Users’ Perceived Value of Electric Vehicles in China: A Latent Class Model-Based Analysis

Abstract

Future promotional strategies for electric vehicles (EVs) need to be tailored to the initial users’ perceptions regarding these vehicles. This study aims to evaluate EV users’ perceived value in terms of the following five key dimensions: economic, environmental, social, emotional, and technological value. Recognizing the diversity of users’ perceived value, a latent class model is employed to categorize respondents, integrating predictive and outcome variables for a comprehensive analysis. The results indicate that 62% of users fall into a high endorsement group, indicating the widespread acceptance of the multidimensional value brought by EVs. Another 21% fall into a moderate endorsement group, signifying the partial approval of select EV values (e.g., emotional and social). Conversely, 17% are categorized as low endorsement users, expressing a low level of acceptance in terms of all the dimensions of EV value. Demographic characteristics such as family size and income significantly influence these user classifications, and there are marked differences in perceptions of certain vehicle attributes.

1. Introduction

1.1. Research Background

Air pollution poses a severe challenge in China, particularly in its many large cities [1]. Transportation-related emissions are a major contributor to environmental degradation. Motor vehicles emit pollutants such as carbon monoxide, hydrocarbons, and nitrogen oxides, making them a primary source of urban air pollution. Additionally, transportation contributes 25% of global CO₂ emissions and is projected to rise to 50% by 2035 [2]. In response to these issues, the Chinese government has expedited the development of green and low-carbon transportation systems to achieve national carbon neutrality goals and ensure energy security. With the continuous advancement of electric vehicle (EV) technology, the large-scale promotion of EVs is viewed as a key strategy to reduce air pollution and foster green development [3].

The Chinese government has implemented several policies to encourage EV adoption, including purchase subsidies, exemptions from acquisition tax, enhanced charging infrastructure, and the provision of free parking with time restrictions, significantly accelerating the market expansion of EVs [4]. Official statistics indicate that EV sales in China reached 6.887 million units in 2022, marking a 93.4% year-on-year increase, indicative of robust market growth. By the end of 2022, the number of EVs in China had reached 13.10 million, accounting for 4.1% of the total number of vehicles [5]. EV sales are projected to increase six to eight times by 2025, further expanding the scale of China’s EV market [6].

Despite this growth, EVs still hold a relatively small market share compared to traditional-fuel vehicles. The promotion of EVs is challenging due to their limited range, costly batteries, and charging limitations [7]. Therefore, lots of early studies focused on the adoption intention and behavior of EV users. As the market share of EVs continues to grow, some scholars are gradually shifting their focus to the perception feedback of EV users in recent years. The main reason for this shift is that to continue growing the market share of EVs, promotional strategies of EVs need to be tailored to the initial users’ perceptions of these vehicles [8]. In addition, not only do we need more new EV users, but also the initial users will need to continue choosing EVs for subsequent purchases [9].

1.2. Literature Review

In order to accurately reflect the perception feedback of EVs, studies currently are targeting consumers with EV-related experience. This is important due to the following considerations: First, consumers without prior experience of using EVs tend to express their purchase intentions and have difficulty accurately describing their feelings about using EVs. Second, understanding the demographic characteristics of existing EV owners (age distribution, gender, income levels, and household composition) offers valuable references for both EV manufacturers and policymakers [10], thereby enhancing EV market penetration. Third, consumer attitudes and behaviors evolve over time [11], and focusing on current EV users facilitates our understanding of the transition process from potential to actual adopters in the EV market, along with the key influencing factors. Fourth, existing users have completed the entire process from purchase to usage, and their behaviors and feedback can authentically reflect product performance in real-world conditions [12]. For instance, users may have heightened expectations and requirements regarding environmental performance, battery lifespan, charging efficiency, driving range, charging convenience, and pricing. By analyzing current users’ actual behaviors, perceived value, and satisfaction levels, we can better comprehend the EV adoption process. These findings provide scientific evidence for market promotional strategies and product improvements, ultimately enhancing consumer acceptance of, recognition of, and actual purchasing behavior regarding EVs. However, studies on EV perception feedback are relatively limited compared to studies on EV adoption. One important reason for this is that surveying EV users is more challenging than surveying potential EV consumers [13].

The current studies on user perception can be divided into two types. The first type of study focuses on the overall perception of EVs and has verified that some factors affect overall EV perception. For example, Hu et al. (2023) found that financial, environmental, and psychological perceptions were positively correlated with perceived value [14]. However, the effects of physical safety and performance risk on perceived value were diverse. The second type of study focuses on surveying perceptions of specific vehicle attributes, such as driving range, charging time, and charging convenience. For example, Anastasiadou et al. (2022) found that EV users in South Korea consider battery range and battery charging as the two most important factors negatively influencing their overall perception [15]. Overall, users have a worse perception of battery range, charging convenience, and charging time [13]. Factors such as after-sales service and battery recycling policies also affect user perception levels [16]. Users with a better perception of EVs are more willing to purchase them again and will recommend EVs to others [17].

One limitation in existing EV feedback studies is they analyze EV users as a whole and do not consider heterogeneous groups [16,18]. There is another limitation in systematically examining the determinants of EV user segmentation and attribute preferences. Current studies adopt fragmented approaches, with some focusing exclusively on economic behaviors and psychological factors for EV user classification [19], while others employ distinct categorization frameworks based on driver-related factors, infrastructure-related considerations, and vehicle-specific characteristics [20]. This methodological dispersion hinders the development of a unified theoretical framework for EV market segmentation. Since EV users’ perceived value differs depending on factors such as gender, age, income, and vehicle size, it is crucial to conduct heterogeneous studies. Existing studies have identified substantial variations in perception levels among EV owners, with some indicating that they would not choose an EV for their next vehicle purchase and would revert to gasoline vehicles [9,18]. These differences in feedback represent a knowledge gap that may impact the development of targeted interventions and policies aimed at promoting EV adoption.

1.3. Contributions

To better elucidate and address perceived heterogeneity among existing EV users, methods such as cluster analysis can be used to categorize EV users. A drawback to cluster analysis, however, is the lack of robust criteria for determining the optimal number of categories; it is better suited to analyzing continuous variables. Latent class analysis, an extension of traditional cluster analysis, focuses more on categorical variables. A key advantage of this is the ability to readily determine the optimal number of categories by comparing models [21]. Additionally, latent class analysis can be used to examine the factors that influence category membership.

Based on the above, this paper puts forward the following questions: How can EV users be classified into different categories based on their feedback? What factors influence the classification of EV user categories? And what disparities exist in attribute preferences for EVs among different user categories? Acknowledging the heterogeneity across user feedback, this study employs a latent class model to categorize respondents. Building upon consumption value theory and incorporating EV-specific characteristics, this study extends five key value dimensions to classify surveyed EV users into three distinct groups: high, moderate, and low endorsement segments. This segmentation enables a systematic exploration of inter-group heterogeneity. By introducing covariates, this study aims to comprehensively understand which factors significantly affect users’ category membership and to discern discrepancies in EV perceived value among distinct groups, thereby informing tailored market strategies and facilitating the deeper market penetration of EVs. Furthermore, through systematic analysis of each segment’s preferences regarding EV attributes, personal characteristics, household profiles, and social factors, this study offers predictive indicators for assessing the likelihood of repeat EV purchases among initial buyers. The findings equip EV manufacturers with actionable segmentation knowledge to optimize customer retention strategies, while simultaneously informing policymakers about effective measures to not only attract new EV adopters but also sustain existing users.

