The Potential of Basic Education Accessibility Across Administrative Boundaries Using a Multi-Scenario Comparative Analysis: How Can Education Equity in the Qinghai–Tibet Plateau Be Better Achieved?

Abstract

Ensuring equitable access to basic education is a core issue for promoting balanced regional development and sustainable educational outcomes. As a vast and sparsely populated region with relatively slow development, the Qinghai–Tibet Plateau faces particular challenges in ensuring educational accessibility and equity. Using a 100 m × 100 m travel time cost raster constructed from OSM road networks and the cost-distance method, together with local spatial autocorrelation, Lorenz curve, and Gini coefficients, as well as the Geodetector method, this study examines the spatial equity and factors influencing the accessibility of primary and secondary schools across 2798 townships at three time points (2016, 2020, and 2024) under three scenarios: Scenario 1 (nearby schooling), Scenario 2 (schooling within the prefecture-level city), and Scenario 3 (schooling within the county). The results show that: (1) Overall accessibility improved from 2016 to 2024, with primary schools being more accessible than secondary schools. Western townships, although initially disadvantaged, experienced the most notable gains. However, accessibility declined markedly when administrative-boundary constraints were imposed, with the greatest losses observed in ultra-high-altitude remote areas such as Ngari and Nagqu. (2) Spatial equity also improved, but when administrative boundaries were imposed, population-weighted inequities became even more pronounced than disparities in accessibility itself. Equity declined most sharply under county-level constraints, with pronounced impacts on both primary and secondary schooling. (3) Spatial variations in accessibility were jointly driven by multiple factors. In Scenario 1, road network density and population density had the strongest explanatory power. Under administrative boundary constraints, however, county type and ethnic autonomy became increasingly influential. In conclusion, in ultra-high-altitude areas where natural conditions remain difficult to overcome, improving educational equity depends less on transport expansion or facility provision and more on relaxing county-level boundary restrictions that constrain access to services. These findings may provide useful evidence to inform targeted policy interventions and resource allocation strategies aimed at promoting equitable access to basic education in underdeveloped and high-altitude regions.

1. Introduction

As China’s education sector gradually shifts toward a stage of high-quality development, regional disparities in educational development remain evident, marked by imbalance and relative inadequacy. These disparities may be further exacerbated through intergenerational transmission. Improving educational equity is closely aligned with Sustainable Development Goal 4 (SDG 4), which calls for “ensuring inclusive and equitable quality education.” Its sub-target SDG 4.1 further states that, by 2030, all children should complete free, equitable, and high-quality primary and secondary education and achieve effective learning outcomes [1]. However, in remote and high-altitude regions of China, the universal provision of compulsory education still faces considerable challenges. In response, the Implementation Plan for Building a Powerful Education Nation during the 14th Five-Year Plan explicitly calls for supporting underdeveloped regions in consolidating the achievements of educational poverty alleviation and accelerating the narrowing of regional education gaps [2]. Likewise, China’s Education Modernization 2035 emphasizes addressing the weak links in education development and achieving the equalization of basic public education services [3]. Promoting educational equity is not only a critical measure in implementing national strategies such as invigorating the country through science and education and strengthening the nation through talent, but also a fundamental prerequisite for advancing high-quality education. In this context, the Qinghai–Tibet Plateau, characterized by ecological fragility, slow development, and a sparse population across a vast territory, exhibits a relatively weak foundation and pronounced spatial imbalances in basic education development. Rationalizing the spatial distribution of primary and secondary schools in this region has thus become a pressing issue that demands urgent resolution [4].

Existing studies on educational equity have primarily focused on dimensions such as educational attainment [5], resource investment [6], and facility accessibility [7,8], employing methods like indicator-based evaluation [9], spatial analysis [10], and questionnaire surveys [11,12]. Among these, the accessibility of educational facilities has become a key research focus in recent years [13,14], as it reflects the spatial rationality of resource allocation. Specifically, most research has concentrated on large urban areas [15,16,17,18], utilizing approaches such as proximity analysis [19], time–cost distance methods [20,21,22], various improved two-step floating catchment area (2SFCA) models [23,24,25], and interview-based surveys [26]. These studies have consistently revealed significant inequalities in educational accessibility, with advantaged areas clustered in urban agglomerations, metropolitan regions, and city cores [27,28].

In the context of the Qinghai–Tibet Plateau, studies on educational development and infrastructure construction have gradually increased in recent years. However, the overall volume of research remains relatively limited. Most existing work highlights challenges such as insufficient teaching staff, suboptimal school location planning, and a mismatch between educational resources and local development needs [4,29]. Yet, given the Plateau’s unique natural and human geographical conditions, comprehensive analyses of spatial equity in basic education remain scarce. Furthermore, due to the region’s vast territory and large administrative divisions, traditional county-scale analyses are often inadequate for capturing internal spatial disparities. Therefore, a pressing need for higher-resolution and finer-scale approaches to measure accessibility and assess spatial equity.

