Assessing Users’ Satisfaction with the Urban Central Metro Station Area in Chengdu: An SEM-IPA Approach

Abstract

An urban central metro station area is a core hub within the high-quality Transit-Oriented Development (TOD) model. This study explores users’ perceptions of built environments around urban central metro stations to investigate the critical determinants of user satisfaction and proposes strategies to enhance the quality of these environments. First, a comprehensive perception system, including location situation, field environment, and urban aesthetics, was developed through literature reviews and expert consultation. Secondly, three typical central metro station areas in Chengdu were selected as study cases, and 425 questionnaires were collected from August to October 2024. The data were analyzed using a structural equation model (SEM) to reveal the impact of built environment perception on overall satisfaction. The results indicate that the field environment has the strongest direct influence on satisfaction. Urban aesthetics impacts satisfaction both directly and indirectly, making its overall effect the most significant. While the location situation does not directly affect satisfaction, it indirectly influences satisfaction through its impact on the field environment and urban aesthetics. Subsequently, based on the satisfaction performance and SEM outcomes, an importance–performance analysis (IPA) was conducted to identify specific areas needing enhancement. Finally, we integrated environmental assessments with the above findings and put forth strategic recommendations to enhance the quality of the built environment.

1. Introduction

A central metro station area (CMSA), as a practical implementation of the Transit-Oriented Development (TOD) model, plays a pivotal role in urban centers and sub-centers. It integrates crucial functions such as transportation, employment, and daily living, positioning itself as a critical public space within the city [1]. TOD has been widely recognized as a sustainable growth strategy, guiding cities towards a more polycentric structure and fostering territorial cohesion [2,3,4]. CMSAs are designed to encourage “metro + walking” travel patterns, where users can complete their remaining activities through short walking or cycling distances after using public transport. This concept aligns with Carlos Moreno’s “15 min city” model, which, when adopted in long-term planning, would improve quality of life by reducing time wasted in traffic and promoting sustainable mobility [5,6].

In practical design, however, due to the complex environment of urban centers and the competition among various elements, the built environment design of a CMSA faces numerous conflicts between economic benefits and user experience [7]. The latest research on public perception of urban space shows that users are more likely to show negative emotions in urban transportation nodes than at other places [8]. These metro station areas located in the center or sub-center of the city bear more social responsibility than other ordinary TOD areas [9], necessitating design considerations that extend beyond mere transportation and economic considerations to encompass public sentiment and perception [10]. The primary way to enhance public satisfaction is to create a people-oriented, efficient, and comfortable pedestrian environment [11]. Consequently, a comprehensive understanding of public perceptions of central metro station areas is imperative for formulating socially oriented design strategies and policies to enhance overall urban livability [7].

The station catchment area represents an intersection of transportation planning and urban design. Currently, traffic researchers focus primarily on the impact of macro-level indicators on passenger travel and the complex interrelationships among these variables [12,13]. In contrast, urban design researchers concentrate on universal methodologies to enhance the satisfaction of urban public spaces [14,15]. However, there exists a significant gap in the literature regarding the in-depth exploration of urban spatial perception within the context of traffic characteristics. The primary reason for this gap lies in the fact that traffic research typically focuses on macro-level indicators such as traffic flow and travel patterns, while urban design research tends to emphasize spatial form and aesthetic experience. The lack of an integrated interdisciplinary perspective has resulted in insufficient exploration of how traffic characteristics influence urban spatial perception. Thus, establishing a user satisfaction research framework that combines the dual attributes of urban public spaces and traffic node functions in central metro station areas is essential. Specifically, the evaluation of transit catchment areas should include the integrated development characteristics of TOD, such as the 5D principles and node place attributes, while also incorporating aspects like urban aesthetics and service experience. This is necessary for shaping dual benefits that balance both urban economic development and a human-centered society [10,16].

This study delves into the determinants of user satisfaction with the built environment of a CMSA through the application of a combined approach involving the structural equation modeling (SEM) and importance–performance analysis (IPA) methodologies [17]. By assessing the comparative impact intensity and satisfaction levels of these factors, this study aims to pinpoint areas requiring enhancement. First, we devised a conceptual framework for the CMSA that includes location situation, field environment, and urban aesthetics. Subsequently, a selection of observed variables was curated through word frequency analysis and expert consultation, culminating in the construction of a second-order structural equation model. Drawing upon the SEM analysis of 425 questionnaires collected from three typical CMSAs in Chengdu, our investigation elucidates the intricate influence mechanisms of location situation, density, diversity, distance to transit, destination accessibility, road networks, connection design, street characteristics, and order management on user satisfaction. Then, we assessed the alignment between the significance and performance of objective environmental elements by IPA. Finally, tailored strategic recommendations were formulated for the CMSA by synthesizing these findings with on-site investigations. This study offers a dual contribution: (1) It develops a perception assessment framework for the CMSA, which aids in evaluating user satisfaction and analyzing potential correlations between the perception of individual environmental elements and public sentiment. (2) The findings provide various policy recommendations for urban designers and TOD developers to enhance CMSA design, ultimately improving user satisfaction and promoting human-centered development.

