing endogenous constructs’ coefficient of determination (R²), which is a measure of model accuracy (Benitez et al., 2020). The results (Table 2) indicate that the model moderately explains the intention to switch (R² = 0.568), with the moderating effect size of the conditional value considered weak (effect size f² = 2.31%) (Hair et al., 2021).

While implementing a marketing strategy in the field, unexpected market responses often occur. The ability to estimate unpredictable conditions is very useful in mitigating the risk of program failure by using a holdout sample (Hair et al., 2021). The PLS path model as shown in Appendix C, Table C6, has a lower out-of-sample predictive error (the root-mean-square error, or RMSE) compared to the naïve LM model benchmark, mostly for indicators of the dependent variable except for item y5 (the possibility to switch). Accordingly, we conclude that the model has moderate predictive power (Hair et al., 2020).

Quadratic, cubic, exponential, and other models may more precisely reflect real-world market occurrences. However, SmartPLS applied linear equations by default to evaluate the relationship among variables (Ringle et al., 2015). The quadratic effect is an optional analysis applied to this study. The result indicates that two of five insignificant paths (H1d: USE → EPIV and H5: RECV → INT) have a more substantial positive impact in quadratic correlation than in linear correlation. As shown in Table 2, R² EPIV increased from 0.368 to 0.390 and R² INT from 0.568 to 0.578. Thus, the quadratic relationship affects higher model accuracy.

4.3. Observed heterogeneity assessment

Typically, traditional segmentation is based on demographic groups such as gender, age, and geographic location. Perceptions about consumption value may vary depending on the benefit receiver (Talwar et al., 2020). Table 3 presents an analysis of observed heterogeneity by gender, age group, and location.

Cooking energy transformation in Indonesia can be divided into two eras: before the 2007 conversion program from kerosene to LPG in Indonesia (represented by people 45 years and older) and people under 45 years old who mostly used LPG-fueled stoves. The result of the age group analysis shows that all correlations in Table 3 have p-values of more than 0.050. It means that there are no significant differences between the two eras. Similarly, respondents from other provincial capitals shared the same views as Jakarta residents. There was only observed heterogeneity in the female vs. male group. The impact of ease of use and time-oriented convenience on epistemic value show significant differences between the female and male segments (Table 4). The effect of ease of use on epistemic value had the strongest effect on the female group and was stronger than the male group. On the other hand, the impact of time-oriented convenience on epistemic value is the second dominant factor in male group perception, which is stronger than female group perception. According to the female group, besides its functionality, an induction stove’s time-oriented convenience influences its consumption value from the perspective of being emotionally aroused too. The MGA contributes to increasing the suitability of empirical data for the proposed model for the male segment.

5. Discussion

By the "electrifying people’s lifestyles" motto, the conversion program from LPG-based to electric-based appliances is expected to improve Indonesia’s energy sovereignty. Based on the theory of consumption value, this study is one of the few that examines the impact of an induction stove’s usefulness, time-oriented convenience, and ease of use from the perspective of consumption value and its impact on the intention to switch. The findings provide a foundation for understanding the determinants of consumers’ intention to switch their cooking energy from LPG-fueled to electric stoves.

The usefulness of an induction stove indicates a positive influence on functional value, recognition value, emotional value, and a quadratic effect on epistemic value, confirming the previous studies that usefulness is a benefit that has a positive impact on perceived value in general (Hsu & Lin, 2018; Lian, 2020; Lin et al., 2012; Sohn & Kwon, 2020; Talawanich & Au, 2020; Vishwakarma et al., 2020). Table 1 shows that usefulness has the most significant effect on emotional value, which is the most significant on switching intention. It clearly explains that consistent quality of results, durability, and reasonable price of an induction stove leads to good impressions and that the arousal of positive feelings is positively caused by faster and improved performance, increased productivity, and effectiveness of cooking activity. The findings of this study are not in line with Chen and Wu (2018) who found that men’s perception of usefulness does not directly affect their value.

