Efficiency and determinants of science and technology investment in Chinese competitive sports: a provincial DEA-Tobit analysis

Abstract

Introduction:

The efficiency with which science and technology (S&T) investment is translated into competitive sports performance is a critical yet underexplored issue in China, despite the rapid growth of related funding in recent years.

Methods:

This study evaluates the efficiency of provincial S&T investment in competitive sports across 31 Chinese provinces from 2018 to 2022. Using a two-stage analytical framework grounded in the Resource-Based View, we integrate Data Envelopment Analysis (DEA), the Malmquist productivity index, and Tobit regression to assess static efficiency, dynamic productivity changes, and their key influencing factors.

Results:

The results indicate that the overall efficiency of S&T investment in China’s competitive sports system remains relatively low, with pronounced regional disparities. Dynamic analysis reveals that total factor productivity declined on average, primarily due to limited technological progress rather than changes in technical efficiency. Tobit regression results show that research staff input and the emphasis on technological R&D are negatively associated with efficiency, while regional economic development improves pure technical efficiency but undermines scale efficiency.

Conclusion:

These findings suggest that the main constraint on the effectiveness of S&T investment lies not in the quantity of resources but in their strategic configuration and conversion into athletic outcomes. Policy efforts should therefore shift from expanding investment scale toward optimizing resource orchestration and strengthening the linkage between scientific research and training practice to enhance the overall efficiency and sustainability of competitive sports development in China.

1 Introduction

Competitive sports function not only as a reflection of national culture and strength but also as an important indicator of economic and social development (1, 2). Over current many years, China has achieved outstanding success in main worldwide sporting competitions. These achievements have been strongly supported by sustained investment in science and technology (S&T). From 2018 to 2022, China’s S&T investment in competitive sports activities elevated by practically 50%, reflecting sturdy nationwide prioritization and an growing reliance on technological development. However, the effectivity of this substantial investment—how successfully assets are transformed into sporting success—stays a vital query that warrants in-depth exploration. The effectiveness of S&T enter subsequently relies upon critically on rational useful resource allocation and environment friendly utilization.

To consider such effectivity, Data Envelopment Analysis (DEA) has been broadly employed in sports activities analysis as a result of it accommodates a number of inputs and outputs with out requiring a pre-specified manufacturing operate (3–5). However, commonplace DEA fashions are restricted in their means to account for stochastic disturbances and endogeneity when analyzing the determinants of effectivity. The Tobit mannequin, which is effectively fitted to truncated dependent variables comparable to effectivity scores, is subsequently generally used to enhance DEA analysis (6, 7). Integrating the DEA and Tobit fashions leverages the strengths of each, permitting not just for an correct evaluation of S&T investment effectivity but in addition for a deeper exploration of the elements that affect it.

Existing analysis on S&T investment in sports activities has primarily targeted on macro-level profit analyses or on the applying of particular applied sciences in remoted domains (8–11). There is a lack of systematic research evaluating the general effectivity of S&T investment and its influencing elements, notably from a complete, provincial-level perspective in China. Moreover, the determinants of effectivity are advanced and multidimensional—encompassing coverage, human assets, and organizational administration—but the mechanisms by way of which these elements function stay insufficiently explored.

To floor this empirical investigation theoretically, the research adopts the Resource-Based View (RBV) as its overarching framework. It conceptualizes the S&T inputs—funding, establishments, and personnel—as strategic assets for competitive sports activities growth. Accordingly, the core analysis query shifts from “what is the level of efficiency?” to “how effectively do different regions convert strategic S&T resources into competitive performance?”. This perspective inherently directs consideration to the conversion mechanisms and contextual elements that designate effectivity disparities, which the following DEA and Tobit fashions are designed to measure and analyze.