In general, a smaller number of schools can facilitate the concentration of resources and improve the efficiency of basic education service provision. However, in sparsely populated regions, this may also increase travel burdens for residents in remote settlements [30,31,32,33]. In China, admissions to primary and secondary schools are primarily linked to administrative divisions, and school catchment boundaries are delineated mainly based on administrative jurisdictions, population distribution, and school layout [34,35]. For the Qinghai–Tibet Plateau, characterised by vast territory and low population density, various proactive initiatives, such as cross-regional school cooperation models, have helped expand access to high-quality educational resources. Nevertheless, after accounting for transport accessibility and geographic conditions, administrative boundaries at different levels may still influence opportunities for students to access educational services. Therefore, incorporating administrative boundaries as spatial constraints in this study has both institutional grounding and practical relevance, and the effects of varying boundary levels can be systematically evaluated through scenario comparisons.

Given this context, this study takes the Qinghai–Tibet Plateau as the research area and selects 2016, 2020, and 2024 as the study periods. These three time points represent consecutive stages before and during the 14th Five-Year Plan, capturing the progressive improvements in regional accessibility and public services, and comparable datasets are available for all three years, ensuring consistent analysis. A 100 m-resolution cost raster is constructed based on the OSM road network, and the cost-distance method is applied at the township/subdistrict scale to calculate travel time to primary and secondary schools under three schooling scenarios: nearby schooling, schooling within the prefecture-level city, and schooling within the county. On this basis, local spatial autocorrelation, Lorenz curves, and Gini coefficients are used to assess the spatial equity of educational accessibility. Furthermore, key explanatory variables, including natural conditions, population distribution, and road network density, are analyzed using the Geodetector model to identify the dominant factors influencing accessibility. The findings aim to provide scientific support for optimizing the allocation of basic education resources in the Qinghai–Tibet Plateau and offer practical guidance for promoting educational equity and improving spatial governance in underdeveloped regions.

2. Study Area and Overview of Educational Development

2.1. Overview of the Study Area

The Qinghai–Tibet Plateau region encompasses the entire Tibet Autonomous Region and Qinghai Province, as well as parts of Sichuan, Yunnan, Gansu, and Xinjiang provinces, covering a total area of approximately 2.6 million square kilometers [36]. As a critical ecological security barrier of China, the region has an average elevation exceeding 4000 m [36], forming a unique high-altitude socio-ecological system. The population and urban settlements are primarily concentrated in lower-altitude river valleys, such as the Huangshui Valley in eastern Qinghai and the Yarlung Tsangpo River Valley in southern Tibet. Notably, 90% of the population resides in areas that account for only 26.49% of the total land area, while regions above 5000 m in elevation generally have population densities below 1 person per square kilometer [37,38]. To maintain the administrative integrity of prefecture-level regions within the Plateau, this study selects 2798 township-level administrative units as the basic evaluation units.

2.2. Overview of Basic Education in the Study Area

In recent years, the development of education and the construction of related facilities on the Qinghai–Tibet Plateau have steadily improved. However, due to its sparse population and highly clustered settlement patterns, basic education resources still exhibit pronounced spatial imbalance at the county level. Based on comparative analyses of data from Qinghai Province and the Tibet Autonomous Region, the core provinces and regions of the Tibetan Plateau, with the national data (Figure 1), reveal the following: in terms of schools per capita, the gaps between Qinghai, Tibet, and the national average are not substantial, and in certain stages even exceed the national level. This suggests that, given the relatively small population base, these regions are not facing a severe shortage of schools per capita. By contrast, when measured as schools per unit area, the number of schools per township in the Plateau is far below the national average. The disparity is particularly striking in Tibet, where the density of primary, junior secondary, and senior secondary schools is only about one-third, one-ninth, and one-sixth of the national average, respectively. Qinghai also falls to roughly half the national level. This indicates that in the vast and sparsely populated Plateau, where modern high-speed transportation networks remain relatively underdeveloped, the per capita school index conceals pronounced spatial disparities in distribution, and the uneven allocation of educational facilities directly translates into challenges of accessibility. Therefore, relying solely on quantitative indicators provides an incomplete picture of service provision; accessibility becomes a crucial lens for evaluating both the level and equity of basic education services on the Plateau.

3. Research Methods and Data Sources

3.1. Research Methods

3.1.1. Cost Distance Method

A previous study has highlighted that, in the context of the Qinghai–Tibet Plateau, both the “proximity” principle based on geographic distance and the “locality” principle based on administrative boundaries may exert substantial impacts on schooling efficiency [39]. To further capture how institutional constraints shape access to basic education, three scenarios with progressively increasing administrative restrictions were developed in this study. Scenario 1 represents an idealized condition of nearby schooling, where students attend the nearest school based purely on travel time without any administrative limitation. Scenario 2 simulates schooling within prefecture-level jurisdictions, reflecting an intermediate degree of administrative management in which cross-prefecture mobility is restricted. Scenario 3 corresponds to schooling within county jurisdictions, aligning with the strictest interpretation of locality-based enrollment regulations commonly adopted in practice.

In constructing the travel-cost surface, road speeds were assumed to be constant by road class. This represents a necessary simplification, as real travel times may vary due to factors such as weather, seasonal conditions, slope, and pavement quality. Due to the lack of fine-grained spatiotemporal data, these factors were not explicitly modeled and are acknowledged as a source of uncertainty. Then, based on the Technical Standards for Highway Engineering (JTGB01-2014) [40] and related study [41], travel speeds were assigned to different road types (

Table 1. Setting of driving speed and comparative analysis scenarios.