This study indicates that ease of use positively impacts perceived value (Vishwakarma et al., 2020), particularly epistemic value because an induction stove is quite a new appliance for Indonesian households. Epistemic value positively affects switching intention; therefore, ease of use has a significant impact on switching intention to an induction stove, which echoes Noverita and Mussry (2020) findings. Ease of use of the product in early adoption strongly influences women and the elderly (Venkatesh et al., 2003). The findings show that ease of use is strongly related to the female segment’s curiosity and willingness to seek information. The reason is that Indonesian women, like most Asian women, are typically in charge of preparing family meals, with 68.28% of those polled belonging to families with children and cooking for them (see Appendix B, table B1), and they are still less familiar with using an induction stove than an LPG-fuel stove.

Time-oriented convenience is shown to have a positive impact on four consumption values, which exert the greatest influence on its functional value. This study reveals the indirect impact of functional, emotional, and epistemic values in the relationship between time-oriented convenience and switching intention. The results emphasize previous research on which dimension of consumption value influences the switching intention from LPG to electricity (Hilmiyati-Mas’adah et al., 2023). The results partly support previous findings about convenience (time and effort) positively influencing the perceived value of department stores (Wong, 2021) and cruise services (Shahijan et al., 2018). Respondents in a random interview stated that the convenience of saving time or doing double work when using an induction stove has a strong impact on emotional value because it makes her feel like a better person who can accomplish household tasks while also enjoying me-time, or as if she makes a significant personal contribution to something better. On the contrary, men appear to spend less time in the kitchen, so an induction stove that is quick to use would save their scarce and valuable time while also making a good impression on others. The multi-group analysis exposes a heterogeneity: time-oriented convenience has a positive impact on epistemic value for the male group. For each gender, different predictors influence epistemic value. If she found it simple to get an induction stove to do what she wanted it to do, it would increase her desire to know more about the induction stove. In contrast, men’s epistemic values are influenced by a time-oriented convenience factor because of technological invention.

This study reveals that not all consumption values are related to switching intentions. The impact of recognition value on switching intentions was not confirmed. Except for the male group’s perception that recognition value, which affects switching intention, is driven by usefulness and time-oriented convenience, the results indicate that conditional value moderates the impact of emotional value on the intention to switch cooking energy. It is worth mentioning that in Indonesia, electricity subsidies are provided to lower-income families who have been verified through DTKS (Data Terpadu Kesejahteraan Sosial/integrated data on social welfare), unlike subsidized LPG distribution. Customers with 1300 VA and above who are not eligible for electricity subsidies believe that receiving subsidies that are not rightfully theirs might have a negative impact on their moral obligation to switch cooking energy. The results show that the model has moderate predictive power and gender-specific tendencies.

5.1. Theoretical contribution

This research has revealed information about the moderating effects of conditional value on the relationship between emotional value and switching intention. It provides empirical evidence supporting Tanrikulu’s (2021) proposition that conditional value moderates the relationship between consumption value switching intention. However, conditional value does not significantly affect the role of functional and recognition value in influencing switching intention. Furthermore, integrating TAM, TCV, and VAM into a research framework suggests a better prediction of consumers’ switching intentions. The perceived value of VAM framework is not examined from a cost-benefit perspective. This study’s distinctive approach of scrutinizing every dimension of consumption value - functional, emotional, recognition, and epistemic value - gives a more exact prediction of which product value affects intention.

5.2. Managerial implication and limitation

These findings have implications for electric appliance manufacturers who want to enhance their new product acceptance through segmented product feature development, communication, and promotion campaigns. Analyzing unobserved heterogeneity could aid related industries in creating segmented products, features, and communication strategies, irrespective of demographic segmentation. In the future, a longitudinal survey should be conducted to compare urban and rural areas, subsidized and non-subsidized customers, and small business/commercial and household customers. In the interview, some respondents explained that their opinions about the importance of time-oriented convenience were influenced by the presence of a housemaid at home. This is because they are not primarily in charge of cooking activities. Independent variables such as peer opinion (Buabeng-andoh & Baah, 2020; Sohn & Kwon, 2020) and the level of involvement in meal preparation may influence consumption values. To investigate the moderating effect of conditional value, future researchers may consider customers’ opinions outside the Java-Bali grid, which includes regions with varying electricity availability (Java-Bali grid area vs. islands) and availability of gas supplies (LPG and Liquified Natural Gas).

