Super-Supportive Corporate Social Responsibility Behaviors in China’s Construction Enterprises

Abstract

Super-supportive CSR behaviors (SSCBs) are integrative actions devised to enhance the effectiveness of CSR initiatives by harmonizing social, environmental, and economic efforts. Despite their strategic role in business operations, SSCBs remain insufficiently addressed, especially within the construction sector. This study utilizes text mining and association rule mining to analyze 211 CSR reports from Chinese construction firms spanning 2010 to 2021. The key findings highlight the pivotal role of 17 SSCBs in strengthening CSR initiatives, revealing three major characteristics: foundational, synergistic, and triggering. Within the construction industry, SSCBs primarily focus on corporate governance, community development, employee welfare, and environmental sustainability, evolving from isolated practices to integrated systems over time. Notably, construction firms tend to adopt SSCB portfolios instead of standalone initiatives. Furthermore, exceeding a certain threshold of SSCBs may increase challenges in coordination and resource allocation. These insights highlight SSCBs as a dynamic, multidimensional construct and provide construction firms with a practical framework to integrate complementary CSR actions, improving coordination, optimizing resources, and strengthening sustainability outcomes in practice.

1. Introduction

In recent years, growing pressures to adopt eco-practices and enhance employee well-being have led businesses to increasingly embrace corporate social responsibility (CSR) as a pathway to comprehensive competitiveness [1,2]. As firms operate ever more interdependently, CSR has expanded beyond traditional philanthropy to encompass a wide range of economic, social, environmental, and stakeholder-related concerns [3,4]. The potential benefits of CSR engagement are considerable, motivating firms to acquire critical resources such as human capital, technology, and marketing channels. According to the International Organization for Standardization (ISO) 2024 survey, more than 4.91 million sites across 172 countries are now certified, with over 2.96 million certificates issued [5]. Meanwhile, the Chairman of Deloitte China stated at the “2024 ESG (Environmental, social, governance) Global Leaders’ Summit” that global ESG investment assets will reach USD 53 trillion by 2025, accounting for one-third of the total global assets under management.

The primary goal of CSR behaviors is to cultivate harmonious relationships between employees and their families, local communities, and business partners [6]. Shaped by resources like human capital and financial assets [7], effective CSR requires strategic resource allocation to balance financial constraints, governance structures, and long-term goals [8]. Under resource limitations, firms should prioritize key initiatives, such as employee well-being and environmental sustainability to maximize CSR’s societal and environmental impact. Beyond this, CSR demands transparent and ethical commitments to stakeholder value creation [9], yet conventional CSR often relies on fragmented initiatives that either limit impact or cause adverse effects. This critical gap between the need for impactful CSR and the limitations of existing practices highlights the necessity for super-supportive CSR behaviors (SSCBs).

Defined as targeted complementary actions to proactively amplify the causal mechanisms of core CSR initiatives, SSCBs differ from “integrated CSR”, “synergistic CSR”, and “complementary CSR”, because these constructs all focus on outcomes rather than behaviors [10]. ”Integrated CSR” aligns cross-domain goals, e.g., merging environmental and social goals [11]; “synergistic CSR” highlights combined efficiency gains; and “complementary CSR” highlights additive benefits [12]. None of these constructs, however, specify what concrete actions firms should take to strengthen their core CSR efforts. In contrast, SSCBs are behavior-focused: they provide actionable steps to enhance impact. For example, pairing green construction with community engagement programs to recruit local “green advocates” amplifies reach, addressing a gap that effect-focused CSR frameworks cannot fill.

While SSCBs lack explicit prior labeling, their significance is implicit in studies advocating for non-isolated CSR practices. Scholars stress that effective CSR requires not only integration but also reinforcement to meet stakeholder expectations [13,14] and cross-functional coordination for sustainability [11,15]. Prior research has examined CSR dimensions such as governance, environmental management, employee well-being, and community engagement, showing the value of interconnected approaches [14,16,17]. However, these studies fail to explain how to design supportive actions that make such integration effective. SSCBs fill this gap by shifting the focus from “which domains to combine” to “how to amplify core impact”, thereby offering a unique theoretical contribution.

This gap is particularly pronounced in the Chinese construction sector, which faces unique challenges due to its complex supply chains, wide-ranging stakeholders, and large social–environmental footprint [11,18,19]. Although construction firms have increasingly adopted CSR strategies covering governance, environmental sustainability, and worker welfare [20,21], many remain trapped in fragmented initiatives that limit overall impact [22,23]. Integrated or complementary CSR frameworks offer little guidance here; they prioritize theoretical alignment over actionable steps, failing to address the sector’s need for resource-efficient, impact-driven solutions. SSCBs, by contrast, are tailored to this context: for instance, using community engagement to boost green construction participation. Though complementary CSR benefits construction firms [19,24], SSCB-driven reinforcement remains unstudied.

To address these theoretical and practical gaps, this study introduces the concept of “super-supportive CSR behaviors (SSCBs),” a novel construct that differs fundamentally from established CSR frameworks like complementary and integrated CSR. And it answers two urgent questions: “What SSCBs are being implemented by Chinese construction firms?” and “What are the relational patterns among these SSCBs?” This focus is timely: amid ongoing volatility in the construction industry, firms need actionable, resource-efficient CSR tools to mitigate risks, optimize resource use, and foster collaboration, and these needs have not yet been met by generic integrated CSR frameworks, which emphasize theoretical alignment over practical action. By defining SSCBs and mapping their application in China’s construction sector, this study advances CSR theory beyond effect-focused frameworks and provides firms with an actionable roadmap for impactful CSR, two distinct contributions that distinguish it from prior work.

