Abstract
The present research conducts a thorough meta-analysis to better understand the relationship and magnitude of the effect of corporate social responsibility (CSR) initiatives on consumers’ behavior and attitudes, including multiple moderators (e.g., type of CSR initiatives, type of industry, and country development level). Data comprises 744 effect sizes, extracted from 165 research articles encompassing 67,270 participants. Altogether, the findings reinforce the importance of CSR to consumers, suggesting that companies benefit from incorporating CSR initiatives in their business practices, having a positive effect of medium magnitude when no specific moderators are considered in the analysis (r = .38; linear models = .25; correlations = .44). However, findings also indicate that when considering specific moderators (e.g., country development, industry type, and cultural characteristics) and consumer responses type (e.g., attitudinal vs behavioral), this magnitude can differ significantly, providing evidence that consumer engagement in CSR activities is multifaceted and does not always translate into the expected outcomes.
Introduction
Corporate social responsibility (CSR) practices have become a growing concern for companies in the last decades (Carroll, 1999; Kim et al., 2016; KPMG, 2020; Öberseder et al., 2013). For example, Fortune 500 companies indicate that they invest circa 20 billion USD annually in CSR efforts (Meier & Cassar, 2018). In line with this, several CEOs advocate publicly for CSR policies across various issues (Global Giving, 2020). Likewise, governments are developing strong CSR legislation (European Union, 2020), and the United Nations is launching several initiatives to foster impactful CSR global initiatives (World Commission on Environment and Development [WCED], 2019). This CSR popularity has led some authors to argue that “its increased popularity inside boardrooms has outpaced the research needed to justify it” (Albuquerque et al., 2019, p. 4451).
CSR was initially defined in the 70’s as “the social responsibility of business encompasses the economic, ethical, legal, and discretionary (philanthropic) expectations, which society has of organizations at a given time” (Carroll, 1979, p. 500). But throughout time, this definition has evolved, with some authors stressing that CSR actions should also correct for adverse effects on society and maximize positive impact on communities (Mohr et al., 2001), emphasizing the importance of the company’s responsibility to improve societal well-being through corporate actions (Kotler & Lee, 2005; Schwab, 2021). This has led to multiple perspectives regarding what CSR is, its goals, and what it encompasses (Dahlsrud, 2008; Freeman et al., 2010; Matten & Moon, 2008). Importantly, despite the definitions of CSR not often converging among authors, the common element among the perspectives is that companies should engage in actions to impact the world positively (Kim et al., 2016; Weber, 2008).
The impact and the outcomes of CSR initiatives were initially approached from the perspective of corporations (Lee, 2008). The underlying idea was that CSR actions could boost companies’ competitive advantage, facilitating their differentiation from competitors and fostering sustainable relationships with their stakeholders (Lee, 2008). Consequently, engaging in CSR activities became essential to companies’ long-term success (Chang & Yeh, 2017), reducing financial risk default (Sun & Cui, 2014), increasing premiums in M&A deals (Ozdemir et al., 2022), firm performance (Mutuc & Cabrilo, 2022), resilience to crisis shocks (Epure, 2022), promoting innovation (Hou et al., 2023), value creation (Sánchez & Sotorrío, 2007), and affecting corporate tax payments (Timbate, 2023).
Part of existing literature also views CSR activities as mechanisms of tempering firms’ reputational loss from fraud or other illegal activities such as excessive tax avoidances. However, there is much to know about CSR’s impact on firm’s performance (Laplume et al., 2022) and on other beneficiaries besides the shareholders (Barnett et al., 2020), such as employees (Gond et al., 2017; Kruse, 2024; Onkila & Sarna, 2022).