The contents are organized as follows: The second section describes the theoretical framework, hypothesis, and CMSA perception index system. The third section details the research method, encompassing the methodology of SEM-IPA, study areas, data sources, and research process. The fourth section presents the empirical study results, focusing on survey data characteristics, model testing and fitness, path impact outcomes, and the alignment identified by importance–performance. In the last section, some design lessons and key points are summarized.

2. Theoretical Framework and Index

2.1. Theoretical Framework and Hypothesis

A conceptual framework is constructed based on the interpretation of the content of the catchment area of a metro station by existing academic theory. The 3D theory (and its extensions) and node–place model are the academic theories commonly used to characterize the built environment of a catchment area. The mesoscale factors of the built environment are gauged within an area and include land use diversity, street designs, and densities. These factors are designated as the 3Ds; later, these 3Ds were expanded to 5Ds by the addition of two more factors: distance to transit and destination accessibility [18]. Banerjee [19] proposed the concept of ‘railscape’, emphasizing the role of streetscape aesthetics, historical imagery, and user perception in the built environment surrounding railways. The node–place model introduced by Bertolini [20] underscores the necessity to characterize the catchment area by considering both macroscopic traffic performance and the mesoscale place design. The descriptions of the built environment by the NP model and “D” theory overlap. Cao et al. [21] took the 5Ds as the second-level index of the NP model. Deng et al. [22] combined the 5Ds and NP model to evaluate the catchment area. They identified density, diversity, and design as indicators of place design and added a ‘linkage’ dimension to represent the distance to transit and destination accessibility.

By the explanation of the built environment of the catchment area in different scales, we intended to deconstruct the theoretical connotation of the built environment into the location situation, field environment, and urban aesthetics. Figure 1 is a diagrammatic representation of built environment quality attributes for user satisfaction as a theoretical framework for this study, which is also the basis for the structural equation model. There are three first-order latent variables and five second-order latent variables.

Hypotheses regarding the direct and indirect effects among latent constructs may be verified after an SEM analysis has been conducted.

Table 1. Theoretical hypothesis.

Hypothesis	Description	Reference
H1	Field environment directly affects satisfaction	[14,23,24]
H2	Location situation directly affects satisfaction	[25,26]
H3	Urban aesthetics directly affects satisfaction	[14,27,28]
H4	Location situation affects field environment	[29]
H5	Location situation affects urban aesthetics	[30,31]
H6	Urban aesthetics affects field environment	[32,33]

shows six hypotheses between latent variables and satisfaction and between latent variables.

2.2. Index System

Based on the conceptual framework, the corresponding observed variables were selected through word frequency analysis and expert consultation to form a measurable model. We selected the indicators through word frequency analysis and expert consultation. As one of the bibliometric methods, the word frequency analysis is a research method that uses quantitative statistical analysis to describe the external characteristics of publications and evaluate the critical spots in a particular research field [34]. Expert consultation is seen as a consensus-oriented technique where experts orient themselves towards a greater or lesser degree of consensus around the critical projection or visions of the future within a specific field [35].

We searched for relevant studies that met two conditions: first, a title containing the words “Station Area”, “Transit Oriented”, “TOD”, “Transit Station”, and so on; second, an abstract or keywords containing words such as “performance”, “influence”, “impact”, “effect”, “quality”, “evaluate”, “assess”, or “measure”. In total, 158 studies from 2019 to 2023 were collected, and 94 indicators were selected. According to the frequency of these words, we divided them into high, mid, and low frequencies. All the high-frequency indicators were adopted, most mid-frequency indicators were retained, the indicators that did not meet the research theme were deleted, and only a few of the low-frequency indicators were retained. Finally, 58 primary indicators were obtained.

We consulted 13 experts on these 58 indicators and asked them to rate the impact of each indicator on user perception on a scale of 1–5. The experts were from university scholars, designers, TOD developers, and metro companies, and 12 experts responded—a positive degree of 93.3%. The index was further selected by calculating and comparing the average value and coefficient of variation of each indicator. The average value reflects the influence of the indicator on the perception, and the coefficient of variation reflects the degree of expert consensus. The calculation results are shown in Figure 2. Indicators with low average values and a high coefficient of variation were deleted.

Table 2. Perception index of the built environment of the CMSA.

Dimension	1st-Order Index	2nd-Order Index
Location Situation	/	L1. Location in the city
		L2. The rationality of subway line planning
		L3. Number of subway lines
		L4. Accessibility to other centers
Field Environment	Density	A1. Development intensity
		A2. Building volume
		A3. Underground space development intensity
		A4. Adequacy of open space
		A5. Crowdedness
		A6. Traffic jam
		A7. The interference between people and vehicles
	Diversity	B1. Satisfaction of basic needs
		B2. Adequacy of function selection
		B3. The rationality of function ratio
	Destination accessibility	C1. Accessibility of commercial facilities
		C2. Accessibility of official facilities
		C3. Accessibility of residential facilities
		C4. Accessibility of public service facilities
	Distance to transit	D1. The convenience of the bus transfer
		D2. The convenience of the bicycle transfer
	Design	E1. Road connectivity
		E2. Road directivity
		E3. The convenience of getting to the destination
		E4. Entrance quantity
		E5. Entrance location
		E6. Degree of difficulty of finding entrance
		E7. Form of entrance
		E8. Entrance connection with building
Urban Aesthetics	/	U1. Proportion and scale of the street
		U2. Color and style of the street
		U3. Street furniture
		U4. Culture of the street
		U5. Street hygiene
		U6. Landscape maintenance
		U7. Perceived safety

shows the index system after preliminary selection. We designed the questionnaire based on the indicator system. Specifically, for each indicator in the table, respondents were asked to provide their perceptions, which were quantified using a Likert scale. A score of 5 indicated “very satisfied”, 4 indicated “somewhat satisfied”, 3 indicated “neutral”, 2 indicated “somewhat dissatisfied”, and 1 indicated “very dissatisfied”.