2. Literature Review and Theoretical Foundation

2.1. Rethinking CSR Constructs

Over the past decade, CSR has evolved into a prominent paradigm in enterprise management [25]. Ref. [26] introduced the foundational concept of CSR by posing the question, “What responsibilities to society may they reasonably be expected to assume?” This question has sparked ongoing debate [27]. One of the most influential answers is Carroll’s four-level pyramid model, which categorizes CSR as comprising society’s economic, legal, ethical, and discretionary responsibilities [28]. Grounded in stakeholder and ethical theories, this model aligns with stakeholder theory’s focus on addressing the interests of primary (e.g., shareholders for economic responsibility) and secondary (e.g., communities for discretionary responsibility) stakeholders, while its ethical and legal dimensions reflect ethical theory’s emphasis on moral and regulatory obligations. Researchers have highlighted the diverse constructs of CSR and concur that CSR initiatives must be strategically integrated within value chains to yield actionable outcomes [29].

The World Business Council for Sustainable Development deepens CSR by emphasizing corporate commitment to key stakeholders, including shareholders, employees, local communities, and the environment [6]. This framing echoes stakeholder theory (ST). Rooted also in ethical theory, this conceptualization suggests businesses balance profit with social and environmental responsibilities [30,31]. In response to external pressures such as regulatory changes and societal expectations, enterprises are supposed to secure tangible benefits through CSR fulfillment [1]. Scholars note that embedding CSR into corporate missions, which aligns with ethical duty and accountability, boosts stakeholder engagement [32]. While debates over CSR scope persist, mainstream perspectives grounded in both theories advocate CSR strategy integration to drive corporate success and social welfare promotion [28,33], enabling sustainable development and ethical practices [8].

The current research on corporate social responsibility (CSR) in the construction industry has become a hotspot. In China’s construction sector, CSR efforts are centered on corporate governance, environmental management, occupational health and safety, economic responsibility, and community development [18]. However, these efforts failed to address the integration of CSR with broader business development goals. While international companies tend to align CSR initiatives with long-term strategies, many domestic construction firms regard CSR as a regulatory obligation rather than a strategic priority [23]. For instance, [24] observed that Chinese international contractors prioritize construction quality and safety over broader sustainability concerns. Reference [21] found that community development initiatives are often secondary to quality management and customer service. Despite some progress in areas like communication, biodiversity, and staff training [19], construction firms still face stakeholder conflicts as well as ethical and environmental violations [34].

2.2. Definitions of Super-Supportive CSR Behaviors

As outlined by [35], SSCBs are traceable in terms of competitive strategies, policies, and procedures. However, providing a definition of SSCB is crucial for advancing the discussion. Specifically, SSCBs expand the scope, improve the efficiency, and amplify the societal impact of core CSR initiatives to pursue long-term business resilience amid competition. Unlike traditional CSR practices that focus on outcomes, SSCBs are integrated behaviorally to strengthen core CSR mechanisms. This approach goes beyond isolated or compliance-driven CSR practices, helping firms to enhance sustainability, enhance their competitive positioning, and secure long-term benefits.

Some SSCBs are illustrated in Figure 1, which aligns with the above definition. In Figure 1a, an individual SSCB signifies the basic CSR behavior (denoted as C) paired with complementary actions (denoted as A or B), forming high-frequency complementary pairs. These pairings allow CSR behaviors to be reinforced by accompanying actions, thereby enhancing their overall impact. As shown in Figure 1c, complementary pairs, such as C-A and C-B, can further evolve into synergistic networks, indicated as AB-C and C-AB, respectively. These networks promote cross-functional collaborations and facilitate the implementation of multiple cohesive actions. This extended model demonstrates how individual CSR behaviors and actions can be integrated into a robust CSR strategy.

Both individual and combined SSCBs can thrive in real-world scenarios. As shown in Figure 1c, the foundational and compound structures represent the hybrid form of SSCBs. These hybrid models encourage firms to balance straightforward and effective actions with complex strategies. In line with organizational complexity theory, as outlined by [36], such hybrid CSR models emphasize the importance of balancing simplicity and complexity in strategic decision-making, particularly when firms implement a mix of basic and advanced CSR activities. Prior studies [23] emphasize the need for stronger CSR integration in the construction sector and point to the widespread challenge of limited CSR awareness. Building on this, SSCBs add novelty by framing integration not merely as a combination of CSR dimensions but as a deliberate process of sequencing and reinforcing actions, thereby generating synergies that extend beyond conventional CSR practices.

2.3. Attributes of SSCBs

SSCBs are characterized by their capacity to go beyond basic CSR efforts, creating substantial impacts within organizations and the broader community [37]. Building on three core attributes, foundational, synergistic, and triggering, SSCBs enable firms to engage in sustainable and long-term CSR practices [38,39]. These attributes are not abstract but explicitly anchored in three foundational theories, resource-based view (RBV), stakeholder theory (ST), and institutional theory, each explaining a distinct dimension of SSCBs and their strategic value.