The current focus of CSR initiatives on consumer behavior and attitudes suggests an unclear picture, with divergent findings, both in the magnitude of its impact on consumers’ behavior and attitudes and its valence (Bergkvist & Zhou, 2019; Peloza & Shang, 2011). Although consumers seem to believe that companies should be accountable for benefiting society and the environment (Berens et al., 2005; Öberseder et al., 2013), some studies question whether CSR initiatives indeed impact consumers’ behavior and to what extent such relationship might be influenced by other variables (Van Doorn et al., 2017). Moreover, some authors have raised the importance of making a distinction between passive CSR initiatives, typically described as economic or legal obligations, versus active CSR initiatives, characterized by the proactive adoption of corporate practices to foster economic, social, and environmental development beyond what companies are forced to comply by law (Kim, 2017; Windsor, 2001). We build on this and also analyze the impact that this dichotomy between active versus passive CSR initiatives may have on the observed effect of CSR on consumer responses.
Overall, the complexity of this phenomenon urges for a systematic overview and a meta-analytical approach to allow a better understanding of how and under which circumstances CSR actions can impact consumers’ behavior and attitudes toward the companies. Although some prior meta-analyses have also examined the impact of CSR on consumer responses (e.g., AlJarah & Emeagwali, 2017; Peng et al., 2024; Vieira et al., 2023), they differ from this current work on some dimensions. First, most of them treat CSR as a single construct, not analyzing whether the nature of the CSR initiatives can generate and impact different consumer responses, as we propose in the current work. Second, instead of aggregating in one single construct the consumer responses, or focusing on one single type of responses, we propose to offer a more detailed approach, taking into consideration the several different types of consumer responses (e.g., loyalty, WTP, quality perceptions), distinguishing then between behavioral and attitudinal consumer responses. Third, we combine methodological and multiple theoretical moderators (e.g., industry type, country individualism, country development level) in one model, testing for potential interaction effects, allowing us to offer an integrative and holistic perspective regarding the CSR impact on consumer responses. We study this through a thorough meta-analysis covering 744 effect sizes extracted from 165 research articles, encompassing 67,270 participants, to provide a meaningful understanding of this phenomenon.
Theoretical Background
In the last decade, CSR has become a popular topic for managers and academics (Liu et al., 2014; Meier & Cassar, 2018). Initially studied from the shareholder’s perspective (Barnett & Salomon, 2006; Waddock & Graves, 1997), only recently have researchers started to focus on the extent to which CSR initiatives can also impact other stakeholders (Dmytriyev et al., 2021; Weber, 2008), and, in particular, consumers (Ellen et al., 2006; García-Jiménez et al., 2017; Luo & Bhattacharya, 2006; Sen et al., 2006). As Grohmann and Bodur (2015) highlighted, CSR activities can affect multiple stakeholders, with customers being one of the key groups impacted by a company’s initiatives.
Interestingly, the CSR field has grown significantly, encompassing many theories and approaches (Abdeen et al., 2016; Grohmann & Bodur, 2015). From an initial approach that analyzed CSR initiatives as one single construct, some authors have begun distinguishing between them based on their nature. One of the most popular approaches was suggested by Carroll (1979), who stated that CSR should be considered a multidimensional construct encompassing organizations’ economic, legal, ethical, and philanthropic responsibilities. Another approach is derived from the definition of sustainable development, suggested by the UN World Commission on Environment and Development (WCED, 2019), which defines CSR according to three main pillars: economic, environmental, and social. According to this approach, CSR should include the voluntary integration of societal and environmental concerns in the organization’s daily operations, overseeing their impact on the economy while managing their interaction with stakeholders. The importance of encompassing these several dimensions was reinforced by several authors emphasizing the need to study the CSR phenomenon, considering the nature and goal of the different initiatives (Garcia-Piqueres & Garcia-Ramos, 2022; Sun & Cui, 2014; Weber, 2008). A more recent approach has also outlined the importance of distinguishing between passive and active CSR initiatives (Kim, 2017; Torugsa et al., 2012). In this case, the authors outline the importance of differentiating between CSR initiatives undertaken by companies with the goal of compliance with legal requirements and initiatives that reflect a higher level of concern for responsible business practices above and beyond what is legally required.