Road/Boundary Type	Speed Assignments (km/h) and Boundary Friction Coefficients Under Scenario 1	Speed Assignments (km/h) and Boundary Friction Coefficients Under Scenario 2	Speed Assignments (km/h) and Boundary Friction Coefficients Under Scenario 3
motorway	110	110	110
trunk, primary	80	80	80
secondary, motorway_link	60	60	60
trunk_link, primary_link	50	50	50
tertiary	40	40	40
secondary_link	35	35	35
tertiary_link	25	25	25
Residential, service, track, living_street, busway	20	20	20
unknown, cycleway, bridleway	15	15	15
footway, pedestrian, paths, steps	5	5	5
Friction coefficient across prefecture-level administrative boundaries	1	0	0
Friction coefficient across county-level administrative boundaries	1	1	0

) and converted into time costs to construct a cost-distance raster at 100 m resolution. Administrative boundary friction coefficients were then superimposed to evaluate school accessibility and its disparities under varying administrative constraints. The model formula is as follows:

$$T_{ab}=min{\sum }_{i=1}^n{cost}_i$$

(1)

where $T_{ab}$ represents the minimum time cost from raster township a to school b, and ${cost}_i$ is the travel cost associated with the i th raster cell along the path. Thus, educational accessibility in this study is operationalized as the minimum travel time from each township to the nearest primary or secondary school. Compared with supply–demand approaches such as 2SFCA and threshold-based measures (for example, 30 or 60 min), raster-based continuous travel time is frequently applied to evaluate service accessibility in mountainous regions. This approach is particularly suitable for plateau environments, where physical travel conditions vary markedly across space. Continuous time measures preserve these pronounced spatial differences in travel effort and better reflect the heterogeneous nature of school access in such settings [42,43,44].

3.1.2. Spatial Autocorrelation

Local Moran’s I was applied to identify spatial clustering patterns of school accessibility across the Qinghai–Tibet Plateau, capturing spatial dependencies in the distribution of primary and secondary education facilities [45]. The formula is as follows:

$$I_i={\displaystyle \frac{(x_i-\overline{x}){\sum }_{j =1}^n{w}_{ij}(x_j-\overline{x})}{{\sum }_{i =1}^n{(x_i-\overline{x})}^2}}$$

(2)

where I_i represents the Local Moran’s I index for unit i, with values ranging from −1 to 1. A positive I_i indicates spatial clustering of similar values, while a negative value suggests dissimilarity or spatial heterogeneity. A value close to zero implies spatial randomness. n denotes the total number of spatial units, x_i and x_j represent the variable values for unit i and its neighbor j, $\overline{x}$ is the global mean, and w_ij is the spatial weight matrix [46].

3.1.3. Equity Measurement

To assess spatial equity in education accessibility, the Lorenz curve and Gini coefficient were employed. In this study, accessibility equity reflects how evenly travel times to the nearest schools are distributed across townships. A smaller variation in travel time indicates a more equitable distribution of educational opportunities, whereas a larger variation denotes inequity. The Lorenz curve plots the cumulative percentage of population against the cumulative percentage of travel time, indicating the level of distributional inequity. The closer the curve is to the diagonal, the more equitable the distribution [47]. The Gini coefficient further quantifies this inequality, ranging from 0 (perfect equality) to 1 (maximum inequality). Population-weighted Gini coefficients were used to capture not only spatial variation in accessibility but also differences in the number of people exposed to varying travel times. A higher value indicates greater inequality, meaning that a larger share of the population faces disproportionately long travel burdens, whereas lower values suggest a more equitable distribution. The Gini coefficient was adopted because it provides a concise and distribution-sensitive measure of spatial equity, allowing meaningful comparison of inequality across regions, scenarios, and years [48].

3.1.4. Geodetector

Geodetector is a statistical tool based on variance decomposition, used to quantify the influence of explanatory variables on spatial heterogeneity [49]. If a factor significantly affects the spatial distribution of education accessibility, its spatial stratification should align with that of the dependent variable. This study used the factor detection module of Geodetector to evaluate the explanatory power of 11 selected variables, including road network density. The model is expressed as:

$$q=1 - {\displaystyle \frac{\sum _{h = 1}^L{N}_h{\sigma }_h^2}{N{\sigma }^2}}$$

(3)

where q$\in$ [0, 1] represents the explanatory power of a given factor; higher values indicate stronger influence. N and $\sigma$² are the sample size and variance for the entire study area, while $N_h$ and ${\sigma }_h^2$ are the sample size and variance within stratum h. Based on the unique geographical characteristics of the Plateau, together with prior studies and the availability of data, eleven variables were selected to reflect both supply and demand side determinants of school accessibility (

Table 2. Indicator setting for factors influencing primary and secondary school accessibility at the township scale on the Qinghai–Tibet Plateau.