The foundation of SSCBs emphasizes the strategic integration of internal and external resources, which establish trust and accountability, directly aligning with RBV. RBV posits that firms gain competitive advantage via combinations of valuable, rare, and inimitable resources, and SSCBs operationalize this logic by reconfiguring both tangible and intangible resources to strengthen CSR outcomes [40]. In practice, this means that SSCBs go beyond compliance-driven initiatives by creating resource bundles that are hard to imitate. For example, in construction, Skanska’s integration of BIM technologies with sustainable material supply chains creates resource bundles that competitors cannot easily replicate. By optimizing resource allocation across CSR dimensions, the foundational attribute of SSCBs amplifies the impacts of core initiatives and addresses resource constraints [41,42], directly translating RBV into actionable CSR behavior.

From a synergistic perspective, SSCBs amplify the impact of interconnected initiatives, an idea grounded in stakeholder theory [30]. Unlike traditional CSR programs that address stakeholder demands as discrete and sequential, SSCBs’ synergistic attribute recognizes stakeholder interdependence and aligns initiatives to generate multiplier effects exceeding the sum of individual effects [1]. This is evident in cases like the “Living Places Copenhagen” project, which demonstrates how energy-efficient construction simultaneously satisfies investor ESG requirements and SDGs, improves community livability, and ensures regulatory compliance. Here, synergies among SSCB actions generate virtuous cycles where individual CSR investments yield disproportionate returns across multiple stakeholder groups. This aligns with stakeholder theory’s focus on balancing multi-group interests, as SSCBs’ synergies cultivate stakeholder loyalty and partnerships while advancing CSR objectives [7,43].

Triggerability is another characteristic of SSCBs that encourages businesses to respond to external pressures such as regulatory changes, market competition, and societal expectations. Rooted in institutional theory, triggerability highlights how organizations tailor their behaviors in line with external forces, including laws, social norms, and industry standards [44]. SSCBs operationalize this by turning external constraints into catalysts for proactive adaptation. Unlike static CSR responses, SSCBs use triggerability to transform external shocks or pressures into strategic opportunities [45]. For instance, ACS Group’s post-Grenfell fire safety reforms demonstrate how SSCBs can transform coercive regulatory pressures into industry-leading practices that redefine market standards. Triggerability thus extends institutional theory by illustrating how firms not only conform to institutional pressures but also reshape them, using SSCBs as mechanisms for organizational learning, field-level innovation, and legitimacy building. In this sense, SSCBs embody a form of “institutional entrepreneurship,” where compliance evolves into leadership, and firms move from passive adaptation to active transformation.

These three attributes of SSCBs, each tied to a core theory, align with the broader CSR literature emphasizing complementary strategy integration to maximize value. As illustrated in Figure 2, the evolution of CSR behaviors revolves around leveraging resources (RBV), fostering synergy (ST), and responding to external pressures (institutional theory). Integrating multiple theoretical frameworks is essential to encapsulating the multifaceted nature of SSCBs and their contribution to sustainable and strategic CSR practices.

3. Research Methods

A mixed-methods approach was employed to investigate the relationships among super-supportive CSR behaviors (SSCBs) through four key steps. The overall research process is illustrated in Figure 3. First, 211 CSR reports from the top 100 Chinese construction firms were collected, recognizing their value in analyzing CSR behaviors [46]. Second, 83 CSR behaviors were preliminarily identified through a literature review, the application of CSR tools, and content analysis. Third, text mining techniques were used to quantify these behaviors within the collected reports. Finally, association rule mining was applied to examine the relationships among SSCBs, uncovering patterns and interconnections.

3.1. Samples

In China, all listed enterprises must release CSR reports, which provide valuable resources for sample collection. As performed by [18], we utilized the list of the “Top 100 Comprehensive Strength Firms of China’s Construction Industry” published by the China Construction Industry Association https://www.chyxx.com/top/202110/977571.html (accessed on 6 October 2021). CSR reports were collected from various sources including company websites and stock exchanges. The collected materials encompass CSR; sustainability; and environmental, social, and governance (ESG) reports from 2010 to 2021. A total of 211 CSR reports were collected, comprising 187 reports from listed firms and 24 from non-listed firms.

3.2. CSR Behavior Identification

Following the approach outlined by [47], this study identified the CSR behaviors of construction enterprises by integrating three methods: a systematic literature review, analysis of CSR tools, and content analysis of typical CSR reports. First, a systematic literature review (SLR) was conducted. In March 2024, the Web of Science and Scopus databases were searched using the Boolean query (“corporate social responsibility” OR “CSR”) AND (“construction” OR “contractors” OR “project”). After reviewing titles and abstracts, 130 relevant publications were initially identified, and subsequent filtering for relevance and the removal of duplicates narrowed the selection down to 39 key publications (see Table S1 in Supplementary Materials).

Second, an indicator system for CSR tools was constructed by drawing on the standards, principles, initiatives, and guidelines referenced in the CSR reports of Chinese construction contractors, as well as the CSR tools mentioned in the aforementioned 39 CSR-related publications on contractors. This system included 13 CSR tools covering both international and domestic categories, such as the Global Reporting Initiative (GRI) G4 [48], ISO 26000 [9], and the Guidelines for the Preparation of Social Responsibility Reports (GB/T 36001-2015) [49] (see Table S2 in Supplementary Materials).