Independently of the several definitions and conceptualizations adopted by authors, the fact is that CSR is a well-established and growing phenomenon, with many companies adopting this practice as part of their strategic decisions (Weber, 2008) or their intrinsic values (Matten & Moon, 2008). From an initial phenomenon mainly developed inside the companies (Dmytriyev et al., 2021), companies are now communicating these practices outside of the corporate world, using their CSR initiatives as a way to capture consumer and other stakeholders’ attention (Bakker et al., 2020), signaling their commitment to embracing positive causes (Conte et al., 2023). CSR initiatives have become part of most companies’ strategic decisions, often heavily communicating them to consumers to influence their behaviors and attitudes toward the companies.
Consumers, in turn, are becoming increasingly interested in CSR initiatives (Öberseder et al., 2013) and concerned with companies’ impact on societies and the environment (Palacios-Florencio et al., 2018). Consumers are not only concerned about the effects of their purchases beyond the return it can bring them directly, but a new conscientious consumer has emerged, caring if their purchases positively impact the world (Bulut et al., 2021). This conscious consumer wishes to purchase from socially responsible companies (Auger et al., 2008; Peng et al., 2019) and has a high opinion regarding those that act responsibly (Morales, 2005), purchasing from brands with CSR policies rather than less responsible companies (Auger et al., 2008).
Many of these consumers also advocate against brands that harm the environment and society (Nyilasy et al., 2014), boycotting companies that do not have CSR policies (Cotte & Trudel, 2009), and showing a willingness to exert effort to participate in CSR actions (Iglesias et al., 2020). Besides, their judgment about the company goes beyond actions, reaching the perceived intent level, and screening for companies’ sincerity, punishing the ones whose efforts are perceived as insincere (Nyilasy et al., 2014; Sen & Bhattacharya, 2001).
Indeed, findings from several articles suggest that consumers have more positive attitudes toward CSR practicing companies (Bhattacharya & Sen, 2003; Brown & Dacin, 1997), higher company–customer identification (Marin et al., 2009; Salmones et al., 2005), and loyalty (Marin et al., 2009; Pérez & del Bosque, 2015). CSR and its outcomes have also been reported to translate into higher purchase intentions and behavior (Bhattacharya & Sen, 2004; Mohr & Webb, 2005), the spread of positive word-of-mouth (Bhattacharya & Sen, 2004), and increased brand awareness (Tian et al., 2011). Therefore, we propose a first general hypothesis that overall, CSR initiatives are expected to affect consumer responses positively:
H1. Corporate engagement in CSR initiatives positively impacts consumer responses toward the company.
Interestingly, as Peloza and Shang (2011) highlighted in their systematic review, several prior empirical studies provide indefinite conclusions concerning the magnitude of consumers’ responses to CSR activities. Indeed, while the literature suggesting a positive effect of CSR actions on consumer behavior is extensive, there is also evidence that this might not always be the case. For example, the general consumer awareness of CSR activities can be relatively low (Sen et al., 2006), leading to low behavioral responses to CSR initiatives, and consumers are becoming more sensitive to perceived corporate hypocrisy (Yue et al., 2023). These reported inconsistencies might also result from measuring consumer responses through different variables. Responses requiring more intent might depend more on the knowledge and conscious decision-making toward a brand.
In contrast, more automatic responses might depend less on the conscious availability of CSR information in one’s mind (Martin & Morich, 2011). Also, behavioral responses might be more challenging to change than perceptual and psychological ones. For instance, changing a brand’s image might be easier than leading consumers to spread word-of-mouth (WOM) about the brand. Indeed, while Mohr and Webb (2005) found CSR to affect purchase intentions positively, even more so than price, other studies failed to find such effects on paying a premium for a product with CSR stewardship (Atalik & Eratik, 2015). This suggests that the magnitude and valence of the impact of CSR’s initiatives on consumer responses are affected by the type of response studied. As such, we hypothesize that:
H2. CSR initiatives will have a greater impact on consumer attitudes than on consumer behaviors.
