Determinants of Citizen Satisfaction with Toll Road Infrastructure: A Hierarchical Regression Model from Mexico with Potential Implications for Other Emerging Countries

Abstract

Background: Public satisfaction with public transport infrastructure is a factor in the social legitimacy of infrastructure investment policies. Methods: This study analyzes the determinants of citizen satisfaction with toll roads in Mexico using a hierarchical regression model applied to a nationally representative survey. Results: Satisfaction does not depend primarily on sociodemographic factors, but rather on users’ overall perception of the quality, safety, and management of the road system as a whole. Furthermore, the pattern of predictors varies according to usage experience, suggesting that satisfaction is influenced by different factors among users and non-users of these facilities. These findings support a contextual evaluation model, in which citizen assessments are based more on systemic interpretations than on isolated experiences. Conclusions: The study has direct implications for public policy design and infrastructure management in contexts where the use of toll roads responds to structural constraints rather than voluntary decisions. Although the study focuses on the Mexican case, its contributions offer useful interpretative insights for other countries with similar challenges in terms of mobility and institutional legitimacy.

1. Introduction

Road infrastructure is much more than just a technical support system for the movement of people and goods, as it is a key component of economic development, territorial cohesion, and collective well-being. Its layout, accessibility, and maintenance reflect decisions that, although technical in appearance, are deeply rooted in political and social priorities. In this sense, communication routes are also a tangible indicator of the link between citizens and institutions [1].

According to the World Health Organization (WHO) [2], traffic accidents cause around 1.19 million deaths and 50 million injuries per year. More than 90% of these deaths occur in low- and middle-income countries, highlighting not only deficits in investment and planning, but also unequal protection for the most vulnerable users of the system. This reality has prompted the adoption of strategic frameworks such as the “safe system” approach, which forms the basis of the 2030 Road Safety Strategy promoted by the General Directorate of Traffic (DGT). This perspective recognizes the inevitability of human error and proposes that the road system be designed to minimize its consequences, integrating coordinated actions in infrastructure, speed regulations, vehicle safety, and post-accident response [3,4,5].

From this framework, it becomes clear that traditional technical indicators, although necessary, are insufficient to understand the complexity of the road environment. Citizen perception (i.e., how people experience and interpret aspects such as safety, justice, or quality of service) offers an additional dimension that can be crucial both for the legitimacy of public policies and for the effectiveness of interventions [6,7,8,9]. Exploring how these perceptions are shaped in contexts marked by structural inequalities allows us to identify patterns that are useful not only for local analysis but also for anticipating challenges shared by other countries with similar characteristics. This work focuses on a national case, but with a keen eye on dynamics that transcend the local. By combining empirical data with structural elements, it seeks to provide insights that, without claiming to be universal formulas, may be useful for rethinking public policies and mobility strategies in other contexts facing similar challenges.

1.1. Infrastructural Inequalities and Public Trust in Latin America: The Case of Mexico

Latin America and the Caribbean are characterized not only by high levels of economic inequality, but also by deep gaps in road safety. There are substantial differences between countries in the region. While some, such as Uruguay, have made significant progress in safe infrastructure and sustained prevention policies, others continue to experience high accident rates, associated with both structural deficits and negative public perceptions of road safety [10,11]. Although this heterogeneity is striking, it does not prevent comparison between contexts; on the contrary, it reinforces the usefulness of studies such as this one, where the analysis of public perception can provide key elements to guide public policies, urban or interurban design interventions, and adaptive strategies that can be replicated in environments with similar challenges [10].

In general terms, Latin America faces a severe impact from road accidents. Each year, approximately 110,000 deaths are recorded from this cause, with economic losses equivalent to 2–3% of collective GDP [11,12]. In countries such as Mexico, the annual average number of road traffic deaths ranges from 12,000 to 16,000, placing it among the five Latin American countries with the highest absolute number of victims [13]. Although indicators vary between countries, the underlying factors tend to coincide: poor road design, regional inequality in infrastructure provision, and institutional limitations in risk management.

Many of these challenges stem from structural deficiencies in the road network. It is particularly revealing that nearly 75% of trips in the region are made on roads rated less than three stars according to the International Road Assessment Programme (iRAP), which highlights persistent shortcomings in quality, design, and maintenance [12]. This methodology, implemented in more than 100 countries, rates roads from one to five stars according to the estimated risk to users, taking into account factors such as speed, visibility, geometry, access control, and protective elements [14]. In Mexico, companies such as Aleatica have incorporated this system into their technical audits to prioritize investments that contribute to safer mobility [15].

However, these limitations not only compromise objective safety, but also the public perception of justice and institutional trust. Studies conducted in cities such as Bogotá, Lima, and Santiago de Chile show that citizens perceive deficiencies even in infrastructure that, from a technical point of view, meets certain minimum standards but has visible flaws (such as insufficient lighting or poor signage) that affect the daily experience of users [16,17]. This disconnect between the normative and the lived experience often fuels a perception of inequality, especially when certain sectors (usually peripheral or with less political influence) receive less attention in the allocation of resources. As has been documented, this perception of infrastructural inequality can undermine the legitimacy of public policies, even when there are objective improvements in indicators.

In Mexico, this phenomenon is particularly evident in the context of rapid growth in the number of vehicles and the limited adequacy of road infrastructure. In 2023, there were more than 58 million registered vehicles in the country, with a 160% increase in motorcycles over the last decade [18]. For its part, public and private investment in infrastructure has not kept pace, because of the estimated total of 780,511 km of national road network, only 51,159 km correspond to federal highways, of which 10,527 km are toll highways managed through a mixed model between private concessionaires and Caminos y Puentes Federales (CAPUFE) [19,20]. These highways account for a substantial portion of national traffic and, as studies in countries with similar concession schemes have suggested, their performance can influence public perception of fare equity, model efficiency, and institutional transparency [8].