Finally, content analysis was performed on the reports of typical Chinese construction enterprises. Considering this study’s focus on China’s construction industry, the top five Chinese international contractors from the 2020 ENR TOP 250 International Contractors list were selected as case companies. Three professionals in related fields conducted content analysis on these five reports, manually documenting the CSR practices identified by each member and using a binary coding method. Through this coding, 1167 pieces of specific CSR activity information were obtained, with a total of 821 CSR activities coded. Additionally, Cohen’s kappa coefficient was calculated. A higher value of this coefficient indicates greater consistency, and the calculation yielded an average value of 0.936. This result demonstrates strong consistency among the research team members and confirms the reliability of the identified CSR behaviors.

3.3. Text Mining

A text mining approach was used to analyze 211 CSR reports through a structured process, as shown in Figure 4. After collecting the reports, the specific text mining process proceeded as follows:

(1) Text preprocessing was conducted, which included cleaning the reports, performing Jieba Chinese word segmentation on the corpus, and removing stop words. For stop words, this study selected a comprehensive and commonly used domestic Chinese stop word list (containing 1208 stop words) to ensure the effectiveness of stop word filtering in subsequent text analysis.

(2) Corpus selection and preprocessing were implemented. Specifically, open-source online tools were used to extract text data from Wikipedia, followed by Jieba Chinese word segmentation and stop word removal on the corpus. To capture semantic relationships among words, the Skip-gram model was then adopted, following the word2vec approach [50]. The gensim.models.word2vec package was installed and utilized to train themodel on the preprocessed corpus. This preprocessed corpus was constructed based on the CSR behaviors identified in the previous step. The trained model was saved for subsequent analysis. Next, a keyword corpus was constructed. Drawing on machine learning-based textual analysis approaches that have been increasingly applied in CSR studies [51], cosine similarity was calculated via the most_similar method, and the top 20 words most similar to the initial core keywords were identified. The research team further discussed, screened, and reviewed the constructed keyword list to ultimately confirm a CSR keyword corpus containing 439 terms.

(3) The collected CSR reports were quantified using a 0–1 matrix with reference to the CSR keywords. (4) The reliability of the entire text mining process was verified.

Statistical testing was performed to confirm the reliability of the results. The following equation was used:

$$\mathrm{n}={\sigma }^2/(e^2/{z_{\alpha }}^2+{\sigma }^2/N)$$

(1)

where ${\sigma }^2$, $e^2$,$z_{\alpha }$, and $N$ are the variance, standard deviation, confidence level, and sample size, respectively. For a 90% confidence level (±10% margin of error) with σ = 0.5, e = 0.1, and $z_{\sigma }$ = 1.645, the required sample size was 52. These were randomly selected for manual validation. Using confusion matrix metrics (TP = 1622; FP = 46; FN = 421; TN = 2227), we calculated reliable values for text mining [52], as shown in

Table 1. Reliability of text mining.

Reliability Index	Equation	Validation Results
Accuracy	$(\mathit{TP}+\mathit{TN})/\mathit{All} \mathit{sample}$	97.24%
Precision	$\mathit{TP}/(\mathit{TP}+\mathit{FP})$	92.75%
Recall	$\mathit{TP}/(\mathit{TP}+\mathit{FN})$	79.34%
F1-score	$2 \times recision \times Recall/(Precision+Recall)$	0.874

.

The frequency of 83 CSR behaviors was calculated using the following equation:

$$Frequency {CSR}_i={No. CSR}_i/211$$

(2)

where ${CSR}_i$ denotes type $i$ (max $i$ = 83) CSR behavior. The frequency is then divided from (0.0, 1.0) into 10 equal parts; that is, 1 = [0.0, 0.1), 2 = [0.1, 0.2), …, 10 = [0.9, 1.0].

3.4. Association Rule Mining

ARM is intended to identify frequent co-occurrence patterns among CSR behaviors, thereby revealing their underlying relationships [53]. Let $\mathrm{I}=\{i_1,i_2,\dots ,i_n\}$ denote the set of n distinct CSR behaviors, ${\mathrm{n}}_{max}=\sum N_{CSR-A}$; $\mathrm{D}$ represent the CSR report database ($m\times n$ matrix); and $\mathrm{A}\subseteq \mathrm{I}$ be an itemset of co-occurring CSR behaviors. The support of itemset A can be defined as follows:

$$\mathrm{s}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\left(\mathrm{A}\right)=\left|\mathrm{R}:\mathrm{A}\subseteq \mathrm{R},\mathrm{R}\subseteq \mathrm{D}\right|/\left|D\right|=A.count/\left|D\right|$$

(3)

Itemset $A$ is considered a frequent itemset (FIS) if $support \left(A\right)\ge min\_support$, where $min\_support$ is a predefined threshold. For an association rule $\mathrm{A}\to \mathrm{B}$, where $\mathrm{A},\mathrm{B}\in \mathrm{I},\mathrm{A}\cap \mathrm{B}=\mathrm{\varnothing }$, $A$ represents the antecedent (left-hand side, LHS), and $B$ is the consequent (right-hand side, RHS). The strength of the association was measured using the following standard ARM metrics: support, confidence, and lift. In this context, SSCBs refer to the CSR that meets the ARM threshold; that is, CSR behaviors that demonstrate synergistic integration beyond conventional practices. The equations used were as follows:

$$\mathrm{s}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\left(\mathrm{A}\to \mathrm{B}\right)=\mathrm{s}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\left(\mathrm{A}\cup \mathrm{B}\right)=P\left(\mathrm{A}\cup \mathrm{B}\right)=\left(\mathrm{A}\cup \mathrm{B}\right). count/\left|D\right|,\hspace{1em}\hspace{1em}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{e}\left[\mathrm{0,1}\right]$$