Although there is no difference in response between the Jakarta area and other provincial capitals in Java for those aged 45 and above, this study has some limitations in generalizing the findings. This study only considered a limited number of benefits. First, because the induction stove is new to some respondents in Indonesia, it is better to have a trial session with an induction stove before filling out the questionnaire to ensure the same understanding of its use. Second, research on the small-scale enterprise users and food stall segments will provide valuable information, either for electric cookware manufacturers to develop new products or for PLN to provide electricity to support people’s mobility.

6. Conclusions

This research examines how the usefulness, ease of use, and time-oriented convenience of an induction stove as a replacement for an LPG-fueled stove affects its consumption value (functional, recognition, emotional, and epistemic values). It also investigates the moderating effect of conditional value on the intention to switch household cooking energy by integrating TAM and TCV into VAM. The empirical data targeted LPG users who utilized electricity at 1300 VA or above on the Java-Bali grid.

Based on the results, 12 out of 19 hypotheses were supported. The data show that usefulness has a substantial impact on functional, recognition, and emotional value. These findings confirm prior studies showing usefulness is a determinant that promotes overall product perceived value (Lin et al., 2012), (Hsu & Lin, 2018), (Vishwakarma et al., 2020), (Sohn & Kwon, 2020), (Talawanich & Au, 2020), (Lian, 2020). The empirical evidence also supports (Vishwakarma et al., 2020) that ease of use has less impact than usefulness. Moreover, ease of use merely affects epistemic value in influencing switching intention. Time-oriented convenience is also an important factor in increasing functional, emotional, epistemic, and recognition value. Time-oriented convenience is also an important factor in increasing functional, emotional, epistemic, and recognition value. This finding is like earlier research on the positive impact of service convenience on perceived value (Wong, 2021), (Pham et al., 2018). In addition, assessing quadratic correlations indicates the significant impact of usefulness on epistemic value and recognition value on switching intentions.

The result indicates that conditional value has a moderating effect on weakening the impact of emotional value on the intention to switch while the moderating effect on recognition and emotional value, as Sweeney and Soutar (2001) argue, is not significant. As a result, Tanrikulu’s (2021) proposition was only partially proven in this study. In addition, gender heterogeneity in response is shown, especially the effect of ease of use and time-oriented convenience on epistemic value. Integrating TCV into VAM by applying a multidimensional approach to perceived value indicators proves that functional, emotional, and epistemic values significantly affect switching intention, which is consistent with Ali et al. (2019), Yang and Lin (2017), Yoo et al. (2013) and Zailani et al. (2019). Moreover, the recognition value does not support the VAM framework (Yoo et al., 2013).

Author contributions

Conceptualization, H.H-M., A.S., F.R. and A.Y.; methodology, H.H-M., A.S., F.R. and A.Y.; software, H.H-M.; validation, A.S., F.R. and A.Y.; formal analysis, H.H-M.; investigation, A.S., F.R. and A.Y.; resources, H.H-M.; data curation, H.H-M.; writing—original draft preparation, H.H-M.; writing—review and editing, H.H-M., D.F.H., H.N.; visualization, H.H-M.; supervision, H.H-M., D.F.H., H.N.; project administration, H.H-M.; funding acquisition, H.N.

Citation information

Cite this article as

Acknowledgements

PT PLN (Persero)/PT PLN Indonesia Power, is a state-owned electricity company in Indonesia.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data available on request

Additional information

Notes on contributors

Hilda Hilmiyati-Mas’adah

Hilda Hilmiyati-Mas’adah has been working in a power generation company since 2000 and is a Ph.D. student at ¹. She holds a master’s degree in marketing management from the PPM School of Management in Jakarta.

Achmad Sudiro

Achmad Sudiro is a professor at ¹. He has extensive research collaborations in Asia region, a chief editor of Nationally Accredited Journals, and served as Chair of the on-campus Ph.D. at¹

Fatchur Rohman

Fatchur Rohman is a professor at ¹ with research interests in marketing, operations, and strategic management.

Agung Yuniarinto

Agung Yuniarinto is a senior lecturer at ¹ and hold a Ph.D. in management.