In mild of these concerns, this research employs the DEA-Tobit mannequin to systematically consider the effectivity of S&T investment in China’s competitive sports activities growth throughout 31 provinces from 2018 to 2022 and to investigate the important thing elements influencing this effectivity. The goal is to supply theoretical assist and sensible steerage for optimizing S&T useful resource allocation and enhancing the standard of competitive sports activities growth.

2 Literature review

2.1 The application of the DEA-Tobit model in the field of sports

The DEA–Tobit model, which combines Data Envelopment Analysis (DEA) for efficiency measurement with Tobit regression for determinant analysis, has become a widely adopted approach in studies of sports efficiency. DEA has been extensively used to assess the efficiency of public sports services, the sustainable development of competitive sports, and the integration of the sports industry with tourism (12–17). For instance, prior research have utilized DEA to measure provincial effectivity in public sports activities providers by incorporating environmental variables comparable to GDP, and to investigate growth traits in competitive sports activities (12, 13). Tobit regression is then employed to look at the consequences of environmental and managerial elements on DEA-derived effectivity scores, thereby addressing DEA’s limitations in dealing with exogenous influences. Collectively, these functions show the mannequin’s utility in diagnosing effectivity and guiding useful resource optimization. However, present analysis has predominantly targeted on common sports activities providers or industrial effectivity, whereas systematic and longitudinal analyses particularly analyzing the effectivity of S&T investment in competitive sports activities—and its determinants throughout numerous regional contexts—stay scarce.

2.2 The impact of scientific and technological input on the development of competitive sports

Science and technology (S&T) input is widely recognized as a critical driver of modern competitive sports, underpinning improvements in training effectiveness, competitive performance, and decision-making processes. Technological tools—including biomechanical analysis, physiological monitoring, smart wearable devices, and data mining techniques—play an increasingly important role in informing athlete training and tactical planning (15, 16). For occasion, in desk tennis, information mining programs allow fine-grained technical and tactical analysis, whereas in monitor and subject, superior sensor applied sciences improve the accuracy of movement seize (15, 16). Furthermore, functions in sports activities drugs and digital actuality (VR) coaching assist athlete well being administration and the event of real-time decision-making capabilities (17, 18). These instances underscore that S&T investment permeates all points of competitive sports activities, from expertise cultivation to elite efficiency. However, a lot of the prevailing literature examines the consequences of particular applied sciences in isolation. A complete, system-level analysis of the effectivity of bundled S&T assets—together with funding, establishments, and personnel—stays underexplored.

2.3 Existing studies on the influential factors of S&T investment

Prior studies on the determinants of S&T investment efficiency emphasize the roles of the policy environment, human resources, and research and development (R&D) systems. Government policy support and fiscal investment are widely regarded as important drivers of technological development and resource allocation efficiency in the sports sector (19–21). In addition, the effectivity of public R&D investment and the construction of fiscal decentralization have been proven to affect technological innovation and its diffusion and utility on the native degree (22, 23). Nevertheless, present research have paid restricted consideration to the mechanisms by way of which multidimensional elements—comparable to regional financial growth, administration enter, and the alignment between analysis actions and sports activities follow—work together to form the effectivity with which S&T assets are transformed into athletic outcomes. This hole highlights the necessity for an built-in analytical framework that concurrently evaluates effectivity and examines the affect of contextual elements—an goal addressed in the current research.

3 Materials and methods

3.1 Materials

3.1.1 Evaluation index selection

Guided by the Resource-Based View (RBV), this study conceptualizes science and technology (S&T) input as a systematic configuration of financial, human, and institutional resources dedicated to enhancing athletic training, performance, and competitive outcomes. Consistent with RBV principles, the empirical analysis follows a two-stage logic. First, the Data Envelopment Analysis (DEA) model is employed to assess the relative efficiency with which bundled S&T resources are transformed into competitive sports achievements, thereby capturing regional capabilities in resource conversion. Second, a Tobit regression model is used to examine how contextual and environmental factors influence this conversion process and to explain observed efficiency differentials across regions.