The remainder of this paper is structured as follows: Section 2 details the theoretical framework underlying the proposed model design. Section 3 elaborates on the questionnaire design, methodology, and data collection process used in this study. Section 4 presents the empirical analysis results, and Section 5 offers a detailed discussion of said results. Section 6 provides a brief summary and our concluding remarks.

2. Theoretical Framework

Previous studies have found that customers purchase not a product itself but the value it provides [22]. To accelerate the continuous growth of the EV market, it is essential to analyze users’ feedback regarding the value of EVs. Such analyses could provide guidance for manufacturers, policymakers, and stakeholders to grasp the actual needs of consumers, thereby informing practical measures to enhance user satisfaction and ensure EV market stability. The Sheth–Newman–Gross consumption value theory was introduced in 1991, proposing five dimensions of value that products can offer to customers: functional value, social value, emotional value, cognitive value, and conditional value [23]. These dimensions explain why consumers purchase and prefer a particular product over others [24]. The Sheth–Newman–Gross consumption value theory has been widely used in analyzing consumer behavior. For example, Lin and Huang (2012) applied the Sheth–Newman–Gross consumer value to determine the factors influencing consumer choice behavior regarding green products [25]. Alganad et al. (2023) analyzed the roles of consumption values in the green automotive industry [22].

The existing literature investigating determinants of green consumption behavior predominantly adopts a multidimensional analytical framework encompassing the following: individual factors (attitudes, personal values [26], and demographic characteristics), socio-cultural influences (social norms, social effects [27], and cultural context), environmental awareness/knowledge, product attributes, and external contextual factors (policy incentives and market conditions). The Sheth–Newman–Gross consumption value theory primarily examines consumers’ perceived value regarding products or services and its impact on purchase decision-making processes. Established research has further validated this theoretical framework’s efficacy in both identifying critical attributes influencing consumer purchasing behavior and explaining choice mechanisms among EV adopters [25]. Leveraging the Sheth–Newman–Gross consumer value theory and integrating the unique characteristics of EVs, this paper establishes five value dimensions specific to EV consumption: economic value, environmental value, social value, emotional value, and technological value (see Figure 1).

Functional value denotes the practical utility of a product, indicating the extent to which its features align with consumers’ needs [25]. In the context of EVs, this extends to technological value; it encompasses new driving experiences, technological innovations, and their role in addressing energy-related concerns [28]. Cognitive value involves personal curiosity and the pursuit of knowledge [25], which, in this context, can be translated into consumers’ interest in the environmental benefits of EVs. Environmental value is attained through clean energy usage and carbon emission reduction [29], reflecting the desire of EV users to decrease their carbon footprint and promote sustainable development. Conditional value pertains to the temporary enhancement of functionality and value a product offers under certain conditions [22]. For consumers purchasing EVs, this primarily manifests as economic value, including policy incentives, fluctuations in energy prices, purchase subsidies, and resale prices [30]. Consumers make a comprehensive judgement regarding the economic value of EVs based on their individual circumstances as well as market conditions [31].

Understanding the decision-making processes of consumers in continuing to choose EVs as their mode of transportation necessitates integrating feedback across various value dimensions, as illustrated in Figure 1.

Economic value is a crucial consideration for consumers when making purchasing decisions. It encompasses factors such as acquisition costs, operational expenses, maintenance fees, resale value, and potential government subsidies. Consumers meticulously weigh the cost-effectiveness of their choices, aiming to maximize the return on their investment. The steep battery costs associated with EVs contribute to their relatively high purchase prices, creating a significant barrier to widespread consumer adoption [32]. Moon corroborated this [33], highlighting a significant inverse relationship between the purchase price and consumers’ willingness to buy a product. Conversely, lower operational costs relative to traditional-fuel vehicles make EVs more attractive [31]. EVs offer substantial advantages in terms of energy expenditures, with electricity proving more cost-effective than gasoline or diesel in addition to its increased efficiency. Despite initial cost disparities, EVs generally yield lower total costs over their lifetime of ownership [34].

Resale value also weighs heavily in consumers’ purchase decisions, with a higher resale price increasing the likelihood of future similar purchases by mitigating economic losses. Moreover, resale value serves as a barometer of the durability and reliability of EVs [35]. Vehicle purchase subsidies, as an important policy instrument, mitigate the purchase costs of EVs for consumers, stimulating purchase intentions and market demand. Studies have indicated a significant positive correlation between subsidy policies and EV sales [36]. In summary, economic value is a key driver in the promotion of EVs.

Environmental value refers to consumers’ sense of environmental concern, signifying a readiness to assume greater responsibility for environmental stewardship [29]. EVs play a pivotal role in curbing air pollution and greenhouse gas emissions, reflecting EV users’ commitment to environmental conservation and sustainable development to some extent. Research underscores a close relationship between environmental awareness and individual interest in environmental issues—heightened environmental consciousness prompts proactive measures to combat climate change. Moreover, higher levels of environmental awareness can incentivize consumers to engage in green purchasing and recycling behaviors [28]. In essence, environmental value is an important catalyst in promoting EVs, reflecting consumers’ dedication to environmental quality and encouraging governmental and corporate initiatives toward sustainable development goals.

Social value refers to the extent to which others’ opinions on a product or service affect consumers’ decisions [37]. Social value assumes a critical role in the promotion of EVs, reflecting individuals’ perceptions of how their consumption choices are perceived by others—a factor that significantly influences purchase decisions [38]. Studies also suggest that social experiential value positively impacts consumers through interactions between sales personnel and consumers, thereby potentially enhancing the emotional and functional experiential value of EVs [39]. The views of family, friends, and colleagues significantly affect consumers’ adoption of EVs as well. In summary, social value increases the attractiveness of EVs through social interaction and a sense of identification, fostering not only consumer interest in EVs but also broader societal acceptance of sustainable transportation modes [40].

Emotional value encompasses the affective experiences a product or service provides to fulfill consumers’ emotional needs. Some consumers opt for EVs as they perceive them as a means to attain emotional gratification by embracing green consumption as a superior lifestyle choice [41]. The emotional value derived from EVs stems not only from pleasurable driving experiences but also from the consumer’s belief that they are engaging in environmentally friendly actions, thus enhancing their self-image [42]. Studies have shown that emotional value positively affects EV adoption, especially when such adoption is seen as a manifestation of social responsibility. Leveraging emotional value in EV marketing entails emphasizing EVs as, beyond transportation modes, vehicles for expressing personal eco-friendly philosophies and lifestyles. Reinforcing this emotional connection can heighten consumer favorability and adoption intentions towards EVs, which are crucial for market penetration [43].

Technological value is closely linked to consumers’ curiosity and preference for diversity and novelty, reflecting the demand for performance, functionality, innovation, and broader change [28]. The rapid charging technology and autonomous driving capabilities of EVs, for example, have significantly shifted traditional consumer perceptions, deepening the curiosity and interest in EVs [44]. Technological advancements in EVs address practical performance issues and provide value by fulfilling individual desires to explore new technologies. Highlighting technological value in market strategies can expedite the market acceptance and diffusion of EVs.