Dzikri Firmansyah Hakam

Dzikri Firmansyah Hakam is a lecturer at School of Business Management ITB, Bandung, and hold a Ph.D. in energy economics from University of Dundee. He is a professional has some high-quality publications and a regular reviewer of international journals and conferences.

Herry Nugraha

Herry Nugraha holds a Ph.D. in electrical engineering ITB and interested in electricity system and risk analysis research.

Appendices

Appendix A. The construct’s indicators

Usefulness

x1.1:Using an induction stove would enable me to accomplish my cooking more quickly

x1.2:Using an induction stove would improve my cooking performance

x1.3:Using an induction stove would increase my cooking productivity

x1.4:Using an induction stove would enhance my cooking effectiveness

x1.5:Using an induction stove would make it easier to do my cooking activity

x1.6:I would find an induction stove useful in my cooking activity

Ease of use

x2.1:Learning to operate an induction stove would be easy for me

x2.2:I would find it easy to get an induction stove to do what I want it to do

x2.3:My interaction with the induction stove would be clear and understandable

x2.4:I would find an induction stove to be flexible to interact with

Time-oriented convenience

x3.1:The induction stove is fast to use

x3.2:An induction stove is a means of saving time

x3.3:I’m busy and regard time as scarce/valuable

x3.4:The amount of time required to use an induction stove is specified

x3.5: I can use an induction stove while doing other activities

x3.6:Induction stove gives me more time to relax

Functional value

z1.1:The foods prepared by an induction stove have consistent quality

z1.2:The induction stove is durable

z1.3:The use of an induction stove would save cooking cost

z1.4:The induction stove is reasonably priced

z1.5:The induction stove is a good kitchen appliance for the price.

Recognition value

z2.1:Using an induction stove would improve the way that I am perceived

z2.2:Using an induction stove would make a good impression on other people

z2.3:Using an induction stove would help me to feel acceptable

Emotional value

z3.1:Using an induction stove instead of an LPG-fueled stove would feel like the morally right thing to do

z3.2:Using an induction stove instead of an LPG-fueled stove would make me feel like a better person

z3.3:Using an induction stove instead of an LPG-fueled stove would feel like making a good personal contribution to something better

Conditional value

z4.1:I would use an induction stove instead of an LPG-fueled stove if there were a subsidy for electricity bills

z4.2:I would use an induction stove instead of an LPG-fueled stove when electric power is available all the time

z4.3:I would use an induction stove instead of an LPG-fueled stove under LPG shortage conditions

Epistemic value

z5.1:I like to search for the new different product information

z5.2:I am willing to seek out novel information about the induction stove

z5.3:I would prefer to gain substantial information about the makes and models of electric kitchen appliances before purchase

Intention to switch household cooking energy

y1: I prefer to think of electricity as my primary source of cooking energy.

y2: I’m thinking about boosting my electricity consumption as my cooking energy.

y3: I’m thinking about reducing my LPG consumption as my cooking energy source.

y4: I am determined to switch to use an induction stove

y5: The chance of my switching to an induction stove is high

Appendix B

Descriptive statistics

Table B1. Demographic and characteristics of respondents.

Variable		Count
Gender	female	146 [50.34%]
	male	144 [49.66%]
Age	15–24	38 [13.1%]
	25–34	48 [16.55%]
	35–44	74 [25.52%]
	45–54	83 [28.62%]
	55–64	39 [13.45%]
	Above 64	8 [2.76%]
Household category	1300 Volt Ampere (VA)	115 [39.66%]
	2200 Volt Ampere (VA)	112 [38.62%]
	3500 Volt Ampere (VA)	42 [14.48%]
	4400 Volt Ampere (VA)	9 [3.1%]
	5500 Volt Ampere (VA) and above	12 [4.14%]
LPG consumption	subsidized LPG (3kg cylinder)	100 [34.48%]
	non subsidized LPG	134 [46.21%]
	both	56 [19.31%]
City	Jakarta area	140 [48.28%]
	Outside Jakarta (other big cities: Bandung, Yogyakarta, Surabaya)	150 [51.72%]
Exposure level	partially use an induction stove for cooking	55 [18.97%]
	never use an induction stove for cooking	235 [81.03%]
Employment	Full-time employee in the office/ factory	129 [44.48%]
	students, college students	31 [10.69%]
	Full-time employee work from home	43 [14.83%]
	entrepreneur	27 [9.31%]
	not working on purpose (i.e., taking care of the house)	22 [7.59%]
	retired	18 [6.21%]
	others	20 [6.9%]
Education level	high school	51 [17.59%]
	bachelor	40 [13.79%]
	graduate	134 [46.21%]
	postgraduate	65 [22.41%]
Household/life situation	single and live alone	10 [3.45%]
	single and living with parents/ friends	50 [17.24%]
	partner without children	25 [8.62%]
	couple with children	198 [68.28%]
	others (prefer not to answer)	7 [2.41%]