Based on this analytical framework, an input–output indicator system was constructed. The input indicators reflect four key dimensions of S&T resources: financial investment in scientific research (X1), institutional infrastructure (X2), fiscal support for S&T research and development (X3), and human resource input measured by the number of research personnel (X4). The output indicators correspond to the core objectives of competitive sports development, including the production of elite athletes (Y1), the sustainability and quality of the talent pipeline as proxied by the size of the referee team (Y2), and competitive performance measured by the number of world champions (Y3). Detailed definitions and measurement units of all indicators are provided in Table 1.

The empirical data were obtained from the Statistical Yearbook of Sports Affairs and official publications of the General Administration of Sport of China, covering 31 provinces from 2018 to 2022. These provinces are treated as decision-making units (DMUs) in the efficiency analysis. With respect to the output indicators, the number of elite athletes and world champions was selected as representative measures of competitive sports performance. Although these outcomes may exhibit a time-lag effect—reflecting the cumulative impact of long-term investment rather than immediate technological input—they remain the most standardized and authoritative indicators within China’s provincial sports evaluation system. Moreover, in contemporary competitive sports, technological support is increasingly embedded in the daily training, monitoring, and performance optimization of existing elite athletes. Accordingly, these indicators are considered a valid, albeit lagged, reflection of the effectiveness of the current S&T support system.

3.1.2 Data sources

The data utilized in this study are derived from the China Statistical Yearbook and data compiled by the General Administration of Sport of China. The dataset spans the period from 2018 to 2022 and encompasses 31 provinces and municipalities in China. To ensure data integrity and the reliability of interprovincial comparisons, particular attention was paid to the treatment of missing values. Given that minimum-value substitution may bias efficiency estimates in DEA by artificially distorting the production frontier, a more robust imputation strategy was adopted. Specifically, missing observations for certain provinces were estimated using regional mean values or linear interpolation based on adjacent years, depending on data availability and temporal continuity. This approach helps prevent the introduction of extreme or implausible values, thereby preserving the stability of the DEA frontier and providing a more accurate representation of relative efficiency across provinces.

3.2 Methods

3.2.1 DEA-BCC model

Data Envelopment Analysis (DEA) is a non-parametric method widely used to evaluate the relative efficiency of decision-making units (DMUs) with multiple inputs and outputs. Originally proposed by Charnes, Cooper, and Rhodes (CCR) (3) in the late Seventies, the classical DEA mannequin assumes fixed returns to scale. To loosen up this assumption, Banker, Charnes, and Cooper (24) subsequently developed the BCC mannequin, which permits for variable returns to scale and decomposes total technical effectivity into pure technical effectivity and scale effectivity.

In DEA analysis, inputs and outputs are incorporated into a linear programming framework to construct an empirical production frontier. An input-oriented specification is adopted when the objective is to minimize input usage while maintaining a given level of output. This orientation is particularly appropriate in contexts where output levels are largely constrained, and efficiency improvements are expected to arise from better resource utilization. As a non-parametric approach, DEA does not require prior specification of a production function or assumptions about the functional form, enabling an objective evaluation of efficiency across DMUs with heterogeneous input–output structures.

Within the BCC framework, efficiency outcomes are reported in terms of technical efficiency (TE), pure technical efficiency (PTE), and scale efficiency (SE), with the relationship expressed as TE = PTE×SE. Pure technical efficiency reflects managerial and technological capability under a given scale of operations, whereas scale efficiency captures the extent to which a DMU operates at an optimal scale. A DMU is considered fully efficient only when TE equals 1, which occurs if and only if both PTE and SE equal 1. In the context of this study, a TE score of 1 indicates that a province efficiently converts its science and technology inputs into competitive sports outcomes under the prevailing technological and scale conditions.