1.2. Toll Roads as a Subject of Public Evaluation

Toll roads in Mexico provide a particularly useful setting for observing how citizens interpret the quality and legitimacy of road infrastructure [21]. It is not just a matter of evaluating their technical performance, but also of understanding how subjective dimensions such as perceived safety, fare fairness, and institutional trust are articulated. According to the CAPUFE Satisfaction Survey (2024) [22], users rate the service with an average of 8.13 out of 10, highlighting safety (85.7%) and traffic flow in particular. On the other hand, toll costs and pavement conditions are identified as the main sources of dissatisfaction.

The National Survey on Government Quality and Impact [23] reinforces this perspective, showing that 70% of toll road users express satisfaction, compared to 33% of those who use free roads. This difference reveals how public perception is built not only on direct experience, but also through contrasts between different types of infrastructure and expectations about what a quality public service should offer.

In comparative terms, the Inter-American Development Bank [1] places Mexico in an intermediate position compared to other countries such as Chile or Spain, where concessioned roads have achieved higher levels of public acceptance. In the Mexican context, the perception tends to recognize the superior quality of toll roads, although accompanied by a certain ambivalence regarding the fairness of charges and transparency in the allocation of resources.

To understand this diversity of assessments, it is useful to turn to environmental and transportation psychology, disciplines that distinguish between functional perception (linked to objective quality), symbolic perception (related to social meanings such as reciprocity or exclusion), and institutional perception (associated with trust in service management) [24,25,26]. These approaches allow us to interpret how different people with similar experiences can form divergent opinions about the same road.

Although the technical approach has been predominant in many studies, more comprehensive analyses that incorporate the subjective dimension of the citizen experience are beginning to emerge. Recent research in Latin America shows that aspects such as visible maintenance, signage, safety, and treatment received influence citizens as much as the infrastructure itself [16,17]. In this sense, understanding highways as spaces for social and institutional interaction is key to moving toward more equitable, sustainable, and legitimate mobility models.

1.3. Study Objectives

This study seeks to understand what factors explain citizen satisfaction with toll roads in Mexico, considering both the tangible aspects of the infrastructure and the interpretations that users construct around it. Beyond measuring approval levels, the focus is on identifying which elements influence how the service is valued.

Based on previous evidence, three main assumptions are made. First, it is expected that aspects directly related to the user experience (such as the perception of safety, maintenance, and signage) are closely linked to higher levels of satisfaction [22,26]. Second, it is considered that the perception of fairness in toll payment, understood as the relationship between cost and service quality, may be a key point of friction that affects users differently depending on their profile [27]. And thirdly, it is hypothesised that the pattern of predictors of satisfaction with toll roads differs depending on the reported use of these infrastructure facilities.

2. Materials and Methods

2.1. Participants

The sample consists of 1469 people and is nationally representative with regard to gender and age, with a confidence level of 95% and a margin of error of +/−3.1%. It includes citizens over the age of 18 residing in Mexico. The questionnaire was administered nationwide, with an oversample in Naucalpan, Tlanepantla, Ecatepec, Chimalhuacán, Tultepec, and Mexico City. Priority was therefore given to states with high population densities where tolls are in operation, to ensure that the phenomenon under study was present. This was achieved through stratified sampling, ensuring randomness within each group. The main sociodemographic characteristics of the sample are shown in

Table 1. Sociodemographic characteristics of the sample.

Sociodemographic Characteristics		n	%
Gender	Male	743	50.6%
	Female	726	49.4%
	Total	1469	100%
Age	18 to 24 years old	263	17.9%
	25 to 34 years old	312	21.2%
	35 to 44 years old	263	17.9%
	45 to 59 years old	368	25.1%
	60 years old or older	263	17.9%
	Total	1469	100%
Level of education	Primary education or lower	244	16.6%
	Secondary education	415	28.3%
	High school/preparatory school	529	35.9%
	University studies	256	17.5%
	Postgraduate studies	25	1.7%
	Total	1469	100%
Employment status	Unemployed	75	5.1%
	Retired	93	6.3%
	Student	111	7.6%
	Full-time employee	610	41.5%
	Household	381	25.9%
	Other	199	13.5%
	Total	1469	100%

.

In terms of the characteristics of road users in the sample, 38.8% (n = 570) of participants identified themselves as motor vehicle drivers. Of these, 78.1% (n = 445) had a driver’s license and 21.9% (n = 125) did not. Among drivers, 9.6% (n = 55) reported having received a penalty during the last year, while 90.4% (n = 515) reported no violations. Meanwhile, 32.9% (n = 483) say they have been involved in a traffic accident, compared to 67.1% (n = 986) who have not had this experience (

Table 2. Characteristics of the sample of road users.

Characteristics of Road Users		n	%
Motor vehicle driver	Yes	570	38.8%
	No	899	61.2%
	Total	1469	100%
Driver’s license	Yes	445	78.1%
	No	125	21.9%
	Total	570	100%
Penalties received in the last year	Yes	55	9.6%
	No	515	90.4%
	Total	570	100%
Road crashes experienced	Yes	483	32.9%
	No	986	67.1%
	Total	1469	100%
Frequency of toll road use	Does not use	598	42.6%
	Sometimes	498	35.4%
	Regularly	309	22.0%
	Total	1405	100.0%
Perception of toll prices	Expensive	714	63.0%
	Fair price	384	33.9%
	Cheap	36	3.2%
	Total	1134	100.0%

).

At the same time, variables related to users’ direct experience with toll roads were collected. In terms of frequency of use, 42.6% (n = 598) of participants reported not using them, 35.4% (n = 498) indicated that they used them occasionally, and 22.0% (n = 309) stated that they used them regularly (at least twice a week) (Table 2). Regarding the perception of toll prices, the majority of respondents (63.0%, n = 714) consider them to be expensive.