3. Method and Data

3.1. SEM-IPA Model

3.1.1. Structural Equation Model (SEM)

Structural equation modeling (SEM) is a statistical method for analyzing variables based on their covariance matrix to find their inherent structural relationships. Users provided judgments for each aspect, allowing them to be evaluated as observed endogenous latent variables in the model [36]. A complete SEM includes structural and measurement models. Of these, the former reflects the relationship between latent variables, which is expressed as

$$\mathsf{\eta }=\mathsf{\beta }\mathsf{\eta }+\mathsf{\Gamma }\mathsf{\xi }+\mathsf{\xi },$$

(1)

where η denotes the endogenous latent variable, β is the influence relationship between the endogenous latent variables, Γ is the influence relationship between the exogenous latent variable and the endogenous latent variable, and ξ is the residual term of the endogenous latent variable.

The measurement model reflects the relationship between latent variables and explicit variables, which can be expressed as

$$\begin{gathered} \mathrm{X}=\mathsf{\Lambda }\mathrm{x}\times \mathsf{\xi }+\mathsf{\delta }, \\ \mathrm{Y}=\mathsf{\Lambda }\mathrm{y}\times \mathsf{\eta }+\mathsf{\epsilon }, \end{gathered}$$

(2)

where X is the exogenous explicit variable, Y is the endogenous explicit variable, Λx is the relationship matrix between the exogenous explicit variable X and the exogenous latent variable ξ, Λy is the relationship between the endogenous explicit variable Y and the endogenous latent variable η, and δ and ε are X and Y error terms.

3.1.2. Importance–Performance Analysis (IPA)

Importance–performance analysis (IPA) is a method for identifying the alignment of importance and performance and assumes a distribution of a set of attributes in four sets, according to the four possible combinations, with respect to the performance and importance of attributes [37,38]. Quadrant I is the area of “high importance and high performance”. Quadrant II is the area of “high importance and low performance”. Quadrant III is the area of “low importance and low performance”. Quadrant IV is the area of “low importance and high performance”. The recommendable actions given by the IPA theory for the four areas of appeal are “Keep up the good work”, “Concentrate here”, “Low priority”, and “Possible overkill” [39].

3.1.3. SEM-IPA Method

The SEM-IPA method is also presented as an importance–performance matrix (Figure 3), and the quantification of importance is based on the results of SEM. We calculate the importance of these variables based on their total contribution to satisfaction, as the greater the contribution of a factor to satisfaction, the higher its importance. For both first-order and second-order latent variables, their contribution to satisfaction is the sum of their direct and indirect effects, i.e., the total effect on satisfaction. For observed variables, their importance is determined by the product of their factor loadings and total effects. Similarly, the performance value is defined by the satisfaction score; the higher the satisfaction, the better the perceived performance.

In the IPA results, the alignment of importance and performance is essential for the evaluation of user perception. Specifically, a higher perception of elements should correspond to increased satisfaction, indicating that users have had a positive experience with these environmental factors. Based on the assessment of this alignment, we categorized the quality attributes in the four quadrants into two groups: matched (Quadrants I and III) and unmatched (Quadrants II and IV). When importance and performance align, they can be either high or low; in both scenarios, the strategy is to maintain the current state. Conversely, when there is a mismatch, two situations arise: in Quadrant II, the importance of factors significantly exceeds their performance, necessitating additional investment for improvement. In Quadrant IV, the importance of elements is considerably lower than their performance, indicating a need to reduce input.

3.2. Study Area and Data

Based on the “Guidelines for Integrated Urban Design of Chengdu Metro Stations”, the differences among several city-level central metro stations are primarily reflected in two aspects: location and dominant function. Location differences refer to whether the stations are situated in the new city or the old city. Functional differences are seen in whether the stations are dominated by historical landmarks, comprehensive commercial, business, and administration areas, or large transit hubs. Due to their unique characteristics in terms of function, form, and clientele, stations dominated by large transit hubs have been excluded from this study. The selection of the study samples was based on three principles: first, including the representation of different locations; second, covering stations with different dominant functional types; and third, including stations with a relatively high level of development. Three stations were chosen: Tianfu Square, 3rd Tianfu Street, and Chunxi Road. Tianfu Square, located in the old city center, primarily serves administrative and historical functions; Chunxi Road is a commercial hub; and 3rd Tianfu Street is predominantly a business and office district. These three stations are well-established urban centers, each with distinct functional characteristics (Figure 4).