3.2.2 Malmquist index model

The Malmquist index is a widely used productivity measure for evaluating changes in efficiency and technology over time. Based on Data Envelopment Analysis (DEA) (7, 25), it allows the evaluation of intertemporal variations in the efficiency of decision-making models (DMUs) with out requiring a predefined manufacturing operate. In this research, the Malmquist index is employed to look at dynamic adjustments in the effectivity of S&T investment in China’s competitive sports activities system throughout totally different durations.

Following the decomposition proposed by Färe et al. (26), the Malmquist whole issue productiveness (TFP) index is expressed because the product of technical effectivity change (EC) and technological change (TC). Technical effectivity change displays actions of a DMU towards or away from the manufacturing frontier beneath present technology, whereas technological change captures shifts in the frontier itself, indicating technological progress or regress. When returns to scale are variable, EC could be additional decomposed into pure technical effectivity change (PTEC) and scale effectivity change (SEC), permitting for a extra nuanced interpretation of productiveness dynamics. The components for the Malmquist mannequin (27) is as follows:

In general, a Malmquist index value greater than 1 indicates an improvement in total factor productivity, a value equal to 1 denotes no change, and a value less than 1 indicates a decline in productivity. Similarly, values of EC or TC greater than 1 signify efficiency improvement or technological progress, respectively. By applying this model, the present study identifies the relative contributions of efficiency change and technological change to the evolution of S&T investment performance in competitive sports, thereby providing insight into whether productivity dynamics are driven primarily by managerial improvements or by advances in technology.

3.2.3 Tobit model

In the second stage of the analysis, a Tobit regression model is employed to examine the determinants of efficiency in S&T investment across provinces. The use of the Tobit model is appropriate because the efficiency scores obtained from the DEA analysis are censored, taking values within the closed interval [0,1] (28). Conventional linear regression strategies could subsequently yield biased and inconsistent estimates.

Following standard practice, the DEA efficiency scores are treated as the dependent variable, while a set of explanatory variables capturing economic, institutional, and managerial conditions are included as independent variables. The Tobit model is estimated using maximum likelihood methods, allowing for consistent inference on the marginal effects of the explanatory factors on efficiency outcomes. The Tobit model is as follows:

Through this framework, the model identifies how variations in regional economic development, management input, and research resource allocation are associated with differences in S&T investment efficiency. The estimated coefficients thus provide empirical evidence on the mechanisms through which contextual factors influence the effectiveness with which S&T resources are transformed into competitive sports performance.

4 Results and discussion

4.1 Static efficiency analysis

4.1.1 Technical efficiency (TE) analysis

From 2018 to 2022, the average technical efficiency (TE) of S&T investment in competitive sports across China’s provinces was 0.40, reflecting an overall low level of efficiency with significant regional disparities (Table 2).

Provinces such as Inner Mongolia, Liaoning, Henan, Jiangxi, Guangxi, and Qinghai consistently achieved high efficiency scores, often reaching the frontier value of 1.000 in most years. In contrast, provinces including Beijing, Hebei, Shanxi, Jiangsu, Hubei, Hunan, Guizhou, Tibet, Gansu, and Ningxia showed marked fluctuations and notably low efficiency scores in certain years—for instance, recording values near or equal to 0.000 in 2021 for several of these regions.

4.1.2 Analysis of pure technical efficiency

An analysis of pure technical efficiency (PTE), which reflects managerial and technical capabilities under existing resource conditions, reveals that many provinces achieved optimal pure technical efficiency (PTE) (PTE = 1.000) in at least one year during the period (Table 3).

Provinces such as Tianjin, Liaoning, Shandong, Henan, Guangdong, and Sichuan sustained high PTE scores over multiple years. Notable fluctuations were evident in provinces including Hebei, Shanxi, Hunan, Guizhou, Yunnan, and Tibet. In contrast, Guizhou and Tibet consistently recorded lower PTE scores.