Category	Detection Factor	Detection Indicator
Natural Environment	X1	elevation
	X2	slope
	X3	distance to rivers
	X4	distance to the national border
Demographic and Socio-economic Factors	X5	population size
	X6	population density
Administrative Category	X7	road network density
	X8	township type
	X9	county type
	X10	whether the area is an autonomous prefecture for ethnic minorities
	X11	whether it is under the jurisdiction of a provincial capital city

). These variables were organized into three categories. The first category concerns natural environmental conditions, including elevation, slope, distance to rivers, and distance to national borders; these factors represent travel resistance, ecological constraints, and settlement patterns [50,51]. The second category consists of demographic and socioeconomic characteristics, including population size, population density, and road network density, which indicate the intensity of educational demand and the efficiency of transportation [52,53]. The third category comprises administrative attributes, including township type, county type, ethnic autonomy, and whether a township falls under the jurisdiction of a provincial-level capital, which are related to differences in resource allocation and governance capacity [44,54,55]. While accessibility also involves socio-economic and subjective dimensions (e.g., household income, school choice preference), this study focuses on objective physical accessibility due to the limited availability of consistent township-level time-series data. These aspects can be further investigated when finer micro-scale datasets become accessible. Factor detection was conducted for educational accessibility in the years 2016, 2020, and 2024.

3.2. Data Sources

The dataset used in this study includes school numbers, school points of interest (POI), population distribution, road networks, and topographic information. Specifically, school numbers were obtained from the China Educational Statistics Yearbook (2023), available at http://www.moe.gov.cn/jyb_sjzl/moe_560/2023/ (accessed on 9 September 2025), which provides the most recent official counts. School POI data for 2016, 2020, and 2024 were obtained from Amap (Gaode Maps). Population datasets corresponding to the three study years were sourced from the LandScan Global Population Database, available at https://landscan.ornl.gov/ (accessed on 9 September 2025). Road network data were collected from OpenStreetMap. Topographic data, including elevation, slope, and hydrology, were obtained from the Resource and Environment Science and Data Center of the Chinese Academy of Sciences, available at https://www.resdc.cn/ (accessed on 9 September 2025).

4. Results

4.1. Spatiotemporal Changes in Primary and Secondary School Accessibility

From 2016, through 2020, to 2024, the overall accessibility of primary and secondary schools on the Qinghai–Tibet Plateau has improved markedly, with average travel times continuously declining and the three-hour accessibility rate steadily increasing. Spatially, areas with relatively good baseline accessibility are mainly concentrated in the southeastern part of the Plateau and have consistently maintained high accessibility levels. In contrast, parts of the central Plateau have experienced limited improvements, or even declines and fluctuations, while the western Plateau, which initially had the weakest accessibility, has shown the most significant progress. Overall, the spatial pattern can be characterized as “stable in advantaged areas, marked improvement in peripheral areas, and relatively lagging in hinterland areas” (Figure 2).

In terms of educational stage, accessibility to primary schools is consistently better than that to secondary schools (

Table 3. Average travel time (min) and efficiency loss from various towns to their nearest primary and secondary schools in 2016, 2020, and 2024 under different scenarios.

Scenario	Education Stage	2016	2020	2024
1	Primary School	99	40.8	29.4
	Secondary School	103.8	51.6	42.0
2	Primary School	111	47.4	33.6
	Scenario 2-1 Efficiency Loss	12.12%	16.18%	14.29%
	Secondary School	115.2	58.8	46.2
	Scenario 2-1 Efficiency Loss	10.98%	13.95%	10.00%
3	Primary School	118.8	51.6	35.4
	Scenario 3-1 Efficiency Loss	20.00%	26.47%	20.41%
	Secondary School	125.4	61.8	51
	Scenario 3-1 Efficiency Loss	20.81%	19.77%	21.43%

). Under the “nearby schooling” scenario, the average minimum travel time to primary schools decreased from 99 min in 2016 to 29.4 min in 2024, representing a reduction of 70.3%. By comparison, secondary school accessibility improved from 103.8 min to 42 min over the same period, a 59.54% reduction. Although the improvement is substantial, secondary school accessibility remains relatively weaker, which is closely linked to facility distribution. Primary schools are typically distributed at high density around rural settlements and population centers, demonstrating strong local service characteristics, while secondary schools are more concentrated to achieve resource integration and economies of scale. This results in longer travel times for students in peripheral areas.

A comparison between the “nearby schooling” scenarios and the “local schooling” scenario (Table 3) reveals that restricting schooling within prefecture-level boundaries has relatively limited effects for most areas, with significant increases in travel time observed only in some towns at prefectural margins. However, when schooling is restricted within county-level boundaries, the impacts are particularly pronounced in the highest-altitude regions such as Ngari Prefecture and Nagqu City in Tibet, where travel times increase substantially. Specifically, in 2024, the average travel time to primary and secondary schools increased to 33.6 min and 46.2 min, respectively, under prefecture-level restrictions, with average losses of 14.3% and 10% relative to nearby schooling. Under county-level restrictions, the averages further increased to 35.4 min and 51 min, corresponding to losses of 20.41% and 21.43%.

Overall, administrative boundaries, particularly at the county level, significantly exacerbate spatial inequalities in educational accessibility under the Plateau’s geographic constraints, with the effects being especially pronounced in extremely high-altitude peripheral regions and at the secondary school level.

To complement the continuous measure of minimum travel time, this study also adopts the three-hour accessibility rate, defined as the proportion of townships that can reach the nearest primary or secondary school within three hours. This threshold-based indicator evaluates whether a basic standard of access to compulsory education is being met, which is particularly relevant for remote, high-altitude settlements on the Plateau.