The scale range of the catchment area is defined by the effective perception range of the user in actual use. We simulated the 5 min walk at the outermost entrance and exit of the site through Mapbox, combined with the 5 min range and the shape of the plot, and manually outlined the effective perception range. The overall size is about 400–800 m, which is similar to previous research [41,42].

Based on the theoretical framework, this study utilized a questionnaire consisting of self-formulated questions of the index system. The questionnaire comprised two sections: personal information and a satisfaction survey. The personal information section gathered data on participants’ gender, age, educational background, occupation, purpose of visit, and frequency of visits. The satisfaction survey required participants to evaluate various components of the catchment area using a standardized rating scale ranging from 1 (dissatisfaction) to 5 (satisfaction). Questionnaires were distributed near different entrances of the three stations. This approach ensured a diverse sampling of individuals using these stations, thereby enhancing the representativeness and reliability of the collected information.

3.3. Research Process

We developed the following framework (Figure 5): The initial phase established the research foundation by conducting a literature review to construct a theoretical framework. This phase also involved administering a satisfaction questionnaire to gather fundamental study data. The theoretical framework was developed based on the built environment of the transport area theory, with formulated study hypotheses. The framework was refined through word frequency analysis and expert consultation into a specific and measurable perceptual evaluation index system supported by data from a questionnaire survey.

In the subsequent phase, SEM was employed to depict the interplay among different dimensions of the perception of the catchment area. Concurrently, the IPA method was utilized to evaluate the alignment between the importance and performance of individual spatial elements. A two-order SEM model was constructed based on the theoretical framework and questionnaire data, adjusted iteratively to test the validity of the proposed hypotheses. The SEM results unveiled each factor’s influence mechanisms and intensities in relation to satisfaction, providing an importance value for the IPA analysis. IPA analysis categorized indicators based on perceived importance and performance, distinguishing matched and unmatched types.

In the final phase, the research findings were carefully examined, culminating in the development of strategic recommendations. Drawing from the SEM-IPA analysis, additional insights into the perceptual attributes of elements are presented. These insights, combined with on-site structural assessments, led to the identification of deficient built environment elements and the provision of targeted recommendations. Furthermore, some critical environmental features conducive to enhancing satisfaction in the central catchment area were summarized.

4. Results

4.1. Data Description

In total, 425 valid questionnaires were collected: 167 at Tianfu Square Station, 138 at Chunxi Road Station, and 120 at 3rd Tianfu Street Station.

Table 3. Respondents’ characteristics.

Personal Particulars	Item	Percent (%)	Personal Particulars	Item	Percent (%)
Gender	Male	47.29	Purpose of visit	Transfer	1.65
	Female	52.71		Shopping	15.06
Age	<30	50.82		Work	33.65
	≥30	49.18		Residence	7.06
Education	Sub-bachelor’s	56.94		Affairs	32.94
	Bachelor’s degree or above	43.06		Travel	5.42
Professional relevance	Unrelated	74.82		Others	4.22
	Relevant	25.18
Visiting frequency	<2 times/month	43.53
	3–4 times/month	33.88
	Almost every day	22.59

presents a statistical overview of the respondents’ information, revealing the following key characteristics: (1) The gender distribution of the respondents was nearly balanced, with an average age of 28.6 years. The education level was primarily dominated by bachelor’s degrees. (2) Most respondents have low knowledge of TOD and urban design. (3) The primary purposes for visiting the stations were work, official business, shopping, and entertainment, with a low frequency.

4.2. Reliability and Validity

Using SPSS 25.0, we evaluated the reliability and validity of the questionnaire data to assess their rationality. Reliability is the evaluation of the stability and consistency of the results. Validity is used to assess the consistency between predicted and actual measured data results.

We adopted Cronbach’s alpha (α) to evaluate the reliability of the scale. It is generally believed that alpha should be greater than 0.6 to indicate that the internal consistency of the questionnaire is accepted [43]. The α for the destination accessibility dimension is 0.596. Upon examining the α coefficient after item deletion, it was found that the α coefficient for C3 (accessibility of residential facilities) increased to 0.720. Consequently, the C3 item was removed. The overall α coefficient is 0.871, with the coefficients for each dimension ranging from 0.720 to 0.875 (

Table 4. Reliability analysis of latent variables.

Latent Variable	Cronbach’s Alpha	Latent Variable	Cronbach’s Alpha
Location situation	0.825	Distance to transit	0.835
Density	0.872	Design	0.763
Diversity	0.875	Urban aesthetics	0.801
Destination accessibility	0.720	Satisfaction	0.858

), indicating good internal consistency. The results showed that KMO was greater than 0.8 [44] (

Table 5. Results of KMO test and the Bartlett’s test of sphericity for the questionnaires.

KMO Measure of Sampling Adequacy		0.847
Bartlett Test of Sphericity	Approximate Chi-Square	6977.882
	df	561
	Sig.	0.000

), indicating the data were suitable for factor analysis to test the validity and the variables had good independence.

Exploratory factor analysis strives to uncover the coherence between the variables and the primary component factors extracted from empirical data. It can also conduct mathematical tests on the previously constructed observed variables. With the principal component extraction method and the maximum variance rotation, nine factors with eigenvalues greater than 1.0 were extracted, and the cumulative explanatory variance was 69.014%, which is above 50% and is acceptable [45] (

Table 6. Total variance explained based on the field questionnaire.