(4)

$$\begin{array}{r}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{f}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\left(\mathrm{A}\to \mathrm{B}\right)=\mathrm{P}\left(\mathrm{A}|\mathrm{B}\right)=\mathrm{s}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\left(\mathrm{A}\to \mathrm{B}\right)/\mathrm{s}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\left(\mathrm{A}\right)\\ =\left(\mathrm{A}\cup \mathrm{B}\right).count/A.count,\hspace{1em}\hspace{1em}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{e}\left[\mathrm{0,1}\right]\end{array}$$

(5)

$$\mathrm{l}\mathrm{i}\mathrm{f}\mathrm{t}\left(\mathrm{A}\to \mathrm{B}\right)=\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{f}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\left(\mathrm{A}\to \mathrm{B}\right)/\mathrm{s}\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\left(\mathrm{B}\right),\hspace{1em}\hspace{1em}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{e}[0,\mathrm{\infty }]$$

(6)

The Apriori algorithm is a bottom-up algorithm that uses candidate generation to establish frequent itemsets. It is more suitable for small datasets than the Eclat and FP-Growth algorithms for ARM [53,54]. Given the frequent itemset $F_k \mathrm{f}\mathrm{o}\mathrm{r}\mathrm{s}\mathrm{i}\mathrm{z}\mathrm{e}k$, and its candidate $C_k$, ${\mathrm{C}}_{k+1}$ is generated from frequent itemsets of size k, and frequent itemsets of size k + 1 are candidates. We added items meeting the mini-support requirements to $F_{k+1}$ to yield a new item, as shown in Figure 5.

4. Results of Data Analysis

4.1. Identifying Main CSR Behaviors in Construction

This study identified 83 distinct CSR behaviors, including developing CSR plans, enhancing corporate governance structures, applying supply chain management, implementing employee career management, assessing public and community needs, and conducting environmental training. The identified CSR behaviors are further categorized into seven dimensions, as outlined in

Table 2. A list of CSR behaviors in the construction industry.

Dimensions	Definitions	No. of Behaviors	References
Management of Responsibility (R)	Management of responsibility (R) is the management of corporate social responsibility itself, including planning, formulating, and implementing social responsibility activities, establishing and organizing relevant departments, etc. [55].	11	[11,18,23,56,57]
Corporate Governance (G)	Corporate governance (G) integrates CSR behavior into a firm’s day-to-day operations and management. It is the responsibility of the entire firm to determine the wide range of uses for which organizational resources will be deployed to the firm [58].	8	[11,18,56,59,60]
Safe Construction and Quality (Q)	Safe construction and quality (Q) are the basic requirements of construction firms, including maintaining and improving the safety and quality of construction, reflecting the ability of construction units to safely deliver high-quality products [13].	15	[11,13,56,61]
Good Partnership (P)	Good partnership (P) mainly involves responsibilities toward suppliers, competitors, and other associations and alliances. A responsible construction firm should be able to develop a better relationship with its supply chain and partnerships [62].	7	[14,60,63,64,65]
Workers’ Interest (W)	Workers’ interest (W) is a responsibility concerning employees. It includes management issues such as employment policy, employment relations, workplace safety, and human development and training [66].	14	[52,60,63,67]
Well-being of Local Community (C)	The well-being of the local community (C) consists of community engagement and development. Construction activities directly affect local communities and face issues such as resettlement, demolition, and construction [67].	14	[13,24,52,60,65,68]
Environment Preservation (E)	Environment preservation (E) refers to the attention paid to and the protection of the environment throughout the entire life cycle of firms, including corporate offices, construction activities, and the post-operation and maintenance of buildings [69].	14	[23,47,56,61]

. A detailed list of these behaviors, along with their relevant references, is provided in Table S2 in the Supplementary Materials, which forms the analytical framework (or codebook) for subsequent investigation.

Figure 6a presents the distribution of CSR behavior frequencies. A clear bimodal pattern emerges, with nearly equal proportions of behaviors below and above the 0.5 threshold (48.98% vs. 51.02%). While a small share (2.41%) of behaviors achieved near-universal adoption, a notable proportion (14.46%) remained marginally implemented. Category-level analysis further reveals distinct patterns, as shown in Figure 6b. Governance- and workplace-related behaviors are more frequently implemented (65–75% ≥ 0.5), whereas responsibility-related behaviors are less adopted (70% < 0.5). Other categories show more balanced distributions around the threshold.

4.2. Examining Super-Supportive CSR Behaviors

The identification of SSCBs among all 83 CSR activities was based on the ARM threshold. Due to the lack of a standardized approach for setting the minimum confidence, threshold values were adjusted iteratively until the rules were deemed meaningful [70]. Specifically, we varied min_support from 0.5 to 0.9 and min_confidence from 0.1 to 1.0. For each combination, the Apriori algorithm was applied, and a looped procedure recorded the number of rules generated under each threshold (see Figure S1 in the Supplementary Materials).