2.2. Instruments

For the study, data obtained from a national survey conducted between 2 and 14 July 2024, designed by the Aleatica Foundation and the Institute of Traffic and Road Safety (INTRAS) of the University of Valencia, was analyzed. The anonymous, self-administered questionnaire took an estimated 25 min to complete and was distributed to road users in different regions of the country. Data was collected by SIMO Consulting through the administration of a questionnaire in person at the participants’ homes.

The instrument included thematic blocks focusing on mobility practices, perception of road safety, evaluation of infrastructure and user experience, as well as sociodemographic and driving variables. Within this broader framework, a specific section on toll roads was included, aimed at gathering information on satisfaction with the service, maintenance, signage, cost–benefit ratio, and confidence in management. The results analyzed in this study correspond to this module, with the aim of delving deeper into the factors that influence citizens’ assessment of this type of infrastructure.

The variables analyzed are derived from the following items:

• Satisfaction with toll roads: this was measured through a direct question in which respondents were asked to rate their level of satisfaction with toll roads in general. Responses were recorded on a 5-point Likert scale (1 = very dissatisfied, 5 = very satisfied).
• Perception of toll prices: participants rated the cost of toll roads as “cheap,” “fair price,” or “expensive”. This variable allows us to explore the subjective perception of the economic value of the road service.
• Frequency of toll road use: respondents were asked how often they use these roads. The responses were categorized into three groups: “does not use,” “sometimes,” and “regularly (at least twice a week)”.
• Infrastructure quality perception index: a composite index was constructed from five items that assessed perceptions of various types of road infrastructure: the country in general, urban areas, interurban areas, national highways, and vehicles in circulation. Each item was rated on a Likert scale from 1 to 5. The index showed acceptable internal consistency (Cronbach’s α = 0.728), with all corrected item-total correlations above 0.40.
• Perception of road safety: a specific question was included that asked how citizens rate the country’s road safety. The response scale was from 1 to 5, where 1 indicated “very unsafe” and 5 indicated “very safe”.
• Characteristics as a road user: including items on frequency of car use as a driver, possession of a driver’s license, involvement in road accidents, and receipt of traffic penalties.
• Sociodemographic variables: data was collected on gender, age, educational level, and employment status. These variables were included in order to explore possible differences in the evaluation of tolls according to user profile.

2.3. Data Processing

Statistical analyses were performed using ©IBM SPSS (Statistical Package for Social Sciences) version 26.0 (Armonk, NY, USA) [28]. Specifically, descriptive analyses (frequencies, means, and standard deviations) were performed to analyze Mexican citizens’ social perceptions of traffic, mobility, and road safety issues, as well as their satisfaction with toll roads. In addition, one-way analysis of variance (ANOVA) and Tukey’s post hoc tests were applied to compare levels of satisfaction and perception of road infrastructure between groups. Pearson or Spearman correlations (depending on the nature of the variables) were also calculated to examine the bivariate relationships between the main study indicators.

Finally, a hierarchical linear regression analysis was performed to identify the predictive factors of satisfaction with toll roads. In a first step, the sociodemographic variables were introduced; in a second step, the experiential variables (frequency and perception of price); in a third step, the general perception of road infrastructure; and in a fourth step, the economic and experiential variables of users. This model has made it possible to evaluate the relative weight of each set of variables in explaining citizen satisfaction with toll roads. The block sequence of the hierarchical model follows an explicit theoretical logic grounded in the literature. Firstly, sociodemographic variables are included as control predictors, as they represent structural characteristics of the individual that precede any attitudes or experiences, and whose effect must be isolated before assessing the influence of other factors. Secondly, driving habits are incorporated as indicators of exposure and experience [29,30]. Thirdly, perceptions of infrastructure quality and road safety constitute the evaluative core of the model, as these cognitive assessments represent a relevant direct antecedent of satisfaction [31,32,33]. Finally, economic variables and negative experiences are introduced in the last block as moderating factors that influence satisfaction once general perceptions have been controlled for. In addition, hierarchical linear regression analyses have been carried out specifically on the group of users and the group of non-users to identify potential differences based on experience with toll roads in relation to the predictors of satisfaction with this infrastructure in Mexico.

3. Results

3.1. Descriptive Analysis

The average infrastructure quality perception index was 2.93 (SD = 0.97). Average satisfaction with toll roads was 3.18 (SD = 1.28). The distribution was approximately normal (skewness = −0.04, kurtosis = −0.85). Likewise, a quarter of users (25.8%) expressed dissatisfaction, while 40.1% expressed satisfaction.

Table 3. Descriptive statistics of the study variables and Pearson and Spearman correlations.

	2	3	4	5	6	7	8
1. Age	−0.322 **	0.006	−0.054 *	−0.113 **	−0.070 *	0.044	−0.335 **
2. Level of education	-	−0.203 **	−0.026	−0.026	−0.007	0.183 **	0.043
3. Frequency of car use as a driver		-	−0.009	−0.028	0.005	−0.202 **	−0.046
4. Satisfaction with toll roads			-	0.355 **	0.339 **	0.011 *	0.148 **
5. Infrastructure quality perception index				-	0.634 **	−0.102 **	−0.129 **
6. Perception of road safety					-	0.016	−0.063 *
7. Frequency of toll use						-	−0.001
8. Perception of toll prices							-

* The correlation is significant at the 0.01 level. ** The correlation is significant at the 0.05 level.

presents the relationships between the variables, revealing a positive and significant association between satisfaction with toll roads and the general infrastructure quality perception index (r = 0.355, p < 0.001) and with the specific perception of road safety in Mexico (r = 0.339, p < 0.001). This suggests that as the assessment of infrastructure and road safety improves, satisfaction with toll roads also increases.

Age showed a negative correlation with satisfaction (r = −0.054, p = 0.040), indicating that older participants tend to be less satisfied with toll roads. This variable was also negatively associated with the perception of infrastructure (r = −0.113, p < 0.001) and the perception of safety (r = −0.070, p < 0.05), suggesting that older people tend to evaluate both road quality and overall road safety more negatively. Educational level correlated negatively with age (r = −0.322, p < 0.001), which is consistent with the demographic profile in which younger participants have higher levels of education.