Factors	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	7.435	21.867	21.867	7.435	21.867	21.867	4.207	12.372	12.372
2	3.586	10.547	32.414	3.586	10.547	32.414	3.056	8.987	21.359
3	2.503	7.362	39.776	2.503	7.362	39.776	2.983	8.773	30.133
4	2.145	6.31	46.086	2.145	6.31	46.086	2.679	7.88	38.013
5	1.826	5.371	51.457	1.826	5.371	51.457	2.447	7.197	45.21
6	1.753	5.157	56.614	1.753	5.157	56.614	2.259	6.644	51.853
7	1.612	4.741	61.355	1.612	4.741	61.355	2.069	6.086	57.94
8	1.338	3.937	65.291	1.338	3.937	65.291	1.965	5.778	63.718
9	1.266	3.723	69.014	1.266	3.723	69.014	1.801	5.297	69.014

). Therefore, we adjusted the original index system: The first-order latent variable named design dimension is refined into road network design and connection design, and the first-order latent variable named urban aesthetics is refined into street characteristics and order management (

Table 7. Factor analysis.

	Component	Factor Loading
1	Density	A1/0.576	A2/0.715	A3/0.710	A4/0.669
		A5/0.761	A6/0.806	A7/0.728
2	Connection design	E4/0.797	E5/0.806	E7/0.857	E8/0.830
3	Location situation	L1/0.783	L2/0.796	L3/0.830	L4/0.795
4	Urban characteristics	U1/0.823	U2/0.859	U3/0.837	U4/0.842
5	Diversity	B1/0.762	B2/0.801	B3/0.814
6	Order management	U5/0.845	U6/0.830	U7/0.737
7	Road network design	E1/0.859	E2/0.879	E3/0.838
8	Destination accessibility	C1/0.804	C2/0.732	C4/0.781
9	Distance to transit	D1/0.871	D2/0.859

). After principal component analysis, all factors loading above 0.4 can be retained [46]. Among the observed variables, the standard factor load of E6 (difficulty of finding entrance) is lower than 0.4, so the item is deleted.

4.3. Influence Path

We used Amos 28 software to build and compute the structural equation model, establishing a model that includes four first-order latent variables and eight second-order latent variables, and imported the questionnaire data into the model. The model (Figure 6) demonstrated that location situation (H2, p-value > 0.05, C.R. < 1.96) had no significant effect on satisfaction; thus, these exogenous latent variables were removed. This was represented by the broken lines. The remaining five hypothetical paths are all supported (

Table 8. Model path coefficients and hypothesis results.

Hypothesis	Path	C.R.	Result
H1	Field environment—>Satisfaction	4.789 ***	True
H2	Location situation—>Satisfaction	0.069	False
H3	Urban aesthetics—>Satisfaction	3.242 **	True
H4	Location situation—>Field environment	3.006 **	True
H5	Location situation—>Urban aesthetics	−3.151 **	True
H6	Urban aesthetics—>Field environment	2.932 **	True

* denotes a p-value less than 0.05, ** denotes a p-value less than 0.01, and *** denotes a p-value less than 0.001.

). In addition, items with values > 0.50 were deemed significant [47]. In the current model, all items exceed this threshold. The SEM was rerun after the removal of the exogenous latent variable and the consideration of the parameters to create the final model in evaluating the effect on satisfaction.

Table 9. Fitness of the SEM.

Goodness of Fit Measures	IFI	CFI	GFI	AGFI	x2/df	RMSEA
Parameter estimates	0.949	0.948	0.890	0.874	1.654	0.039
Minimum cut-off	>0.80	>0.80	>0.80	>0.80	1~3	<0.07

demonstrates the IFI, CFI, GFI, AGFI, and RMSEA of the adjusted model. By the standard [48,49], values higher than 0.800 are cut off. The value of x²/df ranges from 1 to 3, and RMSEA < 0.07 is acceptable, which provides the reliability of a good representation of the data observed.

Table 10. AVE and CR scores.

Observed Variable	Latent Variable	Estimate	AVE	CR	Observed Variable	Latent Variable	Estimate	AVE	CR
L4	Location Situation	0.769	0.542	0.827	D2	Distance to transit	0.814	0.719	0.837
L3		0.736			D1		0.881
L2		0.765			E3	Road network design	0.763	0.698	0.873
L1		0.677			E2		0.847
A7	Density	0.667	0.485	0.874	E1		0.891
A6		0.758			E7	Connection design	0.781	0.670	0.89
A5		0.726			E6		0.865
A4		0.605			E5		0.788
A3		0.709			E4		0.836
A2		0.769			U1	Urban characteristics	0.78	0.643	0.878
A1		0.695			U2		0.83
B3	Diversity	0.84	0.701	0.876	U3		0.796
B2		0.789			U4		0.8
B1		0.881			U5	Order management	0.804	0.532	0.769
C3	Destination accessibility	0.758	0.470	0.724	U6		0.791
C2		0.576			U7		0.569
C1		0.709

presents the AVE and CR values of the adjusted model. AVE reflects the explanatory power of latent variables on their respective indicators, while CR is used to assess the reliability of latent variables. The majority of the latent variables exhibit an AVE greater than 0.5. According to Hair et al. [50], factor loadings greater than 0.5 and CR values above 0.70 are typically considered ideal. In this study, a few dimensions have AVE values slightly below 0.5, which can be considered acceptable as they demonstrate moderate explanatory power.