Type of CSR Initiatives
Besides recognizing that the impact of CSR initiatives on consumer responses depends on the type of variables studied, it is also important to analyze whether the responses depend on the type of CSR initiatives. As outlined by several authors, a vast array of actions and practices are commonly considered under the umbrella term of CSR (Carroll, 1999; Weber, 2008; Windsor, 2001). Therefore, the nature of CSR might be an important moderator in explaining discrepancies between findings. Some authors even claim that it is important to distinguish between different types of initiatives due to the complex and intricate nature of CSR initiatives (García-Piqueres & Garcia-Ramos, 2022; Weber, 2008).
For example, some authors found that the degree of influence of CSR on brand image was dependent on the consumers’ perceptions regarding CSR type (Wu & Wang, 2014), while others found that the CSR initiatives that are legally required had a lower impact on brand image in contrast to ethical CSR activities (Lho et al. (2019). Indeed, while economic and legal CSR initiatives are typically required by law, ethical and philanthropic actions are often optional (Windsor, 2001), with some authors even making a distinction between passive versus active CSR activities (Kim, 2017). This distinction between companies that simply comply with their legal CSR obligations (passive) versus those that go one step further and actively engage in CSR initiatives (active) seems to matter (Kim, 2017; Knudsen & Moon, 2022). For example, Ellen and colleagues (2006) found that if a firm’s motivation for CSR was considered selfish, it undermined the effect of CSR on consumer purchase intention, while value-based motivations for CSR did not. Also, Kodua et al. (2016) and Stanisavljević (2017) found evidence for a stronger effect of philanthropic and ethical CSR (active), not finding any effect for legal CSR (passive). Therefore, CSR’s impact may vary according to the nature of the initiative, depending on how altruistic and proactive it is perceived (Peloza & Shang, 2011; Tully & Winer, 2014). As such, we hypothesize the following:
H3. Active (non-legally mandatory) CSR initiatives will have a more positive effect on consumers’ responses than passive (legally mandatory) CSR initiatives.
Type of Industry
Another possible factor that may lead to differences in the impact of CSR on consumer responses is the type of industry engaging in CSR actions. For example, the reaction of customers to CSR activities in the service sector industries may differ from that of other industries due to differences in the level of relationship between the customer and the company (Asatryan, 2013; Moisescu, 2017). For example, the tourism industry consumes a large quantity of natural resources, impacting the environment negatively (Kotler et al., 2017) and thus faces pressure to increase CSR-related activity to minimize its environmental effects (Lee et al., 2018; Wong & Kim, 2020). Likewise, in the airline industry, environmental concerns may be at the forefront of consumers’ minds, placing companies under increasing pressure to reduce their environmental impact (Chen et al., 2012; Hagmann et al., 2015).
Similarly, legal CSR initiatives were found to affect the casino industry (McCain et al., 2019), arguably given the industry’s negative social impacts. Therefore, consumers’ overall perceptions and evaluations seem to depend on the extent to which they will value and trust the initiatives companies adopt, suggesting that this relationship may depend on the specific context of a given industry. Thus, we hypothesize the following:
H4a. The type of industry of the companies pursuing CSR activities moderates the impact of CSR actions on consumer responses.
Moreover, the negative reputation associated with specific industries can limit the perceived value of CSR initiatives (Peloza et al., 2012; Sen et al., 2006), with some authors distinguishing between controversial or non-controversial industries. Some prior studies show that the controversy regarding companies can negatively affect consumer responses, such as consumer identification (Jo & Na, 2012), with CSR efforts in controversial industries leading to less favorable brand outcomes than those in non-controversial industries due to heightened consumer skepticism and attribution effects. Authors have suggested that when companies in sectors such as tobacco, alcohol, or oil engage in CSR, stakeholders may perceive these initiatives as strategic reputation management rather than genuine social responsibility (Palazzo & Richter, 2005; Yoon et al., 2006). This skepticism can trigger consumer reactance or backlash, leading to the dismissal of CSR efforts and more negative attitudes toward the brand. Supporting this view, Vieira et al. (2023), in their meta-analysis, find that controversy moderates the impact of CSR on customer–company identification, suggesting that CSR efforts in controversial industries might be less effective in fostering positive consumer responses. Some industries may suffer more pressure to engage in CSR initiatives due to their negative impact on the environment or society, with some authors finding that companies working in industries perceived as questionable by society (i.e., controversial) are more likely to develop CSR policies and transparency tools (Conte et al., 2023). While CSR in controversial industries can yield benefits, its impact may not be as strong as in non-controversial industries. Based on this reasoning, we propose the following hypothesis:
H4b. CSR initiatives in controversial industries have a weaker influence on consumer responses than those in non-controversial industries.