Initial calculations revealed that min_support ≤ 0.5 produced over 150,000 redundant rules, which were uninformative for identifying valuable patterns. Therefore, to identify the rules where min_support > 0.5, we found the following:

• min_support = 0.6 and min_confidence = 0.8/0.9 generated 47,355 and 22,829 rules, respectively.
• min_support = 0.7 and min_confidence = 0.8/0.9 generated 2052 and 1138 rules.
• min_support = 0.8 and min_confidence = 0.8/0.9 generated 126 and 105 rules.

Considering both research relevance and manageability, min_support = 0.7 and min_confidence = 0.9 were selected, yielding 18 frequent CSR behaviors and 1138 association rules.

To assess the overall structure of CSR relationships, an itemset matrix was calculated to evaluate the density of frequent itemsets. In this context, a matrix is considered sparse if most cells are zero and dense if most elements are non-zero. Unlike traditional market basket analyses, the CSR activity matrix captures both the presence and co-occurrence frequencies of CSR actions. In this study, frequent itemsets covered nearly half of the matrix cells, with the following parameters: 211 transactions, 83 items, 17,513 total cells, and 7492 non-zero entries. Accordingly, the matrix density was approximately 42.78% (Figure S2, Supplementary Materials), indicating a sufficiently dense dataset for reliable SSCB analysis.

Thus, there are 18 frequent itemsets (support ≥ 0.7) screened in line with the above rules, and they are visualized in a CSR word cloud (Figure 7). Information disclosure (G05, support = 0.976) and managers’ participation in CSR (R03, support = 0.900) are the most frequently observed. Other SSCBs above the threshold are concentrated in corporate governance, local community development, employee welfare, and environmental sustainability initiatives, such as legal management (G03) and the application of recycling systems, such as solar and green wind (E04). These results highlight the emphasis on governance and environmental practices within Chinese construction firms, alongside a growing commitment to employee- and community-oriented initiatives.

Next, we selected rules with repeated items and abandoned those with low confidence by adhering to a lift value less than 1. Consequently, 1053 association rules were derived, as shown in Figure 8.

4.3. Exploring SSCBs’ Relationships

Filtered association rules identified combinations of SSCBs with itemset lengths ranging from 2 to 6.

Table 3. Distribution of SSCBs across antecedents (LHS) and consequents (RHS) by itemset length.

Length of Itemset	-2-		-3-		-4-		-5-		-6-		Total No of Rules
No. of Rules	60		280		493		192		288
Position	LHS	RHS	LHS	RHS	LHS	RHS	LHS	RHS	LHS	RHS
G05	0	15	46	72	131	165	92	86	14	14	635
R03	1	10	46	57	147	136	101	78	18	10	604
G03	2	9	41	43	112	78	53	29	19	9	395
E04	2	10	51	58	149	120	90	60	20	8	568
G01	4	8	46	56	154	136	90	64	18	10	586
G08	5	6	48	46	149	107	77	31	21	7	497
E05	2	1	28	4	91	11	22	2	0	0	161
W04	6	1	47	5	106	3	54	2	0	0	224
C09	4	0	8	0	0	0	0	0	0	0	12
Q03	6	0	44	0	100	1	14	0	0	0	165
W13	3	0	3	0	0	0	0	0	0	0	6
C11	7	0	48	1	69	0	15	0	0	0	140
W06	3	0	0	0	0	0	0	0	0	0	3
C13	2	0	0	0	0	0	0	0	0	0	2
W11	3	0	6	0	0	0	0	0	0	0	9
P03	6	0	27	0	7	0	0	0	0	0	40
Q01	4	0	9	0	0	0	0	0	0	0	13
C08	0	0	0	0	0	0	0	0	0	0	0
Total No.	60	60	498	342	1215	757	608	352	110	58	/

Note: Min_support is 0.7, min_confidence is 0.9, and lift > 1. LHS stands for left-hand side, which refers to antecedent; RHS stands for right-hand side, which refers to consequent.

presents their key distribution characteristics, including the total number of rules generated at each length, the average support values, and the number of distinct SSCBs appearing on the antecedent (LHS) and consequent (RHS) sides of the rules. The results show that SSCBs with higher support levels tend to align with complementary or reinforcing CSR behaviors. For instance, governance-oriented activities such as G05 and responsibility practices like R03 frequently appear on both the LHS and RHS, highlighting their central role in catalyzing and sustaining CSR linkages. Conversely, activities such as W13 and W06 display low frequencies, reflecting weaker associations with other actions.

Notably, itemsets of length 4 dominate the distribution (28 rules, accounting for 42% of all filtered rules; average support 0.32). This suggests that combinations of four SSCBs strike an effective balance between feasibility and depth of impact. They are resource-manageable while still being capable of generating synergistic outcomes. By contrast, shorter itemsets lack sufficient synergy, and longer itemsets (5–6 SSCBs) tend to be overly specific and resource-intensive. Additionally, certain activities demonstrate distinct positional roles. For instance, C09 and W13 appear exclusively on the LHS, confirming their role as trigger antecedents. Meanwhile, although activity C08 met the minimum support threshold, it was excluded from the set of SSCBs because it failed to form valid associations (lift < 1) with other CSR behaviors after applying ARM criteria. In line with the definition of SSCBs, such isolated activities cannot be considered super-supportive, as they do not embed within the broader network of CSR actions.