The final model (Figure 7 and

Table 11. Influence path analysis.

Independent Variable	Dependent Variable	Direct Effect	Indirect Effect	Total Effect
Field environment	Satisfaction	0.399	/	0.399
Location situation		/	−0.023	−0.023
Urban aesthetics		0.336	0.126	0.462
Location situation	Field environment	0.206	−0.072	0.134
Location situation	Urban aesthetics	−0.228	/	0.228
Urban aesthetics	Field environment	0.316	/	0.316

) shows the impact path results as follows: (1) Field environment perception has the most significant direct perceptual effect on station satisfaction, and it is also affected by urban aesthetics and location perception. (2) Urban aesthetic perception exerts a significant direct influence on satisfaction and also serves as a mediator by positively impacting the field environment. Consequently, urban aesthetic perception demonstrates the most substantial total effect on satisfaction. (3) The direct influence of location perception on satisfaction is relatively modest. However, it impacts the field environment and urban aesthetic perception, indirectly affecting satisfaction. The positive correlation between location conditions and the field environment suggests that superior location conditions enhance users’ inclusiveness towards the field environment. Conversely, the negative relationship between location conditions and urban aesthetics implies that a more centralized site elevates public expectations regarding its portal effect.

4.4. Importance–Performance Alignment

Figure 8 and

Table 12. The distribution of elements in the IPA quadrants.

Station	First Quadrant	Second Quadrant	Third Quadrant	Fourth Quadrant
Tianfu Square	A1, A2, A3, A4, A5, A6, B1, B2, B3, E4, E6, E7, U1, U4, U6	A7, E5, U2, U3, U5	D1, D2	L1, L2, L3, L4, C1, C2, C3, E1, E2, E3, U7
Chunxi Road	A1, A2, A3, A4, A5, A6, B1, B2, B3, E4, E6, E7, U2, U3, U4	A7, E5, U1, U5, U6	L3, C1, C2, D1, D2, E1, E2, E3	L1, L2, L4, C3
3rd Tianfu Street	/	A1, A2, A3, A4, A5, A6, A7, B1, B2, B3, E4, E5, E6, E7, U1, U2, U3, U4, U5, U6	L1, L2, L3, L4, C1, C2, C3, D1, E1, E2, E3	D2, D7

displays the IPA results. The values of the contribution of each element to perception and the satisfaction scores represent the importance and performance, respectively. Using the average values of all elements’ importance and performance as the thresholds, two dividing lines are drawn vertically and horizontally on the coordinates, categorizing the elements into four quadrants.

Quadrant I: The elements located in this area exhibit both high importance and high performance, indicating a well-aligned relationship between their perception influence and satisfaction situation. These elements represent significant strengths and are the primary contributors to users’ positive experiences within the catchment area. Notably, elements in this quadrant are predominantly found in Chunxi Road and Tianfu Square, accounting for 48.3% and 51.7%, respectively.

Quadrant II: The elements in quadrant II exhibit a stark mismatch between their high importance and their performance, which is notably low. These elements significantly influence user satisfaction but fail to receive the corresponding user praise, leading to a diminished user experience. Elements in this quadrant are predominantly found in 3rd Tianfu Street, accounting for 66.7%. Tianfu Square and Chunxi Road both accounted for 16.7%.

Quadrant III: The elements located in this area exhibit both low importance and low performance, indicating a matching relationship between their perceived influence and satisfaction levels. These elements have a minimal impact on user satisfaction, and their evaluations by users are correspondingly low. Elements in this quadrant are present across all three stations, with proportions of 50.0% at 3rd Tianfu Street, 40.1% at Chunxi Road, and 10.0% at Tianfu Square.

Quadrant IV: The elements located in this quadrant exhibit a mismatch between their importance and performance, with performance significantly exceeding importance. These elements have a minimal impact on user satisfaction, yet they receive disproportionately positive feedback from users. Elements in this quadrant are present across all three stations, with proportions of 64.7% at Tianfu Square, 23.5% at Chunxi Road, and 11.8% at 3rd Tianfu Street.

Figure 9 utilizes radar charts to illustrate the standardized comparison of the importance and performance of different level variables in each station area. These charts facilitate the easy identification of defect elements. An image is formed by connecting the importance and performance values of each area. The higher the resemblance in shape and area between the importance and performance representations, the greater the matching degree of the catchment area. Conversely, notable disparities in shape indicate potential design flaws within the catchment area. A performance graph area surpassing that of importance signifies inadequacy in the station area, prompting a need to “Concentrate here”, while an importance graph area exceeding that of performance suggests a potential “Possible overkill” in the area. Below, we elaborate the presentation results of latent variables and observed variables in the three CMSAs.