Country Development Level
Another relevant perspective concerning CSR is related to differences in consumer response depending on the countries’ development level (e.g., developing vs developed countries) (Jamali & Karam, 2018). Although many of the prior studies focused on developed markets such as North America and Europe (Preuss et al., 2016), more recent studies started focusing on developing markets, pointing toward different levels of impact of CSR activities across countries, depending on their level of development. This difference might be related to the fact that, in more developed countries, compared with emerging countries, the adoption of CSR activities by companies is a well-established practice (Zhang et al., 2018), and thus, it is more likely to affect consumers’ responses. Contrarily, in developing countries, companies often operate in markets struggling with political, environmental, and social instability, having fewer incentives to invest in CSR, generating, therefore, lower CSR awareness (Mombeuil & Fotiadis, 2017; Ramasamy & Yeung, 2009). For example, familiarity and understanding of CSR in Bahrain are very low (Shabib & Ganguli, 2017), and there is a low level of CSR consumer awareness in Malaysia (Abd Rahim et al., 2011). This leads to the following hypothesis:
H5. The higher the country’s development level, the more positive the effect of CSR activities on consumers’ responses will be.
In addition, consumers from developing countries seem to place higher value on companies that create jobs and economic opportunities (Dartey-Baah & Amponsah-Tawiah, 2011) than companies engaging in CSR initiatives to minimize environmental impact (Yen & Fleck, 2020). Also, the importance placed on different responsibilities varies among countries, with economic responsibilities having a greater emphasis in developing countries, followed by philanthropical, legal, and ethical responsibilities (Abd Rahim et al., 2011; Dartey-Baah & Amponsah-Tawiah, 2011). Contrary to what happens in developing countries, companies in developed countries seem to pay more attention to their social responsibilities, focusing more on developing CSR activities that minimize their environmental impact (Desta, 2012). Therefore, it seems relevant to test if the effect of CSR on consumer responses will depend on the interaction between a country’s development level and the type of CSR initiatives. As such, we hypothesize that:
H5a. In developed countries, the impact of non-economic CSR (e.g., environmental or ethical) initiatives on consumer responses will be higher than in developing countries.
H5b. In developing countries, the impact of economic CSR initiatives on consumer responses will be higher than in developed countries.
Combined Effect of Type of CSR × Industry × Country Development Level
In addition, there is also some evidence that the hypothesized moderators can altogether affect consumer responses. For example, companies tend to opt for CSR in domains that relate mainly to the country’s core industries (Jamali et al., 2017). This may affect consumers’ responses toward different CSR activities since consumers will be exposed just to a subset of CSR practices. For instance, the large scale of foreign investment in countries such as Bangladesh, India, and Pakistan related to telecommunications, textiles, and garments contributes to introducing CSR-aligned initiatives in such markets (Alam & Rubel, 2014; Ali et al., 2010). Also banking and insurance are among the most frequently mentioned sectors in CSR studies in emerging countries (Jose & Buchanan, 2013; Lee et al., 2017), while in European and North American countries, the impact of CSR in areas as ready-made goods like organic and green food-related products, fashion and retail brands, are among the most studied sectors (Castaldo et al., 2009; Ferreira & Ribeiro, 2017; Mohr & Webb, 2005). From a different perspective, some authors have identified distinctions between the determinants of CSR disclosure in developed and developing countries, with industry emerging as a key driver of the CSR reporting agenda (Ali et al., 2017), while other report that consumers in developing countries perceive CSR activities differently than consumers in developed countries (Moisescu, 2017). Therefore, depending on the country’s development level, different CSR activities may emerge in the local focal industries, enhancing differences in CSR practices between developing and developed countries (Jamali & Carroll, 2017; Jamali & Karam, 2018). Therefore, we hypothesize that:
H6. The impact of CSR on consumer responses is moderated (negatively or positively) by the interaction between CSR initiative type, the country’s development level, and the type of industry.