4.1.3 Analysis of scale efficiency and Status

Scale efficiency scores indicated the appropriateness of the scale of operations (Table 4). Provinces together with Tianjin, Inner Mongolia, Liaoning, Henan, Jiangxi, Hainan, Qinghai, and Ningxia exhibited fixed returns to scale (SE = 1.000) in most years.

Provinces such as Hebei, Shanxi, Jiangsu, Hubei, Hunan, Guizhou, Yunnan, Shaanxi, and Gansu showed lower or more volatile scale efficiency scores, often with values significantly below 1.000.

4.2 Dynamic efficiency analysis

4.2.1 Overall analysis

The average Total Factor Productivity Change (TFPCH) over the period was 0.914, indicating an overall decline of 8.6% in dynamic efficiency (Table 5). Technical Efficiency Change (EFFCH) confirmed important will increase in Year 2 (10.128) and Year 5 (18.556), however a sharp drop in Year 4 (0.013). Technological Change (TECHCH) had a imply worth of 0.748 (

4.2.2 Provincial decomposition

Considerable variation existed across provinces (Table 6). Provinces like Tianjin (TFPCH = 3.893), Yunnan (1.616), Liaoning (1.466), and Chongqing (1.483) exhibited TFP progress (TFPCH >1). Others, comparable to Jilin (0.133), Guangxi (0.344), and Qinghai (0.291), skilled productiveness decline. The imply TECHCH of 0.748 confirms the restricted function of technological progress on the mixture degree.

4.3 Analysis of influencing factors

4.3.1 Variable description

Five explanatory variables were selected (Table 7): GDP (per capita Gross Domestic Product), MI (Management Input ratio), RSI (Research Staff Input ratio), ICS (Importance of Competitive Sports), and ITR (Importance of Technological R&D).

4.3.2 Regression results

The Tobit regression results are presented in Table 8.

The Tobit regression results (Table 8) point out that regional financial growth (GDP) exerts a twin impact on the effectivity of S&T investment in competitive sports activities. GDP has a important optimistic impact on pure technical effectivity (PTE) (coeff. = 2.735, p = 0.006), suggesting that extra developed areas are higher at managing and using technological assets. Conversely, it has a important adverse impact on scale effectivity (SE) (coeff. = −2.983, p = 0.003), which can mirror diseconomies of scale or diminishing returns because the financial scale expands. Meanwhile, analysis employees enter (RSI) reveals a sturdy adverse affiliation with each technical effectivity (TE) (coeff. = −3.629, p p p = 0.025). In distinction, the consequences of administration enter (MI) and the significance of competitive sports activities (ICS) on the assorted effectivity measures didn’t attain statistical significance in this mannequin.

4.3.3 Robustness checks

Robustness tests, including substituting the dependent variable with Comprehensive Efficiency and performing bootstrap procedures (2,000 iterations), confirmed the stability of the key regression results. The direction and significance of the main predictors remained consistent.

5 Discussion

This study, grounded in the dual theoretical framework of the Resource-Based View (RBV) and Knowledge Conversion Theory, systematically examines the efficiency of provincial S&T investment in China’s competitive sports from 2018 to 2022. The observed pattern of generally low efficiency alongside significant regional disparities cannot be attributed merely to the scale of investment; rather, it must be understood through the lens of resource conversion capability. While RBV emphasizes the heterogeneity and inimitability of strategic resources, this research reveals that possessing abundant S&T resources—such as funding, institutions, and personnel—is only a necessary condition. The critical factor lies in whether regional systems possess the organizational capability to effectively configure and activate these resources. For instance, provinces like Inner Mongolia and Jiangxi consistently remained on the efficiency frontier in certain years, suggesting they have developed unique pathways for embedding S&T elements into their local training ecosystems. Conversely, the fluctuations and periodic declines in efficiency observed in some economically developed provinces likely reflect an imbalance between resource expansion and growing management complexity. This points to a dual dilemma of “resource redundancy” and “diminished management effectiveness”. This finding deepens the traditional resource-based view by suggesting that the value of resources lies not only in their static endowment but, more critically, in the dynamic process of “resource orchestration”.