The radar charts of the first- and second-order latent variables and observed variables illustrate the overall and detailed situations of all station areas. The results indicate that the performance graphs of Tianfu Square and Chunxi Road stations both exceed their importance graphs, suggesting that overall input should be maintained or reduced. Conversely, the performance graphs of 3rd Tianfu Street are significantly lower than the importance graphs, indicating a need for further optimization and improvement. In the first-order latent variables, field environment and urban aesthetics are identified as variables with high perceptual influence. Correspondingly, in the second-order latent variables, density, diversity, connection design, street characteristics, and order management are recognized as having significant perceptual effects. At Tianfu Square station, these variables fall into the “high importance–high performance” category, suggesting that maintaining the status quo is appropriate. At Chunxi Road station, the first-order variable of urban aesthetics falls into the “high importance–low performance” unmatched category. A detailed analysis reveals that while street characteristics at Chunxi Road are excellent, the evaluation of order management is poor and requires improvement. In the 3rd Tianfu Street station area, two important first-order variables also fall into the “high importance–low performance” mismatch category. Among the second-order variables, only order management performs well, while the others perform poorly and need significant improvement.

The radar chart of observed variables can highlight specific environmental elements that need optimization. For example, at Chunxi Road station, the order management dimension includes U5 (street hygiene) and U6 (landscape maintenance), which are high-impact elements but exhibit low performance, categorizing them as “high importance–low performance” elements. There are more than two-thirds of such elements in the 3rd Tianfu Street station area. Improving these built elements will result in a noticeable enhancement in evaluations.

5. Discussion and Conclusions

5.1. Discussion

5.1.1. Field Environment

The field environment has the most significant direct effect on the satisfaction of the CMSA. Here, density, diversity, and connection design are the most important second-order latent variables. The density and diversity perception are related to the land use, while the connection design is the pedestrian planning and design with the metro entrance as the core. We discuss some of the key design points of land use and pedestrian design.

We listed a part of land use indicators for the three cases (Figure 10) and found that the apparent differences in density perception among the three stations are building height and open space. We infer that building height and open space can directly affect the user’s perception of density. Another noticeable difference is the proportion of commerce. 3rd Tianfu Street needs a higher proportion of commercial function, which may reduce the diversity of users’ activity choices. Based on the above analysis, we recommend the primary strategy to enhance land use satisfaction is to dispel the feeling of high-density development and increase the number of business-oriented functional types. High-density development does help improve land use efficiency and promote urban vitality; it is also considered a critical factor in the successful implementation of TOD, but a common problem central-type sites face is the poor sense of repression and congestion it brings [51]. One of the strategies to solve this problem is to appropriately add small-scale buildings in the station area so that the buildings have a difference of height and height. This strategy is conducive to creating a sparse and staggered urban spatial pattern in old cities and maximizing the agglomeration effect of land development [52]. Another suggestion is to increase the development area of the underground space and release the ground space, thus providing citizens with more open spaces like plazas and parks. Open space can also reduce the degree of interference between people and vehicles and enhance pedestrian safety and comfort.

In terms of diversity, a study showed that neighborhoods with a relatively higher land-use mix have a significantly positive association with walking activity [53]. In the planning concept of the “15 min city”, residents should be able to enjoy a higher quality of life within walking or cycling distance, where they can effectively fulfill six essential urban functions to sustain a decent urban life. These functions include living, working, commerce, healthcare, education, and entertainment [6]. These functions are integrated into TOD planning, expanding outward in concentric circles in decreasing order of economic value and development intensity, to enhance land-use diversity and the completeness of urban functions. Moreover, to maximize these benefits, diversity needs to be implemented at various scales within the city—not only at the urban level, across broader areas, but also at finer scales, such as at the building level.

A suggestion about pedestrian design is to strengthen the walking continuity and enrich the entrance’s form. The planning and design of a station play a significant role in the construction of the station area walking system [54]. For example, the distance to the entrance of Tianfu Square station is more than 380 m, crossing several main roads to achieve passenger–vehicle separation and providing a comfortable underground walking space in extreme weather. Its entrance and exit of Tianfu Square take the form of a sunken square (Figure 11a) to organize passenger flow into the subway station and the commercial complex built on the station. Chunxi Road features a metro entrance that connects directly to the underground of the commercial center (Figure 11b). These designs not only facilitate the management of pedestrian traffic and enhance the integration of the station with the city but also help regulate the microclimate at the metro station entrance [55]. The form of the subway entrance is also an essential factor. Subway stations are usually built underground without building facades, and entrances are vital nodes for users to recognize and remember environmental information. Li’s finding verifies that the traffic demand is the priority in the landscape design of the metro station entrance [56]. The entrances and exits of Chunxi Road adopt various forms of glass to give users a novel experience (Figure 11c). In conclusion, the planning of station entrances and exits should adopt a field-oriented approach, enhancing crowd diversion efficiency by integrating urban main roads and linking key areas and buildings. Additionally, the design of entrances and exits should be both memorable and aesthetically pleasing, while also reflecting the cultural characteristics of the city.