The frequency of car use as a driver did not show a significant relationship with satisfaction (r = −0.009, p > 0.05), perception of infrastructure (r = −0.028), or perception of road safety (r = 0.005). Furthermore, it correlated negatively with educational level (r = −0.203, p < 0.001), indicating that the higher the level of academic training, the less use of the car as a driver. Regarding the frequency of toll use, no significant correlations were found with satisfaction (r = 0.011), perception of infrastructure (r = −0.102), or perception of safety (r = −0.063, p < 0.05). Although the latter is significant, the effect size is very small, reinforcing the idea that frequent use does not imply greater satisfaction. Perception of price was positively related to satisfaction (r = 0.148, p < 0.001), indicating that those who consider tolls to be fairly priced tend to report greater satisfaction. A negative correlation was also observed between this variable and the perception of infrastructure (r = −0.129, p < 0.001), suggesting that those who perceive the infrastructure as poor also rate the price negatively.

3.2. Hierarchical Regression Model

A hierarchical linear regression analysis was performed in four sequential blocks to identify predictors of satisfaction with toll roads (

Table 4. Hierarchical regression model summary: Block-level fit statistics predicting toll road satisfaction.

Step	Additional Variables	R	R2	ΔR2	F Change	Sig. Change	Adjusted R2
1	Sociodemographic -     Age -     Level of education -     Gender	0.072	0.005	—	0.856	0.464	–0.001
2	Driving habits -     Frequency of car use -     Frequency of toll road use	0.075	0.006	0.000	0.093	0.911	–0.005
3	Road infrastructure: -     Infrastructure quality perception index -     Perception of road safety	0.329	0.108 *	0.102	27.867	<0.001	0.095
4	Economic variables and experiences -     Perception of toll prices -     Road crashes -     Penalties	0.352	0.124 **	0.016	2.862	0.036	0.106

* The correlation is significant at the 0.01 level. ** The correlation is significant at the 0.05 level.

). In the first step of the model, only age, educational level, and gender were included, and this did not explain significant variance in satisfaction (R² = 0.005, F(3,489) = 0.856, p = 0.464). No sociodemographic predictor reached statistical significance. In a second phase, variables related to driving habits were incorporated, such as frequency of car use as a driver and frequency of toll road use, which also did not improve the predictive power of the model (ΔR² < 0.001, F(2,487) = 0.093, p = 0.911), suggesting that these variables do not determine satisfaction.

The substantial change occurred with the inclusion, in a third stage, of perceptions about road infrastructure. This block significantly increased the explained variance by 10.2 percentage points (ΔR² = 0.102, F(2,485) = 27.867, p < 0.001), reaching an R² = 0.108. Both the general infrastructure perception index (β = 0.180, p = 0.002) and the specific perception of road safety in Mexico (β = 0.176, p = 0.003) were identified as significant and independent predictors. For its part, the final model, in its fourth stage, incorporated users’ perception of toll prices, experience of road crashes, and penalties. This block added a small but significant improvement (ΔR² = 0.016, F(3,482) = 2.862, p = 0.036).

Therefore, the complete model explained 12.4% of the variance in toll satisfaction (adjusted R² = 0.106, F(10,482) = 6.805, p < 0.001). The significant predictors in the final model were: (1) Overall infrastructure quality perception index (β = 0.167, 95% CI = [0.067, 0.366], p = 0.005): For each point of improvement in the perception of overall infrastructure, toll satisfaction increases by 0.217 points on the original scale; (2) Perception of road safety (β = 0.187, 95% CI = [0.067, 0.281], p = 0.001): Additional and independent effect of specific perception of national roads; and (3) Perception of price (β = 0.116, 95% CI = [0.057, 0.457], p = 0.012): Those who perceive tolls as fair or cheap report greater satisfaction than those who consider them expensive. However, having experienced road crashes (β = 0.009, p = 0.828) and having received penalties (β = −0.058, p = 0.182) did not significantly predict satisfaction (

Table 5. Hierarchical regression model coefficients predicting toll road satisfaction.

Predictors	B	E	Beta	t	p	95% CI		Tolerance	VIF
Step 1
(Constant)	1.598	0.437	—	3.660	<0.001	0.740	2.456	—
Age	0.000	0.004	0.004	0.089	0.929	−0.007	0.008	0.795	1.258
Level of education	0.017	0.053	0.014	0.316	0.752	−0.088	0.122	0.910	1.099
Gender	−0.106	0.118	−0.040	−0.900	0.368	−0.337	0.125	0.935	1.070
Step 2
Frequency of car use	−0.001	0.021	−0.002	−0.041	0.967	−0.042	0.040	0.925	1.081
Frequency of toll road use	0.068	0.068	0.044	0.999	0.318	−0.065	0.200	0.943	1.060
Step 3
Infrastructure quality perception index	0.217	0.076	0.167	2.851	0.005	0.067	0.366	0.529	1.889
Perception of road safety	0.174	0.054	0.187	3.205	0.001	0.067	0.281	0.535	1.870
Step 4
Perception of toll prices	0.257	0.102	0.116	2.526	0.012	0.057	0.457	0.861	1.161
Road crashes	0.023	0.105	0.009	0.217	0.828	−0.183	0.229	0.965	1.037
Penalties	−0.095	0.071	−0.058	−1.338	0.182	−0.235	0.045	0.984	1.017

).