Table B2. Means and standard deviation of constructs’ indicators.

Item	Ref	Mean	SD	Kurtosis	Skewness
y.1	(Wu et al., 2017)	3.417	1.064	−0.697	−0.101
y.2		3.310	1.067	−0.614	−0.046
y.3		3.317	1.042	−0.539	0.052
y.4		3.207	1.056	−0.552	0.072
y.5	(Chang & Chen, 2014)	3.334	1.068	−0.642	−0.033
x1.1	(Davis, 1989)	3.379	1.051	−0.530	−0.090
x1.2		3.245	1.101	−0.470	−0.201
x1.3		3.348	1.040	−0.362	−0.201
x1.4		3.441	1.050	−0.432	−0.258
x1.5		3.700	1.032	−0.193	−0.546
x1.6		3.845	0.976	0.238	−0.715
x2.1	(Davis, 1989)	4.021	0.958	0.406	−0.871
x2.2		4.007	0.928	0.496	−0.819
x2.3		4.034	0.894	0.810	−0.854
x2.4		3.938	0.904	−0.251	−0.525
x3.1	(Gross & Sheth, 1989; Hilmiyati-Mas’adah et al., 2023)	3.976	0.896	−0.699	−0.386
x3.2		3.734	0.930	−0.797	−0.147
x3.3		3.676	0.953	−0.620	−0.152
x3.4		3.762	0.911	−0.579	−0.199
x3.5		3.872	0.918	−0.432	−0.417
x3.6		3.717	0.930	−0.656	−0.158
z.11	(Zailani et al., 2019)	3.517	0.997	−0.274	−0.195
z.12		3.586	0.910	−0.579	0.004
z.13		3.276	1.079	−0.550	−0.121
z.14		3.317	1.035	−0.514	−0.066
z.15		3.690	0.961	−0.211	−0.350
z.21	(Zailani et al., 2019)	3.321	1.161	−0.585	−0.290
z.22		3.376	1.151	−0.582	−0.349
z.23		2.990	1.152	−0.592	0.020
z.31	(Mutum et al., 2021)	3.097	1.141	−0.540	−0.162
z.32		3.097	1.094	−0.497	−0.066
z.33	(Zailani et al., 2019)	3.334	1.041	−0.400	−0.207
z.41	(Zailani et al., 2019)	3.717	1.128	−0.577	−0.499
z.42		3.841	1.011	−0.591	−0.501
z.43		3.986	0.958	−0.063	−0.730
z.51	(Mutum et al., 2021)	3.886	0.938	−0.058	−0.577
z.52		3.831	0.959	−0.251	−0.505
z.53	(Zailani et al., 2019)	4.055	0.904	0.404	−0.784

Appendix C

Inferential statistics

Figure 1. Subsidized LPG for poor households.

Source: (Directorate General of Oil & Gas, 2023), (Central Agency on Statistics, 2023)

Figure 2. LPG import-demand ratio.

Source: (Indonesia Ministry of Energy & Mineral Resources, 2020)

Figure 3. Empirical research model.

Table C1. Cross loading of Indicators.