A comprehensive value framework for analyzing the EVs’ value was developed based on these principles, as illustrated in Figure 2. This framework includes not only traditional dimensions of consumer value (economic, emotional, and social) but also integrates unique factors associated with EVs, embodied in environmental and technological values. This diversified perspective allows the framework to capture the key elements driving EV users’ perception more accurately.

Building upon this foundation, a latent class analysis was employed to categorize respondents based on their perception feedback regarding EVs while also incorporating a range of covariates including personal attributes (e.g., age, gender, and educational background) and vehicle characteristics (e.g., fuel type, price, and battery range) for an in-depth analysis. The goal of this analysis is to explore effective strategies for enhancing user retention and brand loyalty, as well as to identify ways to improve customer experiences, facilitating the widespread and long-term adoption of EVs. The results offer strategic insights for manufacturers to better meet consumer needs and expectations, while also serving as a scientific basis for fostering sustained market growth for and the broader acceptance of EVs.

3. Data and Methodology

3.1. Methods

3.1.1. Instruments and Measures

This study examines the dimensions of economic, environmental, social, emotional, and technological value of EVs, drawing from established scales to craft a questionnaire aligned with previously stated research questions (see Tables S2 and S3) [22,28]. For instance, economic value in this case encompasses factors like the vehicle’s purchase price, travel costs, and resale value. Specific items are listed in

Table 1. Measurement items and EVs’ value.

Value Dimension	Item
Economic value	I purchased an EV due to its reasonable price compared to gasoline-powered vehicles.
	I purchased an EV due to its lower daily driving costs compared to gasoline-powered vehicles.
	I purchased an EV due to its high resale price compared to gasoline-powered vehicles.
Social value	It is important for me to buy an EV that is recommended by others.
	Owning the same EVs that other people use enhances my sense of belonging.
	My purchase of an EV has been fully recognized by the people around me.
Emotional value	Driving an EV assists in making a positive impression on others.
	Using an EV leads others to perceive me as someone with a strong sense of social responsibility.
	The positive societal feedback from using an EV meets my expectations (the social recognition and appreciation received from others after purchasing an EV).
Technological value	EVs meet my expectations for innovative solutions to energy problems.
	EVs meet my expectations for automotive technology innovation.
	EVs provide me with a novel driving experience that adds to the pleasure of driving.
Environmental value	EVs play a significant role in reducing greenhouse gas emissions.
	EVs can greatly help reduce the environmental pollution caused by personal travel. EVs can bring me a sense of fulfillment through its environmental benefits of energy conservation and emission reduction.

, and questionnaire items reliability and validity testing are shown at Table S1 (all items used a 3-point Likert scale, in which 1 = disagree, 2 = uncertain, and 3 = agree).

3.1.2. Latent Class Model

The latent class model (LCM) is a statistical technique for determining associations among manifest indicators through discrete latent variables, thus preserving their local independence. In this study, LCM was utilized to divide respondents into multiple homogeneous categories to capture the diversity among their preferences. The number of latent classes was initially unknown, so exploratory latent class analysis (LCA) was employed. This involved first assuming the existence of a single class, i.e., that observable variables were completely independent (also known as a “null” or “independence” model). Secondly, the number of latent classes was increased, and parameters were calculated for each resulting model. Thirdly, the models were compared based on fit indices to determine the optimal one before proceeding with classification and category assignments.

Mplus 8.3 was employed to conduct exploratory LCA on the questionnaire data. Models with 1–6 classes were constructed to determine the optimal one. Assessment measures included the Akaike information criterion (AIC), Bayesian information criterion (BIC), and the sample-size-adjusted Bayesian information criterion (aBIC), each employed to determine the model’s fit by comparison against expected and observed values, with lower values indicating better fit. Among these, BIC is commonly used as a primary indicator of model adequacy. Generally, the model with the smallest BIC is deemed optimal. Entropy, ranging from 0 to 1, is commonly used to assess the precision of model classification; values closer to 1 signify higher precision. Typically, an entropy value of 0.8 indicates classification accuracy exceeding 90%. A bootstrapped likelihood ratio test (BLRT) and Lo-Mendll-Rubin (LMR) test were further utilized to compare the fit between models with k − 1 and k classes. When a particular class contained a small proportion of respondents, separate analysis was unnecessary, so reducing the number of classes could be considered.

3.1.3. Regression Mixture Model

LCMs partition the sample into multiple latent groups, thereby revealing the group heterogeneity of the sample. These groups may exhibit distinct behavioral patterns and characteristics. A regression mixture model (RMM) is generally adopted to examine the differences between groups further. RMM enables more precise capture of both within-group and between-group differences, thereby enhancing the model’s explanatory power [45].

The RMM has been widely applied to explore the heterogeneity of latent group structures. For example, Shahrier and Habib (2025) utilized an LCM to segment respondents into three groups [46]. They then employed an RMM to investigate how the perception of EVs, plug-in hybrid electric vehicles (PHEVs), and internal combustion engine vehicles (ICEVs) varied across these groups, and the relationships between group membership and factors such as job nature and income. Similarly, Hajhashemi, Sauri Lavieri, and Nassir (2024) classified EV users based on LCM [47]. By combining RMM, they confirmed that distinct consumer groups exhibit unique charging demand. Yang et al. categorized EV owners into three groups based on their sensitivity to charging costs, time, and station distance by an LCM [48]. They simulated the impact of different scenarios on grid emissions by incorporating time tariffs and charging infrastructure completeness as covariates.

LCMs can be transformed into regression mixture models (RMMs) by incorporating covariates and linking group heterogeneity to economic, individual, and social attributes [45]. Predictive variables were incorporated in this study to analyze the factors influencing consumers’ category membership, as well as outcome variables to examine the differences in EV attribute preferences among different categories. Category variables were treated as moderating variables to explore the impact of personal attributes on EV users’ perceived value. Covariates included personal attributes, household characteristics, subsidy policy, and vehicle attributes.

In this study, personal attributes included gender (male/female), education (college or below/Bachelor’s/Master’s or above), age, marital status (married/unmarried), and place of residence (third-tier city/second-tier city/first-tier city). Household characteristics included the number of family members (2/3/4/5), disposable household income (below 100,000 CNY/100,000 CNY–150,000 CNY/150,000 CNY–200,000 CNY/200,000 CNY–250,000 CNY/250,000 CNY–300,000 CNY/300,000 CNY–400,000 CNY/400,000 CNY–500,000 CNY/500,000 CNY and above), whether there is a fixed parking spot at home (yes/no), whether there are children under 10 years old at home (yes/no), whether there are seniors over 60 years old at home (yes/no), the number of private cars in the family (1/2/3/4), and the number of years of EV ownership. Social attributes included whether the city has policies for limited-time free EV parking (yes/no), fuel vehicle purchase restrictions (yes/no), EV travel restrictions (yes/no), and EV purchase subsidies (yes/no). Vehicle attributes included fuel type (battery electric vehicles (BEVs)/PHEVs), vehicle size (mini/small/compact/mid-to-large), type (sedan/sport utility vehicle (SUV)), price, range, and charging time.