In order to verify that the conclusions of the hierarchical linear regression model are not sensitive to the treatment of the dependent variable as continuous, ordinal logistic and ordinal probit regression models were estimated using the same structure of predictors. Both models showed a significantly better fit than the null model (ordinal logit: χ²(21) = 77.07, p < 0.001; ordinal probit: χ²(21) = 72.37, p < 0.001), with Nagelkerke pseudo-R² values of 0.152 and 0.144, respectively, of a similar magnitude to the adjusted R² obtained in the linear regression (0.106). The pattern of significant predictors was identical across the three models. Thus, the infrastructure quality perception index (logit: b = 0.390, p = 0.002; probit: b = 0.202, p = 0.006) and the perception of road safety (logit: b = 0.310, p < 0.001; probit: b = 0.166, p = 0.001) were the only predictors that reached statistical significance in all three specifications, whilst the sociodemographic variables, driving habits, accidents experienced and fines received were not significant. The consistency of results across the three specifications confirms that the study’s findings are robust to the distributional assumptions of the linear model, thereby justifying the use of this approach in the subgroup analyses to be presented subsequently.

Figure 1 shows the breakdown of the variance explained in the regression model. The analysis reveals that the block relating to infrastructure perception represents the most substantial component, explaining 82.3% of the total variance (specifically 10.2 of the 12.4 percentage points explained). Secondly, the block related to the perception of price and user experiences contributes an additional 12.9%. In contrast, the blocks corresponding to sociodemographic characteristics (age, gender, educational level) and mobility habits (frequency of car use, use of toll roads) together explain less than 5% of the variance and are not statistically significant.

This pattern is particularly relevant, as it shows that perceptual factors about the infrastructure (rather than the user’s structural variables) are what really predict satisfaction with toll roads. Although the overall model does not explain a very high proportion of variance, what it does explain is mostly due to the subjective assessment of infrastructure quality. In fact, the user’s experience as a driver, their level of car use, and sociodemographic characteristics do not play a significant role in predicting satisfaction, even though in other studies these factors have been relevant for other dimensions of mobility.

3.3. Hierarchical Regression Model: Users and Non-Users

In order to examine whether the predictors of satisfaction with toll roads differ according to the reported use of these infrastructure facilities, hierarchical regression models were estimated separately for users (n = 807) and non-users (n = 598) (

Table 6. Hierarchical regression model summary by user status: Block-level fit statistics predicting toll road satisfaction.

	Users					Non-Users
Model	R	R2	ΔR2	F Change	Sig	R	R2	ΔR2	F Change	Sig
1	0.114	0.013	0.005	1.533	0.206	0.227	0.051	0.031	2.492	0.063
2	0.117	0.014	−0.001	0.121	0.886	0.261	0.068	0.041	2.475	0.118
3	0.319	0.102	0.084	16.838	<0.001	0.415	0.172	0.135	8.480	<0.001
4	0.353	0.124	0.099	2.897	0.035	0.473	0.224	0.171	2.917	0.037

and

Table 7. Hierarchical regression model coefficients by user status predicting toll road satisfaction.

	Users				Non-Users
Predictors	β	SD	t	Sig.	β	SD	t	Sig.
Step 1
(Constant)	1.375	0.587	2.343	0.020	2.433	0.735	3.311	0.001
Age	0.001	0.005	0.206	0.837	−0.002	0.006	−0.362	0.718
Level of education	0.055	0.065	0.853	0.394	−0.140	0.101	−1.393	0.166
Gender	−0.062	0.143	−0.431	0.667	−0.247	0.206	−1.198	0.233
Step 2
Frequency of car use	0.016	0.026	0.625	0.533	−0.028	0.035	−0.816	0.416
Frequency of toll road use	0.049	0.126	0.389	0.698	-	-	-	-
Step 3
Infrastructure quality perception index	0.179	0.093	1.915	0.056	0.358	0.134	2.678	0.008
Perception of road safety	0.185	0.066	2.809	0.005	0.099	0.099	1.003	0.318
Step 4
Perception of toll prices	0.367	0.125	2.936	0.004	−0.088	0.177	−0.495	0.621
Road crashes	−0.043	0.126	−0.344	0.731	0.275	0.200	1.374	0.172
Penalties	0.009	0.090	0.101	0.920	−0.316	0.115	−2.752	0.007

). The final model explained 12.4% of the variance in user satisfaction (R² = 0.124) and 22.4% in the case of non-users (R² = 0.224).

In both groups, the first block did not explain any significant variance in satisfaction, and no sociodemographic predictor reached statistical significance (all p > 0.16). Similarly, the inclusion of driving habits in the second block did not significantly improve the model’s predictive power. It should be noted that the variable ‘frequency of toll road use’ was not included in the non-users’ model as it was a skip question not applicable to this subgroup.

The substantial change occurred in the third block, with the inclusion of perceptions regarding road infrastructure, although with a different pattern of significant predictors between groups. Among users, this block increased the explained variance by 8.4 percentage points (ΔR² = 0.084, p < 0.001), with perceived safety on Mexican roads being the significant predictor (β = 0.185, p = 0.005). Among non-users, the increase was greater, at 13.5 percentage points (ΔR² = 0.135, p < 0.001), and the pattern was reversed, as the infrastructure quality index was the significant predictor (β = 0.358, p = 0.008), whilst perceived safety did not reach statistical significance (β = 0.099, p = 0.318).

The fourth block added a small but significant improvement in both groups (users: ΔR² = 0.099, p = 0.035; non-users: ΔR² = 0.171, p = 0.037). However, the profile of significant predictors again differed between groups. Among users, perceived toll price was the only significant predictor in this block (β = 0.367, p = 0.004), indicating that the economic valuation of the service is a relevant determinant of satisfaction among those who use it regularly. Among non-users, however, traffic fines received were the significant predictor (β = −0.316, p = 0.007).

3.4. ANOVA Analyses

Table 8. ANOVA Satisfaction with tolls according to perceived price.