Type of Culture
Finally, evidence shows that the effect of CSR on consumer responses is affected not only by economic and country developmental characteristics but also by cultural factors (Becker-Olsen et al., 2011; Vieira et al., 2023). For example, in cross-cultural work covering the United States, France, and Germany, Maignan (2001) found that consumers ranked the importance of Carroll’s (1979) four CSR-related dimensions differently. Differences in this relative ranking have also been found for other countries, such as Malaysia (Abd Rahim et al., 2011) and Thailand (Nochai & Nochai, 2014). In fact, there is growing evidence suggesting that certain countries are more prone to care about CSR than others due to differences in cultural values. Moreover, more collectivistic societies expect higher levels of CSR from corporations (Lee and Lee, 2015; Moon et al., 2015). Eisingerich and Rubera (2010) add that collectivist societies’ requirement for greater consideration of the effect of business on society makes them more sensitive to differences in CSR activities between companies. Likewise, recent work (Vieira et al., 2023) suggests that differences in individualistic and collectivistic countries can moderate the effect of CSR on consumer response. This leads to the following hypothesis:
H7. The relationship between CSR activities and consumers’ responses is stronger in collectivistic than in individualistic societies.
Altogether, we propose a set of relationships between the different types of CSR initiatives and the different consumer responses, as suggested in the conceptual framework presented in Figure 1.
Conceptual framework of the meta-analysis.
Reasoning for a Meta-Analysis
The diversity of results embedded in different economic and social characteristics and the myriad of different conceptualizations of CSR initiatives make it challenging to generalize the impact of CSR activities on consumer responses. Moreover, some types of CSR and specific aspects of consumer behavior may have received more attention than others, which might bias the estimation of its impact, calling for empirical approaches to study CSR impact (Schreck et al., 2013). As highlighted by several authors, adopting a meta-analytical analysis is especially relevant since it allows investigating a specific phenomenon over time, exploiting variations across different settings and studies (Aguinis et al., 2011; Geyskens et al., 2009).
Although some prior meta-analyses have also examined the impact of CSR on consumer responses (see Table 1), they differ from this current work on several dimensions.
Meta Analyses on CSR Effects on Consumer Responses.
First, most studies have considered CSR as a single construct, not considering that the nature of the CSR initiatives can generate different consumer responses, as we propose in the current work. Only Vieira et al. (2023) and Peng et al. (2024) attempted to do so, but without following the same taxonomy or scope of analysis. While Vieira and colleagues (2023), who were mainly interested in understanding the mediating role of Consumer Company Identification (CCI), made a binary distinction between atomistic and holistic CSR actions, Peng and colleagues (2024) made a distinction between core versus extended CSR, based on the targeted stakeholders. We contribute to this body of research by distinguishing between the nature of CSR initiatives (e.g., environmental vs social CSR) and active versus passive initiatives.
Second, we take into consideration the several different types of consumer responses (e.g., loyalty, WTP, quality perceptions), distinguishing between behavioral and attitudinal consumer responses. The majority of prior meta-analyses do not offer this detailed and integrative perspective, joining in one single construct the consumer responses, or focusing on one single type of response. An exception is the work of Peng and colleagues (2024) that distinguishes between three types of consumer responses (evaluation, intentions, and relationships).
Third, we combine methodological and theoretical moderators in one approach, testing for potential interaction effects in a complete and integrative perspective. Importantly, we include a continuous indicator of country development (HDI index), while most prior meta-analyses adopt a binary or categorical approach. In addition, we assess potential differences driven by industries using extensive categorization coding (we followed the Global Industry Classification Standard—MSCI), while most prior works adopt a binary approach, limiting their ability to identify the industries where CSR has the most significant impact.
In light of the above, the contributions of the present meta-analysis compared with prior work devoted to the same theme are fourfold: (1) it measures the magnitude of the effects of CSR initiatives on different consumer responses; (2) it examines if the characteristics of the CSR initiative, such as the initiative type, impact the hypothetical relationship between CSR initiatives and consumer responses; (3) it analyzes how contextual factors such as countries’ development, industries, or culture influence CSR’s impact on consumer responses; and (4) it tests for potential significant interactions between the moderators, to better understand the drivers of consumer responses.
Method
Data Search
In order to identify the highest number of articles associated with the current research questions, a typical meta-analysis data collection procedure was adopted, starting with data search, followed by screening by title and abstract, data adequacy analysis, and finalized with data coding (Cooper et al., 2019; Lipsey & Wilson, 2001).
We started by conducting an extensive search in all main scientific databases of published articles across mainly used academic databases (EBSCO, Emerald Insight, Science Direct, Scopus, Google Scholar, SSRN, Web of Science), as well as several conferences’ websites (e.g., Marketing Science, EMAC, ACR), and additional journal lists not included in previous databases.
When searching for articles, we used multiple possible combinations of CSR with the keywords of interest (“consumer attitude,” “consumer behavior,” “consumer choice,” “product attitude,” “product choice,” “consumer response,” “loyalty,” “purchase intentions,” “purchase behavior,” “willingness to pay,” “attitudes,” and “beliefs”) both in title and abstract (e.g., CSR in the title, with Keyword X in the title, followed by CSR in the title and keyword X in abstract, CSR in abstract and Keyword X in the title, and finally both CSR and Keyword X in the abstract). This procedure aimed to maximize the likelihood of hits, covering all the possible combinations of keywords. We also asked for unpublished manuscripts by contacting authors of papers in the area to tackle the file drawer problem. According to Lipsey and Wilson (2001) and, as discussed by Rothstein and colleagues (2005), that is a good way of preventing publication bias since there is a tendency to publish only studies that find significant or positive results. Altogether, this search method identified 471 initial articles after excluding duplicates (Figure 2).
Studies’ selection process.
After analyzing all the articles, 52 articles did not include any CSR manipulation, 90 articles reported measures which were not adequate for the current analysis, 82 did not focus on the relationship under analysis (consumers’ attitudes and behaviors), seven were excluded due to a language barrier, and 75 for insufficient data (e.g., not enough data on the relationship to extract relevant statistics), resulting in 165 articles. Finally, these 165 articles were coded, resulting in 744 effect sizes, encompassing 67,270 participants, comprising articles from 1997 to 2021, covering 24 years of past research on CSR.
Data Coding
Two independent judges were responsible for the coding, and a third judge solved discrepancies. To ensure the maximum level of accuracy, a new independent coder was asked afterward to validate the coding previously done. The data and all relevant variables were then recorded to allow for the effect size calculation (r or betas), and the theoretical moderators were coded (for details, see Table 2).
Coding of Consumer Responses and Theoretical Moderators.
GICS - Global Industry Classification Standard - MSCI.
Human Development Index (HDI) | Human Development Reports (undp.org).
Country Comparison - Hofstede Insights (hofstede-insights.com).
CSR initiatives were coded according to the authors’ classification (of each study) as environmental, philanthropic, social, ethical, legal, economic, and multidimensional. Regarding the passive versus active classification, we built on the taxonomy of Windsor (2001) and Kim (2017), coding as passive CSR initiatives the ones identified as having a legal or economic nature (i.e., mandatory or required) and coding as active CSR initiatives those indicated as having social, philanthropic, environmental, or ethical CRS purpose. As referred to by Kim (2017) and Torugsa and colleagues (2012), these are business practices that are adopted voluntarily by companies and go beyond simple governmental obligations. We also coded certain CSR initiatives as “other” when they were described as multidimensional or had no particular focus on a specific type of CSR.
The Consumer responses were coded in line with the type of consumer responses included in the articles (e.g., loyalty). Instead of treating these responses exclusively as specified in the different studies, we treat them at two different levels of analysis: the level 1 of this DV encompasses the different outcome variables of each study (e.g., image, loyalty, and purchase intentions, among others), while level 2 encompasses the outcome variables grouped and coded as behavioral or attitudinal. This methodology allowed us to estimate different effect sizes at two different levels: attitude versus behavior, and specific consumer responses. Therefore, similar to the reclassification done to the CSR types, consumer responses were also recoded as behavioral (e.g., purchase intentions) and attitudinal responses (e.g., trust).
Regarding the industry sector type, this moderator was coded according to the industry group classification of the Global Industry Classification Standard (GICS), a hierarchical industry classification system categorizing companies according to their principal business activities. Following prior works, the industries of the studies were also coded as controversial versus non-controversial (Peng et al., 2024; Vieira et al., 2023). We built on the work of Oh et al. (2017) and Cai et al. (2012) and classified all the industries related to tobacco, alcohol, gaming, firearm, military, nuclear, oil, cement, and biotech as controversial. In addition, building on the insights from Jansen et al. (2024), we classified the industries related to airplane and shipping, textile and clothing production, and fast food as controversial, leading to a total of 14 papers (45 effect sizes) encompassing controversial industries and 151 papers (699 effect sizes) focusing on non-controversial industries.
The country development level was coded using the Human Development Index—HDI, a well-established measure of the average country achievement in key dimensions of human development (Min = 0.076, Max = .949), while the level of country individualism was coded according to Hofstede ranking level of individualism, both based on the year of publication and the country where the different studies’ data was collected (Min = 8, Max = 91).
Regarding the sample type, it was coded as student versus non-student samples, while the type of research method was coded according to the study methodological types (experimental vs survey).
Overall, the 165 articles and the 744 effect sizes identified after the data search (Figure 2) were coded according to the coding guidelines presented in Table 2 to develop a detailed meta-analysis on the impact of CSR initiatives on consumers’ attitudes and behaviors.
Statistical Method
In the current analysis, we include several measures of both CSR types and consumer responses to understand their overall relationship. The advantage of this approach is that by including a wide range of evidence across all the dimensions on which CSR has been tested and measured, it does not arbitrarily exclude findings, reducing publication bias (Cooper et al., 2019). This implies that multiple measures can be extracted from a single published study. Since the random effects model, commonly used in prior meta-analyses, is unable to cope with such a data structure (due to the violation of the independence of each effect size), we opted to apply a multilevel meta-analysis model. The adoption of this method allows us to account for three sources of variance between effect sizes that are worth outlining: sampling variance (level 1), variance between measured effect sizes within studies (level 2), and variance between studies (level 3). This method extends the two levels of mixed effects models, including an additional level of dependence for effect sizes extracted from the same study (Cheung, 2019).
Moreover, this analysis includes two different effect sizes: correlations and beta coefficients from linear regression models. 4 While the beta coefficients measure the effect of CSR on consumer responses, controlling for other variables included in the model, the correlation effect sizes measure the simple relationship between them. To allow the inclusion of both effect sizes in the model, the values extracted using a Beta Estimation Procedure (BEP) were converted, following Peterson and Brown (2005). Importantly, this procedure is only appropriate when the value for Beta is in the range [−0.5, 0.5]. As such, we excluded a small number of high-effect sizes outside of this range (Roth et al., 2018). Results are reported across all included effect sizes, testing if controlling for effect sizes derived from a BEP would change the results. 5
Results
Data and Meta-Analysis Model Fit
We started by examining the appropriateness of the data and the proposed model fit. To test the imposition of this multilevel structure, we compared the effect of removing different levels and examined the Akaike information criterion (AIC) and the Bayesian information criterion (BIC), following Harrer and colleagues (2019). Results for the high Q statistic indicate that the studies are highly heterogeneous and cannot be considered repetitions of a similar experiment (Q = 34,001.44, p < .001). The removal of either level 2 or 3 was also tested, with findings indicating it would reduce the model’s fit (p < .001) and that each level of the structure explains some degree of variance in the results