3.1.4. Model Specifications

Under the assumption of local independence, any two observed indicators within a category are uncorrelated. Their association is explained through the latent class variable, implying no direct correlation between them. Following the principle that the probability of concurrent independent events equals the product of their individual probabilities, the formula for the LCM is detailed in Equation (1). Here, p(c_i = k) denotes the proportion of the population belonging to a certain category k, also referred to as the latent class probability; p(y_i) represents the probability of individual i choosing a specific option.

$$p\left(y_i\right)=\sum _{k=1}^{K}p(c_i=k)p\left(y_i|c_i=k\right)$$

(1)

Once the optimal LCM is successfully fitted, each individual must be assigned to different latent class. The classification criterion used in LCA is based on Bayesian posterior probabilities, as shown in Equation (2).

$$p\left(c_i=k|y_i\right)={\displaystyle \frac{p\left(c_i=k\right)f\left(y_i|c_i=k\right)}{f\left(y_i\right)}}$$

(2)

When an LCM incorporates covariates, including predictive and outcome variables, it is essential to thoroughly consider potential classification errors or uncertainties introduced by incorporating covariates. Suppose W represents the category membership variable estimated based on the model, which does not perfectly align with the actual category membership variable C, leading to classification uncertainty (as in Equation (3)). Here, C stands for the category membership variable, N denotes the variable for dividing individuals into different latent class categories based on posterior distribution probabilities, and U represents the observed indicators. Nc1 refers to the number of individuals classified into category C1 according to N.

$$p\left(c_1,c_2\right)=P\left(C=c_2|N=c_1\right)={\displaystyle \frac{{\sum }_{N_i=c_1}P\left(C_i=c_2|U_i\right)}{N_{c_1}}}$$

(3)

RMMs typically consist of two parts: an LCA and a regression model. An RMM, with latent class variables serving as moderator variables, can be viewed as a moderating model. Latent class variables modulate the regression coefficients across different categories like grouping variables. The RMM expression is given in Equation (4), where C denotes the latent class variable, and both intercepts and slopes vary across classes.

$$y={\alpha }_c+{\beta }_{c}x+{\epsilon }_c$$

(4)

3.2. Data Collection

Quantitative research methods were employed in this study based on a structured questionnaire distributed online. Prior to the official survey, a pre-test was conducted with team members’ classmates, family, and friends as subjects. Based on their feedback and the overall pre-test results, minor adjustments were made to the questionnaire to remove irrelevant items, rephrase ineffective expressions, and improve the layout. These adjustments were made in line with principles of scientific rigor to ensure the quality and reliability of the questionnaire.

Once the questionnaire content was finalized, the formal survey was conducted in October 2023, targeting consumers who had purchased both BEVs and PHEVs. To ensure diversity and representativeness of the sample, respondents were sourced from first-tier cities (e.g., Nanjing), second-tier cities (e.g., Suzhou, Wuxi, and Changzhou), and third-tier cities (e.g., Yangzhou, Yancheng, and Zhenjiang) in Jiangsu Province. Respondents ranged in age from 20 to 50 years, encompassing a sufficiently wide variety of demographics. A purposive quota sampling method was employed to collect samples with diverse characteristics such as age, gender, and region.

4. Results

This section begins with a descriptive statistical analysis of the sample distribution (Section 4.1), along with tests for the reliability and validity of the survey data (Section 4.2). In Section 4.3, an LCM is applied to analyze the sample, dividing it into three latent groups based on perceived value. Additionally, Section 4.4 utilizes an RMM to examine the influence of demographic variables on EV users’ perceived value across different groups. Finally, Section 4.5 investigates differences in the EV perception attributes among these groups, using relevant vehicle attributes as outcome variables.

4.1. Sample Distribution

A total of 400 questionnaires were gathered after a collection period of six weeks. We filtered them based on the respondents’ answer speed, response patterns, and answers to the control questions, which excluded any questionnaires not meeting the requirements, leaving 337 valid questionnaires for the analysis. A sample size greater than 30 and less than 500 is suitable for questionnaire-based research, offering an adequate sample size with sufficient useful information [49].

Table 2. Descriptive statistics of sample.

Personal Attribute	Response	Number	Percentage
Educational background	College degree or below	73	0.22
	Bachelor’s degree	179	0.53
	Master’s degree or above	85	0.25
Gender	Male	233	0.69
	Female	104	0.31
Age	20–25	23	0.07
	26–30	87	0.26
	31–35	87	0.26
	36–40	82	0.24
	41–45	28	0.08
	46–50	30	0.09
Marriage	Married	260	0.77
	Single	77	0.23
Family size	1–2	11	0.03
	3	150	0.44
	4	154	0.46
	5	22	0.07
Family income	100,000 and below	17	0.05
	100,001–150,000	21	0.06
	150,001–200,000	46	0.13
	200,001–250,000	93	0.28
	250,001–300,000	86	0.26
	300,001–400,000	49	0.15
	400,001–500,000	15	0.04
	500,001 and above	10	0.03
Number of cars owned	1	198	0.59
	2	109	0.32
	3	30	0.09
Age of purchase	1	165	0.49
	2	88	0.26
	3	66	0.2
	4	13	0.04
	5	5	0.01
Fixed parking space	Yes	180	0.53
	No	157	0.47

provides a summary of the descriptive statistical results.

The characteristics of our sample align closely with recent surveys of EV consumers in China. The respondents were mainly aged between 20 and 50 years at the time of their participation in this study, reflecting the national age distribution of Chinese EV consumers. According to the China Consumers Association, 77% of EV consumers were aged 26 to 45 in 2024. The National Bureau of Statistics survey in Suzhou, Jiangsu Province, found that over 80% of surveyed NEV buyers were aged between 21 and 40 (National Bureau of Statistics, 2023). In terms of gender, our sample mirrors existing surveys, with 60% of the respondents being male and 40% being female. This is consistent with national data indicating that male and female consumers account for 61% and 39% of EV buyers, respectively, as well as a Shenzhen government survey indicating a 67% male and 33% female distribution among EV users (Shenzhen government, 2024). Our sample is also highly educated, with 78% of the respondents holding a Bachelor’s degree or above. This is comparable to an official survey in Changzhou, Jiangsu Province, where over 82% of EV users reported similar education levels (Changzhou government, 2024). In summary, the sociodemographic characteristics of our sample are consistent with the broader population of EV consumers in China, which tends to be young, male, and highly educated.

4.2. Reliability and Validity Test

All scale indices were subjected to reliability and validity tests, as reported in

Table 3. Factor analysis results.

Likert Scale Construct	Number of Items	Cronbach’s Alpha	Standardized Factor Loadings	Composite Reliability	AVE	Eigenvalue	Variance Explained (%)
Economic value	3	0.885	0.668	0.751	0.504	1.471	10.507
			0.789			1.324	9.454
			0.666			1.203	8.594
Social value	3	0.844	0.782	0.769	0.534	1.201	8.581
			0.553			1.169	8.352
			0.827			1.130	8.069
Emotional value	3	0.883	0.808	0.768	0.526	1.123	8.020
			0.636			1.071	7.650
			0.723			1.052	7.513
Technological value	3	0.809	0.818	0.814	0.597	0.935	6.675
			0.632			0.721	5.147
			0.85			0.658	4.700
Environmental value	2	0.873	0.763	0.76	0.514	0.508	3.631
			0.673			0.435	3.107

Note: The data demonstrated excellent suitability for factor analysis (KMO = 0.959; Bartlett’s test of sphericity; p < 0.001), with Varimax rotation applied to facilitate factor interpretation.