					95% Confidence Interval for the Mean
	N	Mean	SD	Standard Error	Lower Limit	Upper Limit	Post Hoc (Tukey)
Expensive	714	3.06	1.324	0.050	2.96	3.16	A
Fair price	384	3.45	1.107	0.056	3.34	3.56	B
Cheap	36	3.58	1.381	0.230	3.12	4.05	AB
Total	1134	3.21	1.271	0.038	3.14	3.28

Note: Groups with different letters differ significantly (p < 0.05). Expensive < Fair price (p < 0.001). Expensive < Cheap (p = 0.040). Fair price vs. Cheap (0.823).

presents an ANOVA confirming significant differences in satisfaction based on perceived toll road prices (F(2,1131) = 13.845, p < 0.001). Post hoc tests (Tukey) revealed that those who perceive tolls as expensive (M = 3.06, SD = 1.32) report significantly lower satisfaction than those who consider them fairly priced (M = 3.45, SD = 1.11; p < 0.001) or cheap (M = 3.58, SD = 1.38; p = 0.040). The difference between fair and cheap was not significant (p = 0.823).

Unlike price perception, no significant differences in satisfaction were found according to toll usage frequency (F(2,1402) = 0.160, p = 0.853;

Table 9. ANOVA Satisfaction according to frequency of toll use.

					95% Confidence Interval for the Mean
	N	Mean	SD	Standard Error	Lower Limit	Upper Limit
Does not use	598	3.17	1.315	0.054	3.07	3.28
Sometimes	498	3.17	1.263	0.057	3.06	3.28
Regularly	309	3.22	1.257	0.072	3.08	3.36
Total	1405	3.18	1.283	0.034	3.12	3.25

). This finding suggests that the use of toll roads is driven more by necessity than by satisfaction. This finding also reinforces the importance of price perception as a tool for managing satisfaction, since, as frequency of use does not translate into greater satisfaction, improving the perception of value for money may be a more effective strategy for increasing user satisfaction than simply modifying toll rates.

4. Discussion

This study examined the determinants of citizen satisfaction with toll roads in Mexico using a hierarchical regression model. The results reveal four main findings: (1) The road infrastructure block is the dominant predictor, as it contributed the greatest increase to the model, accounting for the most substantial contribution to the model’s overall explanatory power (R² = 0.124); (2) sociodemographic characteristics and mobility patterns do not significantly predict satisfaction with these roads; (3) perceived price adds small but significant explanatory power, while previous negative experiences such as traffic fines and road accidents have no effect; and (4) predictors of satisfaction with different toll roads are identified based on user experience.

4.1. Infrastructure Perceived as the Core of Satisfaction

One of the findings of this study is that the general perception of road infrastructure emerges as the factor that most strongly explains satisfaction with toll highways. Specifically, this variable accounts for 82.3% of the explained variance (10.2 out of 12.4 percentage points of R²), above other elements such as perceived price or user characteristics. This suggests that citizens do not evaluate highways as isolated services, but rather within the framework of their daily experience with the entire road system [5,34].

This logic is not unique to Mexico. Various international studies have observed that the perceived quality of infrastructure (understood as a combination of maintenance, signage, fluidity, and safety) is the main basis for building satisfaction with mobility services [35]. In Peru, Arone et al., (2022) [36] found that citizens value roads primarily based on their perceived condition, rather than on personal or sociodemographic variables. In the Dominican Republic, Alonso et al., (2025) [8] showed that the perception of institutional trust and visible maintenance of the network were more relevant determinants than user profile. And in countries such as Ecuador, Paraguay, and Bolivia, reports by the IDB [11] and ECLAC [12] agree that aggregate perceptions of road quality strongly condition the acceptance of toll systems.

This type of overall assessment is based on well-known cognitive mechanisms, such as the halo effect [37]. This phenomenon occurs when a user perceives the road environment as deteriorated, unsafe, or chaotic and tends to project that feeling onto specific elements of the system, even if they are of superior quality. Although this bias was described decades ago [38], recent studies such as that by Conceição et al., (2023) [39] have updated this idea by showing how the emotional experience accumulated during journeys can have a lasting effect on the perception of specific services. In other words, when daily journeys are marked by traffic stress, poor signage, or poor road conditions, that negative experience colors the assessment of the entire system.

This phenomenon is particularly pronounced in contexts such as Mexico, where the structural deficiencies of the road network have been widely documented. According to data from INEGI [40] and CAPUFE [22], federal highways show high rates of deterioration, and chronic problems with signage, maintenance, and safety persist in many regions. It is no coincidence that users internalize this image of precariousness and project it onto toll roads, regardless of their actual condition.

Furthermore, this pattern has also been observed in large Latin American cities such as Bogotá, Lima, and Buenos Aires, where accumulated deficits in road infrastructure have generated collective perceptions of collapse or abandonment. Calatayud et al., (2021) [41] and Brichetti et al., (2021) [42] agree that, in these contexts, the perception of quality is not based on isolated individual experiences, but on the aggregate experience of the system: recurring traffic jams, potholes, lack of signage, or a feeling of permanent risk end up shaping an overall judgment that outweighs any specific attribute [43].

In this scenario, it makes sense that variables traditionally relevant in other studies, such as educational level, gender, or age, have not been significant in our model. While in countries with more developed systems these variables modulate the perception of infrastructure [44,45,46], in contexts where structural deterioration is visible and shared, individual experience can be overshadowed in the face of a collective narrative of dissatisfaction. This does not imply that age or gender do not influence mobility, but rather that their weight is overshadowed when the assessment is strongly determined by the systemic perception of the environment.

From a practical standpoint, these results suggest adopting a more comprehensive view of road satisfaction. Improving a stretch of highway or reducing tolls may have limited effects if the overall perception of the system does not change. In line with this, countries such as Brazil and Chile have implemented citizen audits, transparency schemes, or visible maintenance campaigns to rebuild user confidence in toll concessions [12]. Differentiated pricing strategies and effective public communication have also been explored, as in Santiago and Bogotá, where the perception of fairness in the use of revenues has been key to legitimizing the system [47,48].