Building on this indicator, areas with an accessibility rate above 80% continued to expand, and the overall accessibility level under all scenarios showed a steady improvement trend. In particular, under the “nearby schooling” scenario, the average three-hour accessibility rate for primary and secondary schools increased from 80.9% and 81.4% in 2016 to 96.8% and 95.8% in 2024, respectively. The increase in accessibility for secondary schools was generally lower than that for primary schools (

Table 4. Average 3 h accessibility of townships under the districts and counties to the nearest primary and secondary schools (%) in 2016, 2020, and 2024 under different scenarios, and increase.

Scenarios	Education Stage	2016	2020	2024	Increase 2016–2020	Increase 2020–2024	Increase 2016–2024
1	Primary School	80.9	95.3	96.8	17.8%	1.57%	19.65%
	Secondary School	81.4	94	95.8	15.48%	1.91%	17.69%
2	Primary School	79.1	93.8	95.7	18.58%	2.03%	20.99%
	Secondary School	79.5	91.2	94.3	14.72%	3.4%	18.62%
3	Primary School	74.1	86.4	91.8	16.6%	6.25%	23.89%
	Secondary School	73.1	82	87.4	12.18%	6.59%	19.56%

). In contrast, under the “schooling within county” scenario, the three-hour accessibility rate for secondary schools in 2024 was only 87%, indicating that a considerable share of townships still face long travel times. By comparison, primary schools achieved above 90% coverage across all scenarios, reflecting more extensive spatial availability. While some counties and townships reached nearly 100% accessibility under the “nearby schooling” scenario, accessibility decreased significantly in most remote areas under the “local schooling” scenarios due to school distribution and topographic constraints such as elevation. These declines were further amplified at the county scale, underscoring the potential restrictive role of administrative boundaries on educational service accessibility, with particularly pronounced limitations for secondary education.

4.2. Equity and Its Spatial Characteristics

The population-weighted local spatial autocorrelation of educational accessibility reveals significant spatial heterogeneity across the Qinghai–Tibet Plateau (Figure 3). Specifically, the “low–low clusters” of travel costs remain relatively stable with minor fluctuations, mainly concentrated in urban core areas with favorable topographic conditions, high school density, and along major transportation corridors. These areas may exhibit a path dependence and cumulative advantage mechanism characterized by “resource concentration–spatial advantage–reinforcement.” In contrast, the “high–high clusters” show a pattern of spatial concentration and gradual contraction, predominantly distributed in the Plateau’s hinterland, border areas, and regions distant from major transportation routes. Constrained by shortages of educational facilities, challenging natural environments, inconvenient transport, and sparse populations, these disadvantaged areas display persistent spatiotemporal patterns of “peripheral marginalization.” The “high–low clusters” are mostly observed at urban peripheries or urban–rural transition zones, where educational accessibility is hindered by administrative boundaries, terrain constraints, and insufficient transport infrastructure. These patterns reflect a mismatch between the spatial distribution of educational resources and the populations they serve. By contrast, the “low–high clusters” likely benefit from the spillover effects of neighboring advantaged regions, yet their distribution remains scattered and unstable, highlighting the fragility and uncertainty of regional educational equity. Overall, the results indicate pronounced spatial differentiation in educational accessibility across the Plateau, leaving substantial room for improving the fairness of opportunities to access educational services.

Building on this, the Lorenz curves and Gini coefficients further confirm the improvement trend in spatial equity. Between 2016 and 2024, the Gini coefficients of educational accessibility under both the nearby schooling and local schooling scenarios declined, with Lorenz curves converging toward the equity line. Moreover, secondary schools generally showed higher equity levels compared to primary schools (Figure 4,

Table 5. Gini coefficients for the shortest time to primary and secondary school for each township in 2016, 2020, and 2024 under different scenarios.

Scenario	2016 (Primary/Secondary School)	2020 (Primary/Secondary School)	2024 (Primary/Secondary School)
1	0.68/0.65	0.65/0.59	0.60/0.52
2	0.73/0.71	0.73/0.68	0.67/0.62
3	0.84/0.83	0.89/0.88	0.89/0.86

). Specifically, under the nearby schooling scenario, the Gini coefficients for primary and secondary schools decreased from 0.68 and 0.65 in 2016 to 0.60 and 0.52 in 2024, respectively, reflecting the significant impacts of transport infrastructure optimization and school location adjustments. Nevertheless, by 2024, both values remained relatively high. In practical terms, a Gini coefficient around 0.60 reflects a high degree of spatial inequality, meaning that accessibility advantages are concentrated within a limited number of townships, whereas many peripheral and high-altitude areas face disproportionately long travel times. Thus, despite improvements over time, inequity in school accessibility persists. The comparison of local schooling scenarios further demonstrates the restrictive effect of administrative boundaries on equity. Furthermore, when schooling is restricted within county boundaries, the Gini coefficients remain as high as 0.89 for primary schools and 0.86 for secondary schools, suggesting that administrative boundaries may play a constraining role and are associated with spatial differences in access to educational services.

4.3. Analysis of Influencing Factors

According to the Geodetector results (

Table 6. Results of Factor Detection.