. The Cronbach’s alpha for economic value is 0.885, exceeding 0.6; the composite reliability and AVE are 0.751 and 0.504, respectively, with all standardized factor loadings exceeding 0.65. The Cronbach’s alpha for social value is 0.844, also above 0.6. Its composite reliability and AVE are 0.769 and 0.534, respectively, with all standardized factor loadings greater than 0.5. The Cronbach’s alpha for emotional value stands at 0.883, also above 0.6, with the composite reliability and AVE at 0.768 and 0.526, respectively, and the standardized factor loadings all exceeding 0.6. The Cronbach’s alpha of technological value is 0.809, with the composite reliability and AVE at 0.814 and 0.597, respectively, and all factor loadings above 0.6. Finally, the Cronbach’s alpha of environmental value is 0.873, with the composite reliability and AVE at 0.76 and 0.514, respectively, and all the factor loadings exceeding 0.6. In summary, all indices meet relevant statistical benchmarks, indicating robust reliability and validity.

4.3. Potential Classifications of EV Users

Given the diversity of consumer preferences, this study utilized an LCA to conduct an in-depth analysis of different value dimensions of EVs. According to

Table 4. Latent class analysis: goodness-of-fit and model selection.

Class	AIC	BIC	ABIC	Entropy	LMR(P)	BLRT(P)	CLASS PRO
1	9048.089	9155.051	9066.231	/	/	/	1
2	6555.406	6773.151	6592.339	0.992	0	0	0.77/0.23
3	5935.185	6263.712	5990.909	0.986	0	0	0.62/0.21/0.17
4	5828.693	6268.002	5903.207	0.996	0.0001	0	0.41/0.21/0.2/0.18
5	5778.483	6328.575	5871.787	0.992	0.062	0	0.4/0.21/0.07/0.14/0.18
6	5738.508	6399.383	5850.603	0.952	0.38	0	0.19/0.12/0.26/0.18/0.18/0.07

, the model’s BIC value was minimal when the total sample was divided into three categories, indicating this as the most appropriate classification. The three latent classes showed significant differentiation across all indicators, with an entropy value exceeding 0.8. Additionally, the p-values of the LMR and BLRT tests are below 0.05, further validating the rationality of the three-class division.

After confirming the three classes as the optimal classification, the conditional probabilities of different consumer values were calculated, and the three categories were assigned names accordingly. The first category (“high endorsement group”) comprised 62% of the respondents; its members exhibited significantly higher conditional probabilities across all preference dimensions, indicating their strong recognition of the multidimensional values of EVs. The second category (“moderate endorsement group”) accounted for 21%, with its members demonstrating higher conditional probabilities in certain value dimensions, suggesting they perceive EVs as satisfying their needs only for specific aspects. The third category (“low endorsement group”) contains 17% of the respondents. Its members generally showed low recognition across all value dimensions, suggesting that they do not believe EVs meet their needs. The specific demographic characteristics of the three categories are presented in

Table 5. Demographic characteristics of the three categories.

Demographic Variables	Category	Class 1 (n = 208)		Class 2 (n = 70)		Class 3 (n = 59)
		Number	Percentage	Number	Percentage	Number	Percentage
Gender	Female	62	29.81%	26	37.14%	16	27.12%
	Male	146	70.19%	44	62.86%	43	72.88%
Age	22–30	47	22.60%	32	45.71%	32	54.24%
	31–40	135	64.90%	25	35.71%	8	13.56%
	41–50	23	11.06%	9	12.86%	19	32.20%
	≥51	3	1.44%	4	5.71%	1	1.69%
Income (RMB)	≤100,000	5	2.40%	11	15.71%	1	1.69%
	100,000–150,000	5	2.40%	16	22.86%	0	0.00%
	150,000–200,000	22	10.58%	17	24.29%	7	11.86%
	200,000–250,000	56	26.92%	11	15.71%	26	44.07%
	250,000–300,000	60	28.85%	9	12.86%	17	28.81%
	300,000–400,000	39	18.75%	3	4.29%	7	11.86%
	400,000–500,000	13	6.25%	1	1.43%	1	1.69%
	≥500,000	8	3.85%	2	2.86%	0	0.00%
Education	Below bachelor’s degree	44	21.15%	19	27.14%	10	16.95%
	Bachelor’s degree	110	52.88%	33	47.14%	36	61.02%
	Master’s degree and above	54	25.96%	18	25.71%	13	22.03%
Car ownership	1	133	63.94%	37	52.86%	28	47.46%
	2	55	26.44%	26	37.14%	28	47.46%
	3	20	9.62%	7	10.00%	3	5.08%

.

4.4. RMM Analysis: Predictor Variable

Respondents were divided into three categories by the sizes of posterior probabilities estimated by the LCM. To further investigate the factors affecting EV users’ perceived value and to test the applicability of the proposed framework, an RMM was employed to further assess the effects of personal, economic, and social factors on EV users’ perceived value. The reference group for this model was the low endorsement group. The results are provided in

Table 6. Predictor regression mixture analysis.

	Class 1 (High Endorsement Group)			Class 2 (Moderate Endorsement Group Group)
	Coef	S.E	OR	Coef	S.E	OR
Homeownership	1.957 ***	0.571	7.077	0.98	0.599	2.665
Family size	−0.467 *	0.209	0.627	−0.582 *	0.286	0.559
Age	−0.026	0.024	0.974	−0.019	0.029	0.981
Age ≥ 60	0.189	0.314	1.208	1.064 **	0.409	2.898
Age ≤ 10	0.756 *	0.334	2.13	−0.041	0.407	0.96
Bachelor’s degree	0.333	0.497	1.395	−1.178 *	0.549	0.308
Master’s degree and above	−0.376	0.376	0.687	−1.19 **	0.458	0.301
Gender	−0.307	0.36	0.736	−0.7	0.205	0.496
Marital status	0.738 *	0.351	2.092	0.202	0.617	1.224
Income	0.181 *	0.092	1.199	0.52	0.176	0.594
Subsidy	2.55 *	1.121	12.81	2.849 *	1.15	17.26
Free time	2.01 ***	0.33	7.465	0.637	0.385	1.891
Purchase limit	0.374 *	0.38	1.688	1.78 **	0.52	1.167

Note: *, **, and *** indicate significance levels of 10%, 5%, and 1%; “Homeownership” refers to whether a consumer owns their place of residence; “family size” refers to the number of people in the household; “gender” is used considering females as the control group; and “marriage” is used considering being unmarried as the control group.

.

When examining the impact of individual factors on EV users’ perceived value, variables such as homeownership, marital status, and family size play significant roles. Homeowners are more likely to be in the high endorsement category than the low endorsement one, for instance, with a likelihood 7.077 times greater that of renters. Consumers with children under 10 years old are 2.13 times more likely to be in the high endorsement group compared to those without children. An increase in family size appears to negatively impact the EVs’ perceived value, with each additional family member reducing the probability of being in the high endorsement group to 0.627 times that of the low endorsement group. Marital status is also a significant factor, with married consumers being 2.092 times more likely to highly endorse EVs than unmarried ones, possibly because married consumers tend to naturally consider long-term economic benefits—EVs have lower long-term usage costs compared to traditional vehicles [50]. Consumers with dedicated parking spots are 1.399 times more likely to fall into the high endorsement category than those without, reflecting the convenience of installing private charging facilities. Age is also significant; with every additional year of age, the likelihood of being in the high endorsement group decreases to 0.974 times that of the low endorsement group. Other factors, such as the number of elderly family members and educational background, did not exhibit a significant impact on the EVs’ perceived value.