Thus, the findings of this study support the idea that satisfaction with toll roads cannot be understood outside the ecosystem in which they operate. Therefore, beyond technical improvements, adopting a user-centred road management model designed to build trust, in which investment in infrastructure is accompanied by measures that are visible and coherent from the public’s perspective, may be a more effective approach [49,50].

4.2. Perceived Price as a Factor of Satisfaction

The second factor with weight in the satisfaction model is price perception. Although its impact is lower than that of infrastructure perception (12.9% of the explained variance), it remains an important variable. However, as some researchers point out, the way citizens evaluate the price of a public service is far from purely economic logic [51,52]. Judgment is not based solely on how much it costs, but on how much it seems to be worth, whether it is considered fair, and whether or not there are other viable options [53].

In this study, 63% of participants considered tolls to be expensive. However, their levels of satisfaction do not differ greatly from those who considered them reasonable. This type of contradiction, far from being anecdotal, appears in other research and is often interpreted as a form of “resigned dissatisfaction” [54,55]. In this regard, users perceive that they are paying too much, but they accept the cost because they have no alternatives or because they identify certain compensatory benefits, such as less congestion, greater safety, or savings in travel time.

In contexts where the road system has structural deficiencies, this resignation may be more widespread. This is shown by studies in Asia and Latin America, where the impact of perceived price depends largely on the existence of alternative routes or the clarity with which the value received in exchange is perceived [54,55]. In Makassar, for example, it was observed that the high acceptance of tolls was due less to satisfaction and more to necessity, as the highway was the only efficient route to key areas of the city [53]. In the case of Mexico, where in many regions toll roads are the only viable option in the face of deteriorated or unsafe arterial networks [22], something similar may be occurring.

Another key aspect is the perception of price fairness. It is not just about how much is paid, but whether that payment feels justified. Various studies [50,53] have shown that when the price is considered reasonable and transparent, positive attitudes toward the service are reinforced, even if it is expensive. Irawan and Alversia, (2024) [51] found that the perception of price fairness not only increases satisfaction but also influences the intention to use the service again. This link between price, fairness, and loyalty is relevant in environments where there are not many alternatives, because the margin of choice is smaller and citizens evaluate not only the service but also the legitimacy of the system that manages it. In this sense, the way in which the pricing policy is communicated, the visibility of maintenance, and confidence in the use of revenues play as important a role as the toll amount itself [54].

It should also be noted that price perception does not act in isolation. It is framed within an emotional and practical system that includes factors such as travel comfort, time saved, and a sense of security. In regions such as Surabaya-Malang (Indonesia), users continue to prefer toll roads precisely because of these benefits, despite recognizing high rates [55]. This type of subjective cost–benefit assessment may also be present in the Mexican case and may explain why the impact of price on satisfaction, although significant, is relatively modest.

4.3. Satisfaction with Tolls and Their Independence from Frequent Use

One of the most striking findings of the study is the lack of a significant association between the frequency of toll road use and satisfaction with them. This finding contradicts classic assumptions such as the familiarity hypothesis [56], which argues that repeated use increases acceptance, and the wear-out hypothesis [57], according to which greater exposure intensifies criticism. In the Mexican case, the data show similar levels of satisfaction among regular users, occasional users, and non-users, suggesting that direct experience is not the main determinant in the evaluation of the service.

One possible explanation lies in the structural nature of use. In contexts where mobility alternatives are limited or less competitive, the use of toll roads is driven by necessity rather than preference. This pattern has been documented in countries such as Indonesia, where the scarcity of free or safe routes forces users to choose toll roads even if they are dissatisfied with the service [53,55,58]. In these cases, frequent use is driven by functional considerations such as time savings or greater safety, and not necessarily by an overall positive assessment.

Furthermore, recent research highlights that satisfaction does not automatically imply loyalty or intention to continue use. Studies such as those by Irawan and Alversia [51] emphasize that variables such as perceived fairness or intention to return mediate the relationship between experience and loyalty. Consequently, judgment of the toll system may be more influenced by collective perceptions than by the specific experiences of each trip [58,59].

However, differences were identified in the predictors of satisfaction depending on users’ experience with this type of infrastructure. Among citizens who are users, perceived road safety in the country is the most significant predictor, whereas among non-users, it is the infrastructure quality index. Furthermore, with regard to the other variables analysed, among users, the perception of the toll price was a significant predictor [60]. These results suggest that users with direct experience of toll roads place greater weight on the subjective perception of road safety, to some extent anchored in specific experiences whilst using these roads, whilst those who do not use them base their assessment primarily on general perceptions of the country’s infrastructure quality, as they lack a direct reference to the characteristics of toll roads [33]. This divergence is consistent with studies that conclude that direct experience modifies the evaluation of a service, introducing specific experiential components that are not present in judgements formed exclusively on the basis of general perceptions [61].

From a practical perspective, this finding invites us to qualify the interpretation of usage as an automatic indicator of satisfaction. Policies aimed at increasing frequency of use through operational improvements may not have a proportional impact on satisfaction if they do not also address structural perception frameworks and conditions that limit user choice. Strategies such as strengthening mobility alternatives, increasing the visibility of the destination of revenue collected, or institutional communication about improvements could contribute more effectively to reinforcing the legitimacy of the system and public confidence.

4.4. Contextual Assessment vs. Isolated Assessment: Practical Implications of the Study

The results suggest the existence of a contextual evaluation model, in which the general perception of the Mexican road system has a greater influence on satisfaction with toll roads than the specific attributes of these roads. Consequently, policies aimed at improving satisfaction must adopt a comprehensive approach that considers both the physical condition of the infrastructure and the social perceptions surrounding it [58,62].