5.1.2. Urban Aesthetics

Urban aesthetics had the strongest influence on satisfaction because of its direct and indirect effects. Here, street characteristics are the dominant second-order variable. Color and culture are two important elements that shape street characteristics. In the cases we studied, we found that 3rd Tianfu Street Station is located in the city’s new district, which was completed in a short time, and the government uniformly planned the facade of the building. Therefore, it may lack characteristics compared to Tianfu Square, which has historical charm, and Chunxi Road, which has diversified colorful characteristics (Figure 12). The use of culture as a tool for urban regeneration is not recent; it has been used as a preferred concept to direct the revitalization of urban centers [57]. The color and form of architecture are the design methods that reflect a city’s culture. It is of contemporary significance to retain regional characteristics, stimulate emotional resonance, and enhance the readability of the city image by portraying the urban ethnic image.

Order management, as another second-order variable in urban aesthetics, is not more important than the average in our conclusion, but a variety of studies have shown that a clean, orderly, and safe environment is an essential condition of urban aesthetics and an important guarantee for citizens’ quality of life [58]. This means that order management is also an indispensable part of urban aesthetics. Street hygiene and the pruning of plants are two significant factors. Keeping the streets clean requires adding garbage disposal facilities and regularly cleaning the ground. Vegetation is another fundamental element of urban design. Trees provide the shade necessary to walk and stay in the streets during the day’s hottest hours. Also, vegetation has aesthetic qualities that are indispensable for the design of attractive and pleasing places [59]. The integration of the Smart City concept is also a key factor in improving management levels. For example, within the Smart City framework, factors such as inclusivity, resident participation, and real-time service delivery are promoted through various digital platforms [6].

5.1.3. Location Situation

The location situation does not have a direct impact on satisfaction, but it is not irrelevant. Instead, it indirectly influences satisfaction by affecting the field environment and urban aesthetics. The centers of new urban areas compete with the old urban centers, which tend to have better absolute geographical locations in the process of urban expansion [13]. It is worth noting that there is a negative correlation between location situation and urban aesthetics. For station areas with superior locations, which are more likely to represent the cultural and economic hubs of a city, people tend to have stricter evaluations of their urban aesthetics. From a psychological perspective, there is a tension between functionality and aesthetics. People typically expect high-density areas to not only fulfill basic functional needs but also provide a good visual and environmental experience.

Urban location is a multidimensional performance that is affected by natural and artificial urban areas [60]. Therefore, for stations located in less favorable areas, particularly in new districts, we recommend enhancing the relative centrality of the station area by increasing the number of subway lines at the central station of the new city and ensuring these lines traverse densely populated areas. These two strategies are also key components of our model findings. Studies indicated that increasing the number of subway lines at metro stations and enhancing transportation options to other central areas can improve urban compactness and promote more efficient living for residents [61,62]. There are many examples of urban planning in Japan that we can learn from, the construction goal of which is a “network compact city” [63].

It is worth noting that the relationship between the location situation and field environment is positively correlated, indicating that when the location situation is favorable, people tend to have a higher tolerance for the perceived aspects of the field environment. For instance, when a site is situated in a high-density development area, individuals are more accepting of crowded buildings and heavy pedestrian traffic. Conversely, the relationship between the location situation and urban aesthetics is negatively correlated. This can be understood as follows: These areas are typically located in the traditional center of the city or the old town, characterized by their mixed and chaotic streets. However, they often serve as urban public spaces characterized by gateway effects. People tend to apply more stringent evaluation criteria to these types of stations. Thus, evaluating the location situation provides insights into the design of the field environment and urban aesthetics. For stations with favorable location attributes, greater emphasis can be placed on the relationship between density, business type, and economic value, as people tend to be more accepting of these factors. Moreover, increased investment to enhance the beauty of the streets and elevate the overall image of the city is necessary.

5.2. Conclusions

This study explores the interaction between the built environment and public perception of CMSAs by combining SEM and IPA methods. The results revealed the effects of various variables (location, density, diversity, distance to transit, destination accessibility, road network, connection design, street characteristics, and order management) within the built environment (location, field environment, and urban aesthetics) on satisfaction. The main conclusions of the SEM model are as follows: First, perceptions of the field environment have a direct positive effect on satisfaction. Among these, density, diversity, and connection design are critical factors within the field environment dimension. Secondly, urban aesthetic perception directly affects satisfaction, with street characteristics being a critical factor. Thirdly, the location situation does not have a direct impact on satisfaction, but it indirectly influences satisfaction by positively affecting the field environment and negatively affecting urban aesthetics. Based on this, we used the value of the contribution of each element to perceived importance as the importance score, and the user satisfaction ratings of the elements as the performance score, to conduct an IPA analysis. We compared the importance and satisfaction alignment of various environmental elements across three sample CMSAs in Chengdu, and based on this, we proposed optimization suggestions.

This study aims to deeply understand the complex relationship between perceptions of the built environment and public satisfaction in the CMSAs. We provided an innovative approach to exploring environmental perception and assessing the effectiveness of the built environment. After the application of the proposed SEM-IPA model, the empirical results can provide valuable insights for urban designers and policymakers, as well as an empirical basis for the construction and renewal of central TOD in future cities.

This study has some limitations. Due to the limited sample size, sufficient data were not obtained for certain demographic characteristics, such as elderly individuals and children. Therefore, weight adjustments based on different sample proportions were not considered. In future research, we will aim to collect a broader range of samples using methods such as stratified sampling and targeted interviews with specific populations and further analyze whether significant differences in perceptions exist across various demographic groups.