ITEMS	SI	USE	EOU	TOC	FUNC	RECV	EMOV	EPIV	CODV
y.1	0.838	0.516	0.393	0.552	0.595	0.467	0.530	0.435	0.359
y.2	0.910	0.540	0.390	0.541	0.624	0.535	0.631	0.471	0.401
y.3	0.868	0.529	0.437	0.503	0.549	0.567	0.548	0.488	0.385
y.4	0.924	0.586	0.362	0.563	0.622	0.585	0.658	0.446	0.387
y.5	0.920	0.594	0.363	0.515	0.604	0.526	0.643	0.441	0.456
x1.1	0.460	0.779	0.325	0.546	0.553	0.377	0.467	0.234	0.237
x1.2	0.533	0.865	0.350	0.619	0.649	0.502	0.564	0.324	0.309
x1.3	0.581	0.913	0.427	0.641	0.669	0.482	0.573	0.352	0.342
x1.4	0.557	0.902	0.369	0.680	0.645	0.460	0.578	0.359	0.340
x1.5	0.558	0.855	0.502	0.757	0.659	0.465	0.569	0.393	0.368
x1.6	0.440	0.745	0.465	0.626	0.534	0.433	0.499	0.371	0.379
x2.1	0.359	0.367	0.911	0.489	0.411	0.303	0.271	0.517	0.412
x2.2	0.394	0.433	0.942	0.537	0.458	0.337	0.323	0.498	0.408
x2.3	0.437	0.460	0.949	0.600	0.505	0.385	0.372	0.573	0.442
x2.4	0.409	0.509	0.892	0.615	0.507	0.353	0.375	0.511	0.436
x3.1	0.482	0.568	0.686	0.800	0.609	0.406	0.490	0.477	0.467
x3.2	0.533	0.710	0.551	0.898	0.680	0.452	0.546	0.452	0.339
x3.3	0.492	0.696	0.458	0.887	0.640	0.441	0.523	0.386	0.312
x3.4	0.546	0.740	0.492	0.893	0.694	0.472	0.569	0.475	0.398
x3.5	0.533	0.643	0.524	0.838	0.620	0.437	0.523	0.447	0.371
x3.6	0.534	0.630	0.481	0.896	0.657	0.502	0.559	0.433	0.357
z.11	0.554	0.740	0.469	0.734	0.816	0.608	0.622	0.515	0.471
z.12	0.560	0.626	0.462	0.671	0.873	0.495	0.598	0.462	0.516
z.13	0.600	0.596	0.351	0.598	0.827	0.496	0.614	0.344	0.410
z.14	0.572	0.538	0.337	0.498	0.813	0.441	0.555	0.339	0.421
z.15	0.492	0.515	0.499	0.574	0.815	0.418	0.495	0.500	0.511
z.21	0.530	0.496	0.401	0.478	0.547	0.938	0.668	0.550	0.451
z.22	0.542	0.506	0.431	0.466	0.526	0.942	0.669	0.554	0.441
z.23	0.574	0.472	0.194	0.480	0.567	0.858	0.722	0.447	0.380
z.31	0.611	0.600	0.314	0.541	0.646	0.657	0.935	0.387	0.432
z.32	0.643	0.614	0.335	0.574	0.633	0.774	0.943	0.434	0.437
z.33	0.651	0.599	0.380	0.620	0.692	0.687	0.940	0.459	0.548
z.41	0.285	0.257	0.281	0.299	0.406	0.337	0.418	0.409	0.834
z.42	0.489	0.398	0.450	0.417	0.546	0.482	0.522	0.530	0.917
z.43	0.330	0.330	0.435	0.377	0.480	0.352	0.336	0.486	0.839
z.51	0.462	0.335	0.458	0.441	0.440	0.526	0.375	0.921	0.466
z.52	0.535	0.447	0.549	0.504	0.544	0.580	0.517	0.901	0.534
z.53	0.376	0.299	0.536	0.439	0.425	0.420	0.324	0.896	0.508

INT: Intention to switch cooking energy; USE: Usefulness; EOU: Ease of use; TOC: Time-oriented Convenience; FUNC: Functional Value; RECV: Recognition Value; EMOV: Emotional Value; EPIV: Epistemic value; CODV: Conditional Value

Table C2. Validity and reliability.