The length of 2 in Table 3 reflects the most fundamental SSCB relationships, which are further visualized in Figure 9. This network highlights both homogeneous and heterogeneous associations between pairs of SSCBs. For homogeneous associations, activities such as strengthening legal management (G03) or improving corporate governance structures (G01) are often accompanied by enhanced information disclosure, jointly reinforcing transparency. More importantly, 88.33% of associations are heterogeneous, revealing that most SSCBs span cross-category relationships. For instance, community poverty alleviation initiatives are frequently linked with labor unions mechanisms for denial, reporting, and complaints, illustrating how firms address multiple stakeholder needs simultaneously.

5. Findings and Discussion

5.1. Super-Supportive CSR Behaviors

SSCBs are gaining traction in the construction industry as companies face the growing complexities of social, environmental, and governance concerns. Such behaviors emerge beyond traditional CSR, embodying a proactive and integrated commitment to sustainable development. The rationale behind SSCBs can be explained using multiple theories. These frameworks illuminate how SSCBs emerge as firms embed CSR in their core operations, evolving it into a strategic value-driven model that enhances long-term sustainability and strengthens corporate resilience.

This study identified 17 SSCBs (Figure 7 and Figure 9), suggesting that the scope of SSCBs ranges widely from corporate governance, community development, and employee welfare to environmental issues. In China, these SSCBs are shaped by unique institutional and economic conditions. For example, corporate governance SSCBs extend beyond regulatory compliance, aligning with China’s tightening policies on transparency, accountability, and anti-corruption under state supervision [58]. Such SSCBs not only strengthen organizational legitimacy but also improve competitiveness in public bidding processes, where CSR performance often influences project allocation. Employee-oriented SSCBs are particularly salient in China’s labor-intensive construction sector, where improvements in welfare, safety, training, and human rights can reduce turnover and enhance productivity while responding to the growing societal scrutiny of labor conditions [71]. Community-focused SSCBs resonate with the cultural norms of collectivism and guanxi, helping firms secure local trust, long-term partnerships, and reputational capital [72]. Environmental SSCBs, by contrast, are closely linked to China’s “double carbon” policy, encouraging construction firms to adopt green building standards, reduce emissions, and improve efficiency. These practices enable firms to gain both regulatory compliance and reputational advantages. These behaviors demonstrate how SSCBs not only mitigate risks but also create positive spillovers, positioning CSR as a strategic driver of long-term industry transformation.

Overall, these findings reveal that SSCBs in Chinese construction firms serve a dual role: addressing immediate social and environmental challenges while enabling long-term strategic positioning. Compared to earlier CSR studies [73,74], this research shows the interconnected and context-specific nature of SSCB. Their success depends on how effectively firms mobilize resources, foster synergies across organizational functions, and adapt to China’s unique institutional conditions. As such, SSCBs represent a bridge between global CSR principles and localized realities, showing how firms in emerging markets can translate broad sustainability ideals into actionable, value-driven practices.

5.2. Different Positions of SSCBs in Relationships

Unlike general CSR practices, SSCBs are integrated and highly supportive actions that catalyze chain reactions among initiatives and create a dynamic synergistic framework. As shown in Table 3 and Figure 9, SSCBs act as both initiators and outcomes, reinforcing different facets of CSR and promoting a unified approach to sustainable development. For instance, executive leadership in CSR not only establishes the foundation for advanced initiatives such as pollution control and innovation systems, but it also reinforces the diffusion of CSR values throughout the organization. Stakeholder theory suggests that addressing the expectations of diverse stakeholder is crucial for the integration of economic, environmental, and social dimensions [30]. In line with this perspective, enterprises are shifting from compliance-driven responses to proactive, strategically embedded practices [75].

The associations between SSCBs reveal cascading effects and feedback loops, consistent with the triple bottom line framework [76]. In China’s construction sector, however, these relationships are mediated by distinctive institutional and market conditions. Policy instruments such as the “double carbon” targets drive environmental practices, while competition for government-led projects encourages firms to showcase advanced CSR capabilities as a marker of competitiveness. At the same time, cultural traditions emphasizing collectivism and guanxi create informal networks of reciprocity that amplify the influence of SSCBs across stakeholder groups. As a result, employee-focused SSCBs and community engagement not only enhance legitimacy but also cultivate resilience, facilitate knowledge transfer, and stimulate collaborative innovation across projects.

These findings extend prior research by emphasizing the dynamic, associative, and synergistic nature of SSCBs. Rather than viewing CSR as fragmented or compliance-driven measure [73], this study views SSCBs as constituting a network of interconnected strategies. By amplifying synergies across behaviors, SSCBs enable firms to move beyond discrete initiatives toward holistic, transformative practices aligning with long-term sustainability goals. From a practical standpoint, the position of SSCBs within these relationships varies by firm size and capacity. Large construction enterprises are more capable of integrating governance, environmental, and innovation-oriented SSCBs to optimize resources and reduce costs. Small- and medium-sized firms (SMEs), by contrast, may adopt basic governance or community initiatives as entry points, gradually building synergies while managing costs. By strategically sequencing SSCBs, firms can avoid fragmented practices and develop coordinated portfolios that enhance resilience, competitiveness, and alignment with China’s sustainability agenda.

5.3. From Individual to Combined SSCBs

The progression from individual to combined SSCBs is best understood as a capability-building and resource orchestration process rather than a simple aggregation of actions. Individual SSCB associations act as foundational elements in which individual CSR behaviors serve as antecedents or consequences. These behaviors ensure regulatory compliance and foster stakeholder trust, thereby laying the groundwork for more advanced strategies. For instance, legal management and governance improvements not only help firms to meet regulatory standards but also build transparency and accountability, which in turn cultivate long-term legitimacy and resilience [77]. Such early stage behaviors link compliance and trust-building activities to broader sustainability goals and organizational resilience, setting the stage for more strategic and transformative CSR initiatives.