Education Stage	Year	Scenario	X1	X2	X3	X4	X5	X6	X7	X8	X9	X10	X11
Primary School	2016	1	0.0398	0.0251	0.0045 **	—	0.0447	0.0643	0.0829	0.0313	0.0103	0.0104	0.017
		2	0.0311	0.0249	0.0062 *	—	0.0421	0.0543	0.075	0.0308	0.0173	0.0124	0.016
		3	0.0242	0.0211	0.0061	—	0.0327	0.0472	0.0502	0.0227	0.0154	0.0157	0.0153
	2020	1	0.0497	0.0111	—	—	0.0401	0.0606	0.0512	0.0214	—	0.0031	0.0153
		2	0.0209	0.0055	0.0063 **	—	0.0217	0.03	0.028	0.0212	0.0045	0.009	0.0102
		3	0.0159	0.0063	0.0077	—	0.0197	0.0262	0.0258	0.017	0.0075	0.0108	0.0103
	2024	1	0.0305	—	0.0032	—	0.03	0.0373	0.0387	0.0146	—	0.0146	0.0077
		2	0.0306	0.0101	0.0072	—	0.0307	0.036	0.023	0.0183	—	0.007	0.0093
		3	0.0235	0.01	0.0069 **	—	0.03	0.0345	0.0306	0.0147	0.0074	0.0072	0.01
Secondary School	2016	1	0.0509	0.0315	0.0070 **	—	0.0545	0.0766	0.0899	0.0347	0.0118	0.0099	0.0169
		2	0.0435	0.0312	0.0061 *	—	0.052	0.0705	0.0797	0.035	0.014	0.0123	0.0167
		3	0.0396	0.0249	—	—	0.0388	0.0641	0.0561	0.0315	0.0163	0.0079	0.0159
	2020	1	0.0757	0.0156	0.0053 *	—	0.0528	0.0889	0.0665	0.0292	0.0068	0.005	0.0174
		2	0.0341	0.0085	0.0072	—	0.0289	0.0452	0.0368	0.0249	0.0062	0.0072	0.0122
		3	0.0311	0.0109	0.0055 *	—	0.03	0.0457	0.0363	0.022	0.0107	0.0074	0.012
	2024	1	0.0678	0.0109	0.006	—	0.0584	0.0852	0.0634	0.0248	0.0071	0.004	0.013
		2	0.0619	0.0224	0.0092	—	0.0549	0.078	0.0509	0.0284	0.0093	0.0106	0.0147
		3	0.0481	0.0231	0.0103	—	0.0538	0.0748	0.0429	0.0269	0.0101	0.0131	0.0153

p < 0.001 unlabeled; p < 0.005 **; p < 0.01 *.

), although most variables exhibit relatively low q-values, all of them demonstrate statistically significant spatial explanatory power for educational accessibility across years and school levels (p < 0.001). Their variation trends are consistent with the spatial distribution of accessibility, indicating that these factors exert a stable and meaningful influence on shaping regional disparities in access to education.

Among them, road network density (X7) consistently shows the strongest explanatory power across years and school stages, suggesting that educational accessibility is highly coupled with the distribution of transport infrastructure. Population density (X6) and population size (X5) also demonstrate stable and significant explanatory power, further underscoring the strong linkage between accessibility, population concentration, and the allocation of educational resources. Elevation (X1) and distance to rivers (X3) exert weaker but stable impacts, reflecting the continued constraint of the Plateau’s rugged environment, while slope (X2) was significant only in 2016, suggesting that the expansion of educational and transport infrastructure has gradually offset the barriers imposed by terrain variation. In addition, administrative category factors such as township type (X8), county type (X9), and provincial capital status (X11) exert a stable influence on accessibility. Notably, whether an area is designated as an ethnic autonomous region (X10) was significant in 2016 but not thereafter, reflecting the effects of infrastructure investment and equity-oriented education policies in reducing disparities and weakening the distinctiveness of ethnic areas.

Furthermore, when comparing explanatory power across Scenarios 1 to 3 and across years and school levels, q-values for natural and socio-demographic factors such as elevation, population density, population size, and road network density show a declining trend, with the highest explanatory power under the nearby schooling scenario and the lowest under the schooling within county scenario. This indicates that as the schooling range becomes increasingly restricted, accessibility is less influenced by transport networks, physical geography, and population patterns, and more constrained by administrative boundaries. Conversely, administrative factors such as county type and ethnic autonomy show increasing explanatory power from Scenario 1 to Scenario 3, highlighting that under local schooling arrangements, administrative boundaries have become the dominant constraint on educational equity.

In sum, educational accessibility across the Qinghai–Tibet Plateau is shaped by a multifactorial spatial framework, with road network density and population density consistently emerging as the core drivers. Natural conditions primarily impose baseline constraints, while policies and administrative divisions act as moderating factors. Although most variables achieve statistically significant explanatory power, their overall q-values remain relatively low, likely because the effects of natural environments have already been partially internalized into the spatial layout of schools. Improvements in accessibility for disadvantaged regions thus rely more heavily on policy support and planning interventions. More importantly, the comparative results across scenarios reveal that, relative to transport improvements alone, removing the constraints of administrative boundaries is a more worthwhile strategy to consider for enhancing educational accessibility in high-altitude and underdeveloped areas.