The significant negative correlation between research staff input (RSI) and efficiency warrants particular in-depth analysis. This finding might seem counterintuitive from a mere input-output perspective but finds a coherent explanation through Knowledge Conversion Theory. This theory conceptualizes knowledge creation as a four-stage process: socialization, externalization, combination, and internalization. The current sports science system appears to suffer from a disconnect, particularly in the stages of “combination” (systematizing explicit knowledge) and “internalization” (embedding knowledge into the tacit capabilities of individuals and organizations). An increase in research personnel, if not supported by effective knowledge conversion mechanisms—such as institutionalized dialogue between researchers and practitioners, collaborative research focused on practical competitive problems, and tailored knowledge dissemination pathways for coaches and athletes—can result in an “island effect” in knowledge production. A substantial volume of research outputs remains confined to papers and reports, failing to transform into practical solutions that enhance athletic performance or optimize training processes. This directly leads to the paradox of “increased input without corresponding efficiency gains”. Therefore, this result does not negate the value of researchers; rather, it sharply highlights a core deficiency in the current S&T support system: the weak linkage between research and training.

From a methodological standpoint, this study’s integration of the DEA, Malmquist index, and Tobit models provides a workable framework for evaluating the efficiency of public S&T investment in sports, which is characterized by multiple inputs and outputs. While any model has its boundary conditions—for instance, the DEA assumption of homogeneous decision-making units is challenged by the vast socio-economic and sports-cultural disparities among Chinese provinces—this study partially mitigates the bias of attributing all heterogeneity to managerial inefficiency by introducing environmental variables like regional economic level and management structure into the second-stage Tobit regression. Subsequent research could employ methods like “meta-frontier analysis” to decompose efficiency sources more precisely while acknowledging technological differences between groups. Furthermore, the lagged nature of competitive sports outcomes means that current outputs may carry the effects of historical investments, posing a challenge for precise attribution. Future studies utilizing longer panel datasets and dynamic models could more clearly delineate the time-lag structure and long-term impact between S&T investment and athletic achievement.

Based on the above analysis and findings, this study carries clear policy implications. The primary direction is to shift the management logic of S&T investment from a “scale-oriented” approach to one focused on “efficiency” and “precise fit”. At the macro level, it is advisable to establish a differentiated resource allocation mechanism based on efficiency evaluation outcomes, guiding resources toward regions or domains with stronger configuration capabilities, and to set up cross-regional platforms for knowledge sharing and technology transfer to break down resource barriers. At the micro operational level, it is essential to construct an institutionalized ecosystem for “research-training integration”. This could involve creating dedicated roles such as “S&T Director” at key training bases, mandating that research projects be jointly proposed and implemented by research institutions and training teams, and establishing regular S&T service stations for sports teams. These measures would actively bridge the channel for knowledge to flow from the laboratory to the training ground. Finally, a national-level monitoring and evaluation system for the effectiveness of S&T in competitive sports should be established. Incorporating static efficiency scores and dynamic indicators like Total Factor Productivity change into the regular assessment system for sports development would create a continuous improvement loop of “evaluation-feedback-optimization”. This would ultimately drive the fundamental transformation of China’s S&T support system for competitive sports from extensive growth to intensive, quality-driven development.

6 Strength and limitations

This study has several strengths. Methodologically, it integrates the DEA, Malmquist index, and Tobit models, enabling a comprehensive assessment that combines static efficiency measurement, dynamic productivity change analysis, and the identification of key influencing factors. Empirically, it employs extensive and authoritative panel data (2018–2022) across all 31 Chinese provinces, providing a solid foundation for a systematic national evaluation. Theoretically, applying the Resource-Based View and Knowledge Conversion Theory offers a coherent framework for interpreting the complex mechanisms behind efficiency disparities, moving the analysis beyond mere description to diagnostic insight.