As CSR initiatives mature, construction firms increasingly adopt combined SSCBs, integrating strategies to address interconnected stakeholder needs across economic, environmental, and social dimensions. This movement reflects a strategic reorientation from fragmented efforts to interconnected practices. By embedding SSCBs into broader decision-making processes, firms can better manage diverse stakeholder expectations and respond to sustainability pressures. Scholars argue that integrating CSR into strategic planning enhances long-term competitiveness and resilience, particularly when different initiatives are coordinated and mutually reinforcing [8]. The combination of SSCBs such as governance reforms, community engagement, and environmental control illustrates how complementarities across domains can generate more durable and wide-ranging impacts.

Scaling SSCBs nonetheless presents challenges in resource allocation and coordination, especially in the Chinese construction sector where firms face strict regulatory demands and intense competition for government-led projects. While combining SSCBs can amplify CSR effectiveness and generate synergies, firms must carefully prioritize initiatives and sequence them strategically to avoid overextension. The resource-based view underscores that sustained competitive advantage depends on the effective deployment of valuable and inimitable resources, and combined SSCBs offer a pathway to achieve this by leveraging complementarities across organizational functions [78]. At the same time, firms must balance ambition with feasibility, gradually building interconnected SSCBs that strengthen resilience without undermining operational efficiency.

6. Conclusions

SSCBs represent critical organizational actions that demonstrate deep corporate social responsible commitment. These behaviors extend beyond basic measures to integrate social, environmental, and economic dimensions into cohesive strategies that create meaningful and synergistic impacts. Driven by internal motivations and external pressures, SSCBs form complex integrative patterns that allow firms to optimize resource utilization while enhancing their social contributions. Based on the findings, the key conclusions are as follows:(1) In the Chinese construction industry context, key SSCBs center on corporate governance, local community development, employee welfare, and environmental sustainability initiatives, such as recycling and pollution reduction.(2) Distinct SSCBs occupy varied positions within relationships and play diverse roles, while their connections evolve dynamically from individual and combined forms to hybrid structures that address multiple stakeholder needs simultaneously.(3) Construction firms are increasingly adopting sophisticated SSCB relationships, rather than relying on standalone initiatives. However, once the number of integrated actions surpasses a certain threshold, challenges emerge in coordination, resource allocation, and strategic alignment.

This study contributes to the CSR literature with distinct novelty centered on introducing and theoretically positioning the “super-supportive CSR behaviors (SSCBs)” construct. It formally defines SSCBs, examines their multidimensional integration into China’s construction industry (where behavior-focused CSR solutions are scarce), and opens the “black box” of SSCB relational patterns. Beyond this, this study advances CSR theory through three targeted extensions: (1) It expands the resource-based view from “resource allocation” to “behavioral leverage” by showing how foundational SSCBs optimize resource use for core CSR goals and competitive advantage. (2) It deepens institutional theory’s CSR application by clarifying triggerable SSCBs’ role in turning regulatory pressures into proactive practices. (3) It integrates stakeholder theory to balance investors’, communities’, and regulators’ diverse needs. (4) Supplementing the relevant CSR literature and distinguishing SSCBs, unlike existing “effect-focused” constructs, SSCBs emphasize actionable behaviors that amplify core CSR mechanisms, filling the prior literature’s gap in actionable guidance.

Beyond these theoretical insights, this study provides practical implications: For large firms, SSCBs guide resource allocation, cost optimization, and compliance monitoring to achieve ESG goals. For SMEs, threshold SSCBs offer templates to avoid fragmented CSR, expand synergies, and reduce costs, thereby supporting China’s “double carbon” policy and enhancing long-term resilience. Importantly, in practice, firms should also tailor SSCB adoption to their own capabilities.

Despite its valuable contributions, this study has three limitations. First, although the methodology provides a systematic approach to identifying SSCBs, it lacks detailed validation procedures. For example, additional robustness checks, such as sensitivity analyses for association rule mining thresholds or alternative algorithm comparisons, could further strengthen confidence in the results. Second, non-listed companies are underrepresented in the sample, comprising only 11.4% (24 reports), which limits the generalizability of the findings to non-listed or smaller construction enterprises. Third, the analysis relies on corporate self-disclosed CSR reports from 2010 to 2021. While these reports ensure consistent data coverage, they are subject to potential biases, including positive self-presentation and “greenwashing”. Such biases may introduce subjectivity into the data, potentially affecting the objectivity of observed CSR behavioral associations and the validity of the conclusions.

Future research could further advance this work in several ways. It could integrate discussions of China’s policy shifts to enrich the contextual analysis of CSR behavior drivers, explore how post-COVID-19 changes have reshaped CSR practices given the 12-year data span, and investigate SSCB resource allocation mechanisms and their impact on operational efficiency to deepen insights into CSR strategy optimization. Meanwhile, to address the limitation of self-disclosed reports, future studies could use third-party evaluation data (e.g., industry CSR ratings) to supplement and cross-validate corporate self-disclosures, expand the sample to include more non-listed enterprises, and compare SSCB patterns across industries or regions.