5. Discussion

5.1. Accessibility Patterns on the Qinghai–Tibet Plateau Exhibit Scale Effects

Building on existing research on accessibility in the Qinghai–Tibet Plateau, this study employs 100 m resolution raster data and township-level measurements to examine the spatiotemporal characteristics of primary and secondary school accessibility. The overall pattern shows that accessibility is generally better in the east than in the west, while border and high-altitude areas remain disadvantaged. This finding is broadly consistent with previous studies [56,57].

Time-series analysis at the township scale further reveals persistent spatial heterogeneity in educational accessibility across the Plateau. This differentiation largely reflects underlying disparities in baseline transport infrastructure, settlement concentration, and targeted policy support. Southeastern areas with favorable baseline conditions consistently maintain high accessibility levels. Western regions, despite weak initial accessibility, achieve the greatest improvements under policy support, which is consistent with evidence that improvements in transport accessibility are strongly correlated with baseline levels of connectivity [56]. In contrast, the central interior areas of the Plateau show delayed progress, with some even experiencing fluctuations or declines, forming pockets of stagnation that differ from the generally observed trend of increasing transport equity across the Plateau.

Conventional county-level evaluations often obscure internal spatial disparities. By incorporating township-level assessments and analyzing accessibility under different administrative boundary constraints, this study identifies pronounced intra-county inequalities in educational resource allocation. For instance, some eastern townships, although located in relatively favorable areas, still fall into relative gaps in service coverage because high-quality educational resources are highly concentrated and the spillover effects are limited [56].

Despite methodological refinements such as the construction of fine-grained cost rasters and the simulation of multiple boundary scenarios, certain limitations remain. First, the model relies on static cost rasters for least-cost path analysis and does not fully incorporate seasonal variations [58], topographic adjustment factors, or the impacts of climatic hazards, which may reduce model accuracy in the Plateau’s unique environment. Second, the evaluation focuses mainly on the spatial distribution of facilities and does not include key determinants such as teacher quality, school hierarchy, and educational outcomes, which directly affect the actual service capacity of schools. Future research should therefore expand toward integrated assessments of service provision capacity.

5.2. The Spatial Equity and Influencing Factors of Basic Education on the Qinghai–Tibet Plateau Exhibit Distinctive Features

Previous studies have widely noted that the allocation of educational resources in China is characterized by significant spatial imbalance, which is especially pronounced in regions with complex natural conditions [21,45,52,54,55]. It is particularly noteworthy that accessibility tends to decline as school level increases, and equity also deteriorates, with secondary schools being significantly less accessible than primary schools [24,59,60]. This pattern is also confirmed in the study area.

Existing studies indicate that educational accessibility on the Plateau remains markedly unequal, particularly in areas with complex terrain and weak transport infrastructure [54,61]. These disparities are broadly consistent with our findings. Prior research further reports Gini coefficients of approximately 0.5–0.6 in Tibet and 0.7–0.8 in Qinghai, indicating that spatial inequalities persist across the region [21]. From a regional perspective, accessibility levels are significantly higher in areas with favorable terrain, whereas high-altitude and topographically complex areas still face accessibility deficits. Other studies have also pointed out that within provinces characterized by pronounced geographic variation, educational resources and transport facilities are often concentrated in advantageous areas, while areas with more challenging environments face structural constraints and weaker educational accessibility. This spatial imbalance is particularly evident on the Plateau [39,54]. From the perspective of equity evolution, although this study finds that overall coverage of education services has expanded, Lorenz curves and Gini coefficients continue to reveal internal disparities. At finer spatial scales, the allocation of educational resources remains inequitable, which mirrors trends observed in other regions [60,62,63].

Regarding influencing factors, different regions display multidimensional mechanisms due to variations in geography, social structures, and policy systems. The Qinghai–Tibet Plateau, as an ecologically fragile, sparsely populated, and relatively underdeveloped large-scale region, is shaped by intertwined constraints of natural, social, and policy conditions. Mountainous areas are marked by rugged topography, high road construction costs, and weak connectivity, which directly shape the layout of schools and access opportunities. This finding is consistent with studies of other high-altitude or topographically complex regions [25,62,64]. In contrast, the relatively flat terrain and well-developed transportation networks in eastern and central China weaken such constraints, with population density and land-use efficiency becoming stronger drivers [65]. Consequently, improving network connectivity in the Plateau may represent a critical pathway for optimizing facility distribution and enhancing equitable access. Population size and density are also decisive, functioning as key mediators between supply and demand, a pattern that holds across highland, minority, impoverished, and urban regions [51,54,59,64,65]. Policy mechanisms further play an essential role in improving accessibility in peripheral areas, particularly under extreme geographic conditions [54]. Results of this study indicate that in the Plateau, educational resource allocation relies less on market logic and more on top-down policy interventions. This institutional guidance enables equity to partially overcome natural constraints, echoing findings from Tibet and other minority areas [54].Moreover, when students are free to cross administrative boundaries, accessibility differences are mainly determined by transportation, geography, and population distribution, and improving education in transport-connected and populous areas can effectively narrow gaps. Therefore, promoting cross-boundary schooling and optimizing interregional allocation of educational resources represent possible measures for mitigating uneven accessibility and improving equity in remote areas of the Plateau.