The study also has several limitations. The findings, particularly those from the Tobit regression, indicate correlations rather than established causal relationships. The DEA model’s assumption of homogeneous decision-making units simplifies the substantial contextual heterogeneity among provinces, even though a two-stage approach was used to mitigate this. Furthermore, inherent time-lags exist between S&T investment and measurable sports outputs (e.g., world champions), meaning the analysis captures the efficiency of the current support system rather than fully isolating the long-term causal impact of investment. Data constraints, including reliance on aggregated statistical yearbooks and the potential for measurement error, also pose limitations. Future research should employ methods such as meta-frontier DEA, distributed lag models, and mixed-methods case studies to better address heterogeneity, temporal dynamics, and causal mechanisms.

7 Conclusion

This study demonstrates that the efficiency of S&T investment in China’s competitive sports system depends less on resource endowment than on regional capabilities for strategic resource orchestration and conversion into athletic performance. Grounded in the Resource-Based View and Knowledge Conversion Theory, our DEA-Malmquist-Tobit analysis of provincial panel data (2018–2022) reveals persistently low static efficiency (mean = 0.40) with pronounced regional disparities, while dynamic total factor productivity exhibits modest improvements offset by periodic volatility. Notably, research staff input demonstrates a significant negative association with efficiency, underscoring a systemic “knowledge translation gap” between scientific research and training practice. Regional economic development enhances pure technical efficiency but erodes scale efficiency, exposing inherent tensions between resource availability and managerial complexity. These findings collectively advocate for a policy paradigm shift from quantitative investment expansion toward optimizing strategic resource orchestration, contextual tailoring, and practical application of S&T resources to cultivate a more effective and sustainable ecosystem for competitive sports development in China.

References

• WuD-YTanT-CWangY-T. A new shortcut for competitive sports development? The purpose of and strategy for developing and introducing Sepaktakraw in Taiwan. PLoS One. (2025) 20(1):e0317451. 10.1371/journal.pone.0317451

• RobazzaCVitaliFBortoliLRuizMC. Basic psychological needs satisfaction, coping functions, and emotional experiences in competitive athletes: a multi-states theory perspective. Sci Rep. (2025) 15(1):1854. 10.1038/s41598-025-86072-1

• CharnesACooperWWRhodesE. Measuring the efficiency of decision making units. Eur J Oper Res. (1978) 2(6):429–44. 10.1016/0377-2217(78)90138-8

• WeiCYuanL. A study on the technological innovation efficiency of China’s high-tech industries based on three-stage DEA and Malmquist index. Oper Res. (2025) 25(1):14. 10.1007/s12351-024-00894-x

• DouXWangMChenY. A study of provincial renewable portfolio standards targets allocation in China based on fixed-sum dea model. Energy Sources B Econ Plan Policy. (2025) 20(1):2439404. 10.1080/15567249.2024.2439404

• YenBTHHuangM-JLaiH-JChoH-HHuangY-L. How smart port design influences port efficiency—a DEA-Tobit approach. Res Transp Bus Manag. (2023) 46:100862. 10.1016/j.rtbm.2022.100862

• ChenZZhuWFengHLuoH. Changes in corporate social responsibility efficiency in Chinese food industry brought by COVID-19 pandemic-a study with the super-efficiency DEA-Malmquist-Tobit model. Front Public Health. (2022) 10:875030. 10.3389/fpubh.2022.875030

• YangMZhouHLiYZhangJ. Efficiency evaluation and influencing factors of sports industry and tourism industry convergence based on China’s provincial data. Sustainability. (2023) 15(6):5408. 10.3390/su15065408

• LvCWangYJinC. The possibility of sports industry business model innovation based on blockchain technology: evaluation of the innovation efficiency of listed sports companies. PLoS One. (2022) 17(1):e0262035. 10.1371/journal.pone.0262035