In analyzing the differences between the moderate endorsement and low endorsement consumers, the number of family members emerged as important. Specifically, for each additional family member, consumers were more likely to fall into the low endorsement category. Having elderly individuals (aged 60 and above) in the household also made respondents 2.898 times more likely to be in the moderate endorsement group. The level of education also significantly impacts membership in these two groups. Compared with consumers with lower education levels, consumers with a Bachelor’s degree are 0.308 times to belong to the moderate endorsement group, while consumers with a Master’s degree or above are 0.301 times to belong to the low endorsement group. This suggests that higher education levels may be associated with critical thinking regarding EV technology or sensitivity to purchasing costs [51].

Income level emerged as a significant influencer as well. Compared to the low endorsement group, higher-income consumers are more likely to highly endorse EVs. For every additional 50,000 RMB in income, the likelihood of being in the high endorsement group increases by 19.9%. Additionally, consumers with car purchase subsidies are 12.8 times (for the high endorsement group) and 17.3 times (for the moderate endorsement group) more likely to be in their respective groups than those without subsidies, highlighting the effectiveness of policy incentives in reducing EV purchasing costs and promoting consumer perception [36].

Regarding social policy factors, a consumer’s city’s restriction policies and parking benefits significantly affect the perceived value of EVs [52]. For example, in cities with EV time-limited free parking policies, consumers are 7.465 times more likely to be in the high endorsement group than in the low endorsement group. Additionally, the presence of gasoline vehicle restriction policies increases the likelihood of falling into high and moderate endorsement group categories by 68.8% and 16.7%, respectively.

4.5. RMM Analysis: Outcome Variable

To further delineate differences in EV perceived value among EV user categories, data for the five most pertinent vehicle attributes reported by users were gathered as identified in the literature. These attributes include fuel type, vehicle class, price, range, and charge time. Fuel type and vehicle class are categorical variables; price, range, and charge time are continuous variables. The analysis of these attributes as outcome variables is detailed in

Table 7. Automotive attribute perception analysis.

Class	Prob			Chi-Square Test	Pair by Pair
Fuel type	BEV	PHEV
Class 1	0.597	0.403		40.425 ***	Class 3 > Class 2
Class 2	0.742	0.258			Class 2 > Class 1
Class 3	0.915	0.085			Class 3 > Class 1
Vehicle size	Mini and Small	Compact	Medium and above	Chi-square test	Pair by pair
Class 1	0.119	0.462	0.419		Class 1 > Class 2
Class 2	0.245	0.455	0.3	20.358 **	Class 3 > Class 1
Class 3	0.186	0.373	0.441		Class 3 > Class 2
Price	Mean	S.E		Chi-square test	Pair by pair
Class 1	20.439	0.561			Class 1 > Class 2
Class 2	19.183	1.233		1.296 *	Class 3 > Class 2
Class 3	19.424	1.074			Class 1 > Class 3
Range	Mean	S.E		Chi-square test	Pair by pair
Class 1	703.865	21.539			Class 1 > Class 2
Class 2	629.464	41.953		23.706 ***	Class 1 > Class 3
Class 3	533.388	27.633			Class 2 > Class 3
Charge time	Mean	S.E		Chi-square test	Pair by pair
Class 1	430.886	12.035			Class 2 > Class 1
Class 2	449.546	17.833		5.37 *	Class 3 > Class 1
Class 3	483.187	19.165			Class 3 > Class 2

Note: *, **, and *** indicate significance levels of 10%, 5%, and 1%.

.

The results indicate significant differences in perception of fuel types among consumer categories. The proportions of high endorsement group members purchasing BEVs and PHEVs are 59.7% and 40.3%, respectively; in the moderate endorsement group, 74.2% purchased BEVs and 25.8% purchased PHEVs. In the low endorsement group, 91.5% purchased BEVs and 8.5% purchased PHEVs. Chi-square tests confirm these differences as significant. When comparing vehicle classes, with BEVs as the reference, the order is Class 3 > Class 2 > Class 1. Vehicle sizes were delineated as “mini and small”, “compact”, and “medium and above”. The high endorsement group’s purchase proportions for these classes are 11.9%, 46.2%, and 41.9%, respectively. For the moderate endorsement group, they are 24.5%, 45.5%, and 30%; for the low endorsement group, they are 18.6%, 37.3%, and 44.1%. The chi-square tests show significant differences in vehicle class preferences among consumer categories.

In terms of price, the average purchase price of EVs for the high endorsement group is 204,390 RMB. For the moderate endorsement group, it is 191,830 RMB; for the low endorsement group, it is 194,240 RMB. The chi-square tests indicate significant differences in price preferences among user categories, with the order from highest to lowest average price being Class 1 > Class 3 > Class 2. Regarding range, the average values are 703.865 km for the high endorsement group, 629.464 km for the moderate endorsement group, and 533.388 km for the low endorsement group. The chi-square tests reveal significant differences in range preferences, with the order from highest to lowest range being Class 1 > Class 2 > Class 3. Regarding charge time, the average values are 430.886 min for the high endorsement group, 449.546 min for the moderate endorsement group, and 483.187 min for the low endorsement group. The chi-square tests show significant differences in charge time preferences, with the order from longest to shortest being Class 3 > Class 2 > Class 1.

5. Discussion

The promotion of EVs is recognized as a pivotal strategy in the transition towards greener transportation, offering a potent solution to mitigating GHG emissions. Departing from conventional research approaches, this study focuses on users who have already purchased EVs, seeking to gain insights into their feedback across various dimensions of EV value. EV users were categorized into three distinct groups based on their responses to an online questionnaire.

The high endorsement group constitutes 67% of the respondents, reflecting robust market acceptance that may bolster the confidence of EV companies in their marketing endeavors. These users perceive the purchase price of EVs as reasonable, acknowledging significant reductions in daily travel costs. Moreover, they anticipate future resale values to offset a substantial portion of their initial investment. Within this group, 88.5% find the purchase price reasonable, 98% believe their EV reduces travel costs, and 61.2% view the resale value as acceptable. Previous studies have identified higher purchase prices and a lower resale value as barriers to EV adoption [53]. However, since 2024, decreasing lithium battery production costs and intensifying market competition have led to a continuous drop in EV purchase prices in China, narrowing the resale value gap between EVs and gasoline vehicles. Thus, more users now perceive EV purchase prices as reasonable and find their resale value acceptable. Low travel costs have also consistently been a positive factor in attracting users to EVs; in this study, 98% of the respondents expressed their approval of travel costs. Prior research has indicated that EVs’ low travel costs contribute to long-term savings, helping owners reduce expenses over time [54].

In terms of environmental impact, users regard EVs as effective tools for curbing GHG emissions, reducing urban air pollution levels, and enhancing their overall quality of life, thereby contributing positively to green and sustainable development. Within the high endorsement group, 73.4% believe EVs aid in reducing GHG emissions while 76.9% perceive them as instrumental in emissions reduction at the individual level. This finding is supported by previous studies suggesting that emissions reduction benefits positively influence EV adoption, particularly among environmentally conscious individuals. Similarly, Cheng, Chen, and Lin highlighted environmental value as a primary motivator for EV purchases among consumers in Taiwan [55]. Wang et al. (2024) also found that consumers who identify as environmentally conscious are more inclined to buy EVs [56].

Emotional value, stemming from one’s environmental awareness, positions EVs as eco-friendly transportation options capable of fostering a positive sense of social responsibility. By choosing EVs, individuals actively engage in combating pollution and climate change, thus embodying a strong sense of contributing to society. EV users are often identified as individuals deeply committed to sustainable living and reducing their carbon footprints. Within the high endorsement group of respondents in this study, 73.2% believe EVs contribute to a positive personal image, while 89% feel that EVs elicit positive social feedback. This aligns with previous studies indicating that certain EV consumers associate their green self-image with qualities such as “energy saving”, “environmental protection”, “emissions reduction”, “social responsibility”, and “care for other people” [57].

Concerning social value, the high endorsement group reported receiving affirmative support from friends and social circles upon purchasing EVs, fostering a sense of social belonging and recognition. Within this group, 82.6% believe EV ownership enhances social integration and 77.8% feel acknowledged and supported by their peers. Social value plays a significant role in EV adoption [58]. For instance, as more people begin using EVs, it becomes harder to criticize early adopters, and choosing to drive gasoline vehicles may increasingly be seen as unconventional or even anti-social.

Furthermore, improvements in charging infrastructure and advancements in battery technology have facilitated the widespread acceptance of EVs, thereby fueling market growth. Finally, regarding technological value, the high endorsement group reported appreciating the integration of internet technology in EVs. This advancement appears to offer diverse driving experiences that effectively cater to varied consumer preferences. Innovative features like autonomous driving, automatic parking systems, and faster acceleration compared to gasoline vehicles enhance the appeal of EVs [59]. Within this group, 68.4% perceive EVs as providing a unique driving experience. Existing studies similarly show that intelligence, acceleration, responsiveness, smoothness, and low noise can motivate consumers to adopt EVs [57].

The moderate endorsement group contains 21% of this study’s respondents. This group is characterized by nuanced attitudes towards EVs, with a strong emphasis on environmental protection and a desire to mitigate GHG emissions through EV usage. However, they expressed a low perception of the effectiveness of EVs in reducing emissions. Within this group, 49.8% believe EVs effectively reduce GHG emissions, while 34.2% perceive them only as tools in reducing personal travel-related pollutant emissions. The primary challenge here stems from the reliance on traditional energy sources like coal in China’s current electricity generation structure, undermining the environmental benefits of EVs [60]. Many consumers in China continue to question the environmental benefits of EVs. However, studies have confirmed that EVs can reduce emissions over their entire life cycle [61]. Even in provinces with coal-dominant power structures (e.g., Shanxi and Inner Mongolia), EVs produce lower GHG emissions than gasoline vehicles of the same size [38]. Clarifying EVs’ emissions reduction capabilities in China and informing the public are essential. Increasing the proportion of renewable energy in electricity generation could also further enhance the environmental benefits of EVs and improve consumers’ perceptions of them in the future [62]. Additionally, information asymmetry may prevent EV users from fully understanding the full life-cycle emissions data of EVs, resulting in a superficial perception of their environmental benefits [63].

Regarding price, this group acknowledges the potential of EVs to reduce travel costs, but find their purchase price relatively high compared to traditional internal combustion engine vehicles. Within this group, 47% find the purchase price of EVs to be reasonable, 57% believe EVs could reduce travel costs, and 34.7% find the resale price reasonable. Previous research has similarly shown that certain users believe EVs can save significant money due to their low travel cost, but that this benefit is not significant enough to improve their perception in the short term. Furthermore, within the moderate endorsement group, variations in socioeconomic status and vehicle procurement budgets may induce substantial heterogeneity in price perception [64]. Social factors also shaped their decisions, with many swayed by the opinions of friends when opting for EVs as their mode of transportation. While 51.3% in this group consider others’ opinions important in making purchasing decisions, the complexity of social identity creates ambivalence—they seek social approval through EV adoption yet resist potential ‘environmentalist’ labeling. Moreover, they perceive that the technological value offered by EVs falls short of their expectations—only 14% believe EVs have met their expectations for automotive technological innovation. This phenomenon may stem from consumers’ inflated expectations regarding EV technological capabilities or insufficient understanding of EV performance parameters, resulting in a pronounced expectation/performance gap [65]. The purchase decision-making process is co-determined by multiple interdependent factors, including environmental values, cost considerations, social influences, and technological perceptions. The complex interplay of these determinants generates behavioral ambivalence and decision-making paradoxes, ultimately reflecting the inherent instability in this cohort’s value cognition framework [66].

The low endorsement group contains 17% of the respondents. As illustrated in Figure 3, this group generally believes that EVs fail to meet their expectations across various dimensions of value. Their choices are likely primarily driven by policy incentives such as purchase subsidies, exemptions from purchase taxes, and the provision of free parking spaces for EVs. Shang et al. found that for every 1% increase in purchase subsidies, EV sales grow by 1.36% and the market share of EVs increases by 2.31% [36]. While these policies have bolstered short-term EV sales, they are not sustainable long-term solutions. Over-reliance on government subsidies to promote market share may temporarily alleviate financial barriers to EV purchases but fails to address inherent issues with EVs, such as battery range issues, ineffective charging infrastructure, vehicle performance problems, or consumers’ overall impression of EVs [67]. Prolonged reliance on subsidies not only strains government finances but also distorts the market, impeding natural progress in EV technology and market competition.

Moreover, if consumers base their choices solely on short-term financial incentives, they may revert to traditional gasoline vehicles once government subsidies are withdrawn or reduced. This not only undermines the original aim of promoting EVs but also introduces greater market volatility. According to Hardman and Tal (2021) [9], for instance, up to 20% of PHEV owners and 18% of BEV owners in California were likely to abandon these vehicles due to their dissatisfaction with the limitations of EVs, such as the costs and inconvenience of charging them and insufficient charging stations.

In the endeavor to promote EVs, it is essential to recognize that user perception is profoundly influenced by personal characteristics, social aspects, policy environments, and economic factors. Stakeholders must tailor their marketing strategies accordingly. For instance, homeowners with private parking spaces tend to exhibit higher perceptions of EVs. To enhance the appeal of EVs, offering additional incentives such as convenient home charging facilities for these users may be effective [68]. For renters, policymakers could consider adding or subsidizing public charging stations to alleviate EV-related inconveniences. The increasing size of households can decrease perceptions of EVs due to their limited range and the necessity for frequent charging [69]. Promotional strategies should prioritize offering a diverse range of vehicle options—particularly spacious models designed for larger families—while emphasizing the safety and convenience of EVs [33].

Additionally, older users tend to express a lower perceived value of EVs. This is likely due to their familiarity with traditional gasoline vehicles and limited adaptability to new technologies. Implementing relevant policies is crucial to enhance the user experience for older individuals. This could include launching EV operation education programs specifically targeting older users, facilitating quicker adaptation to new technology. Emphasizing the long-term cost-saving benefits and the environmental advantages of EVs can also promote acceptance among older users, thus furthering the development of the EV market.

For higher-income groups, promotional efforts can focus on highlighting the high-end quality, technological innovations, status signaling, and increased quality of life associated with EVs. Conversely, for lower-income consumers, a dual approach of offering purchase subsidies and charger installment benefits, along with manufacturing more models suitable for their demographic, could be more effective. Social policies such as free parking for EVs within a limited timeframe and restrictions on the purchase of gasoline vehicles can greatly enhance user perceptions of EVs [70], suggesting that policymakers can meaningfully contribute by providing free or discounted urban parking permits for EV users or expanding the integration of EVs into public transportation and shared mobility services. In summary, the promotion of EVs requires comprehensive consideration of the specific needs and perception of various user groups. Developing and implementing multidimensional, targeted policies and marketing initiatives is essential to enhance user perception as well as market acceptance [71].

Understanding the attributes and preferences of different consumer groups provides critical insights for marketing strategies and product positioning, enabling manufacturers and policymakers to better satisfy market demands and thereby promote the adoption of EVs. The high endorsement group in this study, for instance, expects a higher driving range, suggesting that EV manufacturers should prioritize battery technology. For price-sensitive groups, cost control and more competitive pricing strategies may be effective. Users also highlight the importance of charging time as an important indicator of convenience, underscoring the importance of enhancing charging efficiency and expanding charging infrastructure [69]. Moon et al. similarly found that users favor EVs with lower prices, longer battery ranges, better fuel economy, and shorter charging times through discrete choice model analysis [33]. These findings provide valuable consumer behavior information that can support innovative approaches to EV promotion.

6. Conclusions and Policy Implications

Feedback from users regarding the value derived from their vehicles is crucial information regarding the promotion of EVs. This study centers on consumers who have already purchased EVs, analyzing their perception based on five key dimensions of value. Acknowledging the diversity in consumer demands and preferences, this research employed an LCM to categorize respondents and identified three distinct groups as the most appropriate categorization scheme. By computing the conditional probabilities of different values, the respondents were classified into high endorsement, moderate endorsement, and low endorsement categories, with corresponding proportions of 62%, 21%, and 17%, respectively. While the majority of the respondents held a positive attitude towards the values of their EVs, a minority expressed a low perception of the values of their EVs.

To further explore the multiple factors influencing consumer category membership, this study incorporated personal attributes, social attributes, and economic attributes as predictive variables in a regression mixture analysis. The results reveal that variables such as property ownership, family size, marital status, the presence of children under 10 years old or individuals over 60 years old in the household, education level, household income, purchasing subsidies, and policies limiting gasoline vehicles significantly affect category membership. Variables including fuel type, charging time, price, and vehicle size were also analyzed to better understand the perception associated with different EVs among respondents in all categories. The findings reveal significant differences across categories—for instance, the high endorsement group typically favors EVs with shorter charging times and a longer range, while being willing to accept higher prices. Significant differences were also observed in vehicle size and fuel type preferences among the three categories.

Several measures may be implemented to uphold positive attitudes towards EVs among users who already highly endorse them. Firstly, higher levels of after-sales services (e.g., extended warranties, top-quality maintenance services, and swift problem resolution mechanisms) can ensure that EV users feel fully supported and cared for throughout their ownership experience. Secondly, continuous improvements in charging infrastructure are imperative—this should include an increase in the number of charging stations, enhanced charging speed, and the provision of real-time status information on charging stations to enhance convenience [72]. Thirdly, constant innovation and the introduction of new EV models are essential to meet consumers’ diverse needs. Lastly, businesses should establish positive communication channels with users, regularly gather their feedback, understand their concerns, and make targeted adjustments accordingly.

Some EV users have reverted to gasoline vehicles due to having a low perceived value of EVs. In this study, such users primarily belonged to the moderate and low endorsement groups. Several strategies may improve their attitudes towards EVs and prevent a future shift back to gasoline vehicles. In users moderately endorsing EVs, concerns about environmental performance can be addressed by providing more detailed information and research data to emphasize the benefits of renewable, clean energy. For those with higher price sensitivity, more attractive purchasing incentives may be effective (e.g., reducing purchase costs of EVs, offering subsidies, or establishing tax benefits) [73].

It is crucial to fully understand the reasons that some individuals minimally endorse EVs and find appropriate ways to address their dissatisfaction. This may require further investigative research to identify the root causes of this, such as issues with EV performance, charging infrastructure, or battery range. Targeted marketing campaigns can be launched to highlight the performance and convenience of EVs while dispelling misconceptions. Moreover, expanding the charging network can enhance the convenience and availability of EVs across the market [72].

Additionally, strengthening research and innovation in EV-related technologies can improve overall performance to satisfy a broader range of consumer needs. In summary, by adopting comprehensive measures including improving information dissemination, adjusting purchasing policies, enhancing technological performance, and expanding infrastructure, it is possible to effectively change the EV perceptions of moderate endorsement and low endorsement users, thereby encouraging them to continue using them and thereby laying a foundation for green, sustainable transportation in the future.

Latent class variables were employed as moderating variables to explore the impact of individual attributes on attitudes towards EVs across different categories, aiming to gain a deeper understanding of EV users’ perceptions. The findings may represent valuable insights for the development of effective marketing strategies and policy initiatives. It should be emphasized that the environmental benefits of EVs are contingent upon the sustainability of their energy sources. Electrification constitutes the most environmentally benign propulsion system across all vehicle categories, provided the batteries are charged using renewable energy. However, when powered by fossil fuels or non-renewable resources, EVs’ environmental advantages are significantly compromised [74].

Despite these contributions, this study has several limitations. Firstly, while the respondents are distributed across first-, second-, and third-tier cities, this research primarily focuses on one province in China, which may limit the generalizability of the findings. Secondly, it is possible that the set of predictive variables considered in this study was not sufficiently large. In reality, factors affecting user perception are more diverse—they include the impact of cultural and social backgrounds, which may vary significantly across different regions. Thirdly, given the context-specific nature of this study, our findings are strictly applicable to EV adoption analyses and cannot be directly generalized to general populations or alternative vehicle categories.

Future research can be conducted from the following perspectives: Firstly, the carbon reduction benefits of various vehicle types should be considered. The extent of carbon dioxide reduction varies across different vehicles due to differences in technology, energy sources, and policy support. EVs exhibit the greatest potential for reducing carbon dioxide emissions among all vehicle types. Fuel cell vehicles also demonstrate significant carbon reduction potential, particularly when hydrogen is used as fuel [75]. A comparative analysis between fuel cell vehicles, hybrid vehicles, and diesel vehicles can be conducted to evaluate their respective environmental impacts. Moreover, it is important to assess whether carbon emissions may shift from the consumption stage to the production stage and to accurately calculate the full-life-cycle carbon emissions of different vehicle types. Secondly, an intelligent optimization framework should be built based on digital twin technology, and machine learning algorithms should be deeply embedded into the vehicle system development process [76]. Machine learning (ML), as a data-driven and intelligent analytical approach, has been widely adopted in diverse fields such as linguistics, psychology, remote sensing, and urban planning. ML can be integrated with statistical techniques such as random forests, Bayesian models, and traditional population-level statistical methods to enable simultaneous analysis at both individual and group levels, thereby offering a more comprehensive analytical framework. Additionally, alternative methodologies, including the Taguchi method, big data analytics, and the integration of artificial intelligence with statistical tools, can be selectively applied based on the specific research objectives to improve both the efficiency and accuracy of the research.