Construct	Cronbach’s Alpha	ρA	Composite Reliability	Average Variance Extracted (AVE)
Intention to switch cooking energy	0.936	0.939	0.951	0.797
Usefulness	0.919	0.924	0.937	0.714
Ease of use	0.943	0.947	0.959	0.854
Time-oriented Convenience	0.935	0.937	0.949	0.756
Functional Value	0.886	0.890	0.916	0.687
Recognition Value	0.900	0.900	0.938	0.834
Emotional Value	0.934	0.934	0.958	0.883
Epistemic value	0.891	0.899	0.932	0.821
Conditional Value	0.835	0.912	0.898	0.747

Table C3. Discriminant validity: Heterotrait-Monotrait.

Construct	USE	EOU	TOC	FUNC	RECV	EMOV	CODV	EPIV
USE
EOU	0.515
TOC	0.824	0.648
FUNC	0.805	0.556	0.814
RECV	0.591	0.405	0.567	0.664
EMOV	0.693	0.386	0.658	0.764	0.820
CODV	0.431	0.504	0.477	0.640	0.519	0.555
EPIV	0.438	0.617	0.558	0.580	0.627	0.489	0.633
SI	0.666	0.462	0.640	0.735	0.655	0.722	0.480	0.554

Table C4. Inner variance inflation factor (VIF) of structural model constructs.

Constructs	VIF
Usefulness	2.418
Ease of use	1.593
Time-oriented Convenience	2.955
Functional Value	2.317
Recognition Value	2.702
Emotional Value	3.041
Conditional Value	1.735
Epistemic value	1.824

Table   C5. Quadratic effects on unsupported linear correlation paths.

Path	Path coefficient (β)	t statistics	Quadratic effect result
H1d: USE → EPIV	0.139	3.005	Significant (R^2=0.390)
H2a: EOU → FUNC	−0.056	0.881	Not significant
H2b: EOU → RECV	0.073	0.876	Not significant
H2c: EOU → EMOV	−0.078	1.296	Not significant
H5: RECV → INT	0.108	2.110	Significant (R^2=0.579)

Notes: t-values test: 1.96 at sig level 5%

Table C6. Predictive power of structural model

Indicators	Q²_predict	RMSE PLS	RMSE LM	RMSE PLS-LM
y.1	0.298	0.895	0.929	−0.034
y.2	0.310	0.889	0.920	−0.031
y.3	0.288	0.882	0.899	−0.017
y.4	0.327	0.870	0.893	−0.023
y.5	0.313	0.888	0.869	0.019

Table 1. Structural model estimates.

Hypotheses	Path coefficient	t-statistics	Result
H1a: USE → FUNC	0.387	4.964	Accepted
H1b: USE → RECV	0.336	4.026	Accepted
H1c: USE → EMOV	0.415	5.332	Accepted
H1d: USE → EPIV	0.013	0.180	Rejected
H2a: EOU → FUNC	0.079	1.475	Rejected
H2b: EOU → RECV	0.084	1.148	Rejected
H2c: EOU → EMOV	−0.026	0.454	Rejected
H2d: EOU → EPIV	0.409	5.807	Accepted
H3a: TOC → FUNC	0.404	5.155	Accepted
H3b: TOC → RECV	0.212	2.189	Accepted
H3c: TOC → EMOV	0.314	3.834	Accepted
H3d: TOC → EPIV	0.253	2.556	Accepted
H4: FUNC → INT	0.315	4.228	Accepted
H5: RECV → INT	0.081	1.095	Rejected
H6: EMOV → INT	0.327	4.253	Accepted
H7: EPIV → INT	0.151	3.007	Accepted
H8a: CODV x FUNC → INT	0.106	1.684	Rejected
H8b: CODV x RECV → INT	0.019	0.323	Rejected
H8c: CODV x EMOV→ INT	−0.160	2.180	Accepted

Notes: t-values test: 1.96 at sig level 5%

Table 2. Coefficient of determination of endogenous variables: impact of quadratic effect.

Endogenous variables	Linear			Quadratic
	R^2 excl. moderating effect	R^2 incl. moderating effect	Effect size (f^2)	R^2	Effect size (f^2)
Emotional value	0.451	0.451
Epistemic value	0.368	0.368		0.390	3.70%
Functional value	0.627	0.627
Recognition value	0.323	0.323
Intention to switch cooking energy	0.558	0.568	2.31%	0.578	4.66%

f² 0.020-0.150:low; 0.150-0.350: medium; > =0.350: high (Benitez et al., 2020).

for social science research, R²=0.75 high; R²=0.5 medium; R²=0.25 low (Hair et al., 2021)

Table 3. Multi group analysis.