• ZhangKGuoBYangMJiaYZhangKWangL. The assessment of sports performance by grip pressure using flexible piezoresistive pressure sensors in seven sports events. Sci Rep. (2024) 14(1):31750. 10.1038/s41598-024-82274-1

• AssalveGLunettiPDi CagnoADe LucaEWAldegheriSZaraVet alAdvanced wearable devices for monitoring sweat biochemical markers in athletic performance: a comprehensive review. Biosensors-Basel. (2024) 14(12):574. 10.3390/bios14120574

• RenPLiuZ. Efficiency evaluation of China’s public sports services: a three-stage DEA model. Int J Environ Res Public Health. (2021) 18(20):10597. 10.3390/ijerph182010597

• ZhangSOuyangTWeiS-YLiangB-BZhuJ-M. Research on sustainable development of competitive sports in China based on PSR and DEA model. Front Psychol. (2022) 13:925909. 10.3389/fpsyg.2022.925909

• YeJGuoGYuKLuY. Allocation efficiency of public sports resources based on the DEA model in the top 100 economic counties of China in Zhejiang province. Sustainability. (2023) 15(12):9585. 10.3390/su15129585

• WangLZhouZZouQ. Analysis system for table tennis techniques and tactics using data mining. Soft Comput. (2023) 27(19):14269–84. 10.1007/s00500-023-09082-z

• MengYWuY. Track and field competition data collection using a vision sensor system and machine learning. IET Softw. (2023) 17(4):639–48. 10.1049/sfw2.12086

• EnoiuR-SGainariuIMindrescuV. Implementing modern technology for vital sign monitoring to enhance athletic training and sports performance. Sustainability. (2023) 15(3):2520. 10.3390/su15032520

• WangXGuoY. The deep learning-based human action recognition system for competitive sports. J Imaging Sci Technol. (2024) 68(3):030405. 10.2352/J.ImagingSci.Technol.2024.68.3.030405

• OuyangYLeeP-CKoL-M. A systematic review of the development of sport policy research (2000–2020). Sustainability. (2023) 15(1):389. 10.3390/su15010389

• ViolletBScellesNPengQ. From sport policy to national federation sport policy: an integrative literature review and conceptualisation attempt. Sustainability. (2023) 15(4):2949. 10.3390/su15042949

• ShinW-SOhA. Policy recommendations for improving elite sports performance for athletes with disabilities. Korean Soc Sport Policy. (2022) 20(1):157–75. 10.52427/KSSP.20.1.10

• KimS-MHwangW-S이경재. Analysis of the impacts of public R&D investment on stimulating science and technology development in Korea. Product Rev. (2015) 29(4):3–27. 10.15843/kpapr.29.4.201512.3

• ChenZZhaoN. Research on the effect of fiscal decentralization on science and technology input in China. Agro Food Ind Hi-Tech. (2017) 28(3):3014–9.

• BankerRDCharnesACooperWW. Some models for estimating technical and scale inefficiencies in data envelopment analysis. Manag Sci. (1984) 30(9):1078–92. 10.2307/2631725

• PanW-TZhuangM-EZhouY-YYangJ-J. Research on sustainable development and efficiency of China’s E-agriculture based on a data envelopment analysis-Malmquist model. Technol Forecast Soc Change. (2021) 162:120298. 10.1016/j.techfore.2020.120298

• FäreRGrosskopfS. Intertemporal production frontiers: with dynamic DEA. J Oper Res Soc. (1997) 48(6):656. 10.1057/palgrave.jors.2600779

• MalmquistS. Index numbers and indifference surfaces. Trab Estad. (1953) 4(2):209–42. 10.1007/BF03006863

• GuoYLiZLiuPWuY. Modeling correlation and heterogeneity in crash rates by collision types using full Bayesian random parameters multivariate Tobit model. Accid Anal Prev. (2019) 128:164–74. 10.1016/j.aap.2019.04.013