This study also faces several limitations. First, although the analysis incorporates three benchmark years (2016, 2020, and 2024) aligned with key policy milestones, the time span remains insufficient to reveal longer-term spatial dynamics, particularly the coordinated improvements in education and transportation infrastructure during the “14th Five-Year Plan” period, which require further investigation. Second, policy response delays and variations in local implementation capacity may affect actual accessibility, suggesting that future work should integrate local surveys and planning documents to provide empirical support. Third, due to limited data availability, this study captures only objective accessibility based on minimum travel time. The analysis does not incorporate the characteristics of specific school-age groups, such as age structure or mobility patterns, nor does it account for differences in school quality or capacity. In addition, subjective dimensions of accessibility, such as parents’ school preferences and perceived travel burden, are not reflected. These factors may influence real-world access outcomes and could be further examined when finer micro-scale datasets become available in future studies. Finally, the absence of indicators such as local fiscal expenditure and governance capacity constrains the analysis of policy impacts on spatial heterogeneity. Future research should therefore extend across temporal, spatial, and population dimensions by incorporating finer demographic and transportation data, along with field surveys and policy documents, to construct a more systematic explanatory framework for educational accessibility.

5.3. Policy Implications

Based on the above findings, several policy recommendations are proposed to further enhance the accessibility and equity of basic education on the Qinghai–Tibet Plateau, with the goal of promoting a fairer, more coordinated, and sustainable regional education system. These recommendations integrate the differentiated accessibility patterns revealed under multiple schooling scenarios and reflect both natural constraints and institutional barriers identified in this study.

First, improving transportation infrastructure should be regarded as a key strategy for optimizing educational accessibility. Since transport connectivity directly determines the service range and efficiency of educational facilities, continuous investment is needed in remote areas, particularly in the Plateau’s hinterland and peripheral rural settlements. Accelerated construction and maintenance of access roads, bridges, and through a “transportation-first” approach can expand the spatial coverage of education services and significantly improve both accessibility and sustainability.

Second, the spatial layout of educational resources should be optimized in accordance with population distribution and aggregation patterns. Establishing flexible adjustment mechanisms is recommended to ensure adequate resource supply in densely populated areas through additional school places, while in sparsely populated regions, small-scale, multifunctional educational facilities should be encouraged. This approach allows the supply of education to better match population distribution and enhances the efficiency of resource allocation.

Third, educational resources should be deployed in a more refined and differentiated manner, taking local topographic conditions into account. The allocation of facilities should shift from uniform distribution toward efficient coverage. In townships with advantages in transportation, terrain, or population, priority should be given to capacity expansion and the addition of facilities to improve service coverage. Meanwhile, in areas constrained by the Plateau’s unique environment, diversified educational models, such as micro-schools, boarding schools, and distance learning platforms, should be actively promoted to gradually mitigate the barriers imposed by terrain on equitable access.

Finally, mechanisms for cross-administrative coordination and equity assurance must be strengthened. Building interregional education cooperation networks can improve service provision in marginal and disadvantaged areas. A governance system based on regional collaboration and resource sharing would help break down barriers created by administrative boundaries and facilitate the balanced flow and redistribution of educational resources. In particular, minority areas should continue to benefit from targeted support and preferential policies, including greater investment in teacher training, educational facilities, and long-term equity assurance mechanisms.

6. Conclusions

This study combined multi-source geographic data of the Qinghai–Tibet Plateau, using townships and subdistricts as the basic evaluation units, and assessed the spatiotemporal patterns of primary and secondary school accessibility in 2016, 2020, and 2024 under three scenarios: nearby schooling, schooling within the prefecture-level city, and schooling within the county. A “three-hour accessibility rate” indicator was further introduced, followed by spatial autocorrelation analysis, the Gini coefficient, and Lorenz curves to examine spatial equity and its evolution. In addition, the Geodetector method was employed to identify the dominant factors and their explanatory power across different scenarios. The main conclusions are as follows:

(1)

From 2016 through 2020 to 2024, educational accessibility on the Plateau improved significantly, with average travel times decreasing and the three-hour accessibility rate steadily increasing. The spatial pattern revealed in this study exhibits “advantages in the east, progress in the west, and lagging hinterlands.” Administrative boundaries posed potential constraints on accessibility. The shortest average travel times were observed under the nearby schooling scenario, while prefecture-level restrictions had only limited additional impacts. In contrast, county-level restrictions significantly amplified disparities, particularly in high-altitude areas such as Ngari Prefecture, Nagqu City, and other peripheral counties.

(2)

During this period, spatial inequity in school accessibility decreased markedly, as reflected by the declining overall Gini coefficients, especially at the secondary school level. However, under the county-level schooling scenario, spatial inequalities remained pronounced, indicating that resource allocation at finer administrative scales still requires further optimization.

(3)

Accessibility patterns were jointly shaped by multiple factors. Road network density and population density emerged as the strongest socioeconomic drivers, while natural geographic conditions served as fundamental constraints, and administrative factors became increasingly significant in remote and minority areas. As the scope of schooling was restricted, the explanatory power of natural and demographic factors decreased, whereas the influence of administrative divisions tend to increase.