Correlations	Female vs. Male		Jakarta vs. other cities		Under 45 vs. 45 above
	β-diff	p-value	β-diff	p-value	β-diff	p-value
USE → FUNC	−0.128	0.412	0.101	0.474	−0.005	0.967
USE → RECV	0.204	0.238	0.155	0.352	0.173	0.297
USE → EMOV	−0.209	0.188	−0.034	0.817	−0.096	0.532
USE → EPIV	0.131	0.365	0.139	0.351	−0.197	0.182
EOU → FUNC	0.092	0.385	0.134	0.176	−0.007	0.971
EOU → RECV	0.059	0.699	−0.113	0.423	−0.136	0.355
EOU → EMOV	−0.050	0.694	0.015	0.901	0.000	0.978
EOU → EPIV	0.330	0.009	0.137	0.306	0.023	0.855
TOC → FUNC	0.042	0.792	−0.194	0.167	−0.031	0.830
TOC → RECV	−0.295	0.145	−0.114	0.551	−0.126	0.498
TOC → EMOV	0.192	0.285	0.091	0.569	0.100	0.551
TOC → EPIV	−0.400	0.036	−0.319	0.095	0.025	0.917
FUNC → INT	0.112	0.460	0.132	0.368	−0.081	0.605
RECV → INT	−0.317	0.057	−0.240	0.167	−0.078	0.642
EMOV → INT	0.083	0.592	0.132	0.398	0.319	0.055
EPIV → INT	−0.067	0.554	−0.030	0.786	−0.105	0.350
CODV x FUNC → INT	−0.126	0.325	−0.055	0.661	−0.039	0.791
CODV x RECV → INT	−0.160	0.257	−0.052	0.689	−0.229	0.122
CODV x EMOV→ INT	0.116	0.466	0.086	0.581	0.148	0.374

Notes: β-diff = path coefficient difference between two groups; t-values test: 1.96 at sig level 5%

Other provincial cities: Bandung (West Java), Yogyakarta (in the Central Java area), Surabaya (East Java)

Table 4. Multi-group analysis by gender.

Correlations	Original path	β –diff by gender	Female		Male
			Path coef.	p-value	Path coef.	p-value
USE → FUNC	0.387	NS	0.304	0.014	0.432	0.000
USE → RECV	0.336	NS	0.455	0.001	0.250	0.023
USE → EMOV	0.415	NS	0.301	0.011	0.510	0.000
USE → EPIV	0.013	NS	0.076	0.529	−0.056	0.512
EOU → FUNC	0.079	NS	0.123	0.153	0.031	0.634
EOU → RECV	0.084	NS	0.092	0.379	0.033	0.756
EOU → EMOV	−0.026	NS	−0.048	0.520	0.002	0.984
EOU → EPIV	0.409	S	0.536	0.000	0.206	0.032
TOC → FUNC	0.404	NS	0.451	0.000	0.409	0.000
TOC → RECV	0.212	NS	0.061	0.673	0.356	0.013
TOC → EMOV	0.314	NS	0.410	0.001	0.218	0.089
TOC → EPIV	0.253	S	0.092	0.500	0.491	0.000
FUNC → INT	0.315	NS	0.364	0.001	0.252	0.020
RECV → INT	0.081	NS	−0.021	0.807	0.297	0.031
EMOV → INT	0.327	NS	0.364	0.001	0.281	0.017
EPIV → INT	0.151	NS	0.133	0.124	0.200	0.006
CODV x FUNC → INT	0.106	NS	0.031	0.718	0.157	0.092
CODV x RECV → INT	0.019	NS	−0.032	0.708	0.128	0.251
CODV x EMOV→ INT	−0.160	NS	−0.133	0.213	−0.248	0.038
R^2 Intention to switch	0.568		0.517		0.656

Notes: original path coefficient applied on all data; t-values test: 1.96 at a significance level of 5%