This has strategic implications for concessionaires and authorities. The use of toll roads should not be interpreted as an automatic sign of approval, as it may simply be a response to the lack of viable alternatives [53,55]. Thus, if functional free routes were to emerge, users could abandon them even if objective improvements have been made. Conversely, the deterioration of the free road network can have negative effects on the perception of tolls, even if their quality remains unchanged [48,63]. One specific measure that could be implemented to improve customer service is a channel for reporting issues or making suggestions, which would serve as a mechanism for public participation. Furthermore, the relevant authorities should also take into account informal channels such as social media, as social listening tools could be used to gather information on potential measures to be implemented, as well as to obtain ongoing feedback on public perception of these measures.

However, this interdependence also represents an opportunity. Improving the perception of the road network as a whole through visible investments, road safety campaigns, or effective signage can indirectly benefit satisfaction with toll roads. This logic challenges the competitive approach between free and concessioned sections and proposes a systemic vision in which all components are interrelated.

Depending on the type of measures, they should either be universal in nature or tailored to specific population groups [64]. From a communication perspective, since variables such as age, gender, or educational level do not significantly predict satisfaction, public policies can focus on collective messages that reinforce the shared value of infrastructure [59,65]. However, these messages must be backed up by visible and objective improvements, as a prolonged disconnect between discourse and experience can erode the legitimacy of the system.

However, measures specifically targeted at toll road users could be introduced to justify the toll price (given that it is widely considered expensive and has been identified as a predictor of satisfaction among this group). Specifically, measures such as a loyalty scheme for frequent users or a breakdown of the toll receipt detailing how the fee is spent on maintenance, road safety and other areas could be implemented [66]. And precisely to reinforce and highlight these measures, information boards providing real-time updates on traffic incidents or road conditions could be added as a complementary measure.

In this regard, it is recommended to institutionalize periodic assessments of citizen perception using standardized instruments, which allow for timely adjustment of intervention strategies and detection of emerging sources of dissatisfaction. This combination of continuous monitoring, tangible investment, and effective communication is key to advancing toward a user-centered road governance model.

4.5. Limitations of the Study and Future Lines of Research

This study has some limitations that should be taken into account when interpreting the results. The R² value (12.4%), although moderate, is within the usual range in research on citizen satisfaction and perception of road infrastructure [11,67]. Thus, the model’s limited explanatory power may be due to the presence of factors not captured by the instrument, such as the specific quality of the section of road typically used, the available route alternatives, or the measurement error inherent in self-reports, which should be taken into account in future studies [31,68].

In this type of study, explanatory models tend to have lower coefficients than those focused on objective technical variables, since satisfaction is a subjective judgment influenced by attitudinal factors, prior expectations, and personal experiences [48,59,63]. In fact, other studies have noted that reported satisfaction may reflect individual predispositions more than actual service performance [67]. More relevant than the total magnitude of the explained variance is its internal structure: 82% of the predicted variance comes from the perception of road infrastructure, rather than demographic factors or frequency of use. This finding coincides with research that highlights the centrality of perceived service quality as the main determinant of satisfaction [51,54]. Similarly, other studies show that toll usage responds more to structural constraints (such as lack of alternative routes or time pressure) than to a positive assessment of the service [53,55].

Additionally, the use of simplified ordinal scales to measure price perception and frequency of use may have limited statistical sensitivity. This limitation is methodologically recognized in similar studies, where discrete response scales are used [48], which compromises the accuracy of conventional linear models. As Irawan and Alversia [51] point out, in future research it would be useful to explore more flexible models, such as classification trees, that capture the complexity of citizen judgment.

In terms of future lines of research, we propose expanding the model by incorporating psychosocial and attitudinal variables, such as institutional trust [5], perceptions of fare equity and distributive justice [6], and transparency in concession management [12]. These factors have shown a high explanatory capacity in other contexts and would allow us to move toward a more comprehensive understanding of public satisfaction. The use of comparative designs between countries is also recommended, as suggested by the Mexican Transportation Institute [58] and the IDB [11], in order to identify structural patterns and cultural variations in citizen perception. Another valuable line of work would be the incorporation of qualitative methodologies that allow for the exploration of subjective meanings associated with the road experience and perceptions of justice, safety, or emotional well-being [39]. This is especially relevant if the aim is to link technical performance indicators with indicators of perceived well-being, as already pointed out by CAPUFE, (2024) [22] and the IDB, (2022) [1]. Likewise, future studies could evaluate the impact of specific interventions (such as road safety education campaigns, technological improvements, or maintenance programs) on citizen perceptions [64,67].

5. Conclusions

This study has made it possible to analyze social perceptions of toll roads in Mexico in order to identify relevant aspects in the design of public policies and urban-interurban planning aimed at safer, more efficient, and more legitimate mobility. The results obtained show that citizen satisfaction with toll roads is not explained so much by individual variables (such as demographic profile or frequency of use) but rather by a systemic evaluation of the state and management of the road infrastructure as a whole. Specifically, the results reveal that perceptions of infrastructure quality and perceived safety are the key predictors of satisfaction with toll roads [69]. Furthermore, the pattern of predictors varies depending on usage experience, suggesting that satisfaction is influenced by different factors among users and non-users of these infrastructure facilities.

These findings support a governance perspective in which public perception is not a peripheral factor, but a strategic component of the design and implementation of road policies. As evidenced by studies in comparable contexts (e.g., the case of the points-based driver’s license in Spain, where acceptance of the system has been directly linked to perceptions of regulatory consistency and institutional legitimacy [8]), the incorporation of psychosocial and attitudinal variables is key to the success of structural interventions in mobility.

Although the analysis has focused on the Mexican case, the convergence of results with research in other Latin American and Asian countries suggests that the mechanisms observed could be valid in contexts with similar structural conditions: mixed road networks, high rates of captive use, territorial inequalities, and institutional frameworks in the process of consolidation. From this perspective, the results presented here not only offer empirical evidence on the determinants of satisfaction with toll roads but also provide useful criteria for rethinking the role of the responsible authorities (governmental and concessionaires) in the construction of infrastructure that, in addition to being functional, is perceived as legitimate and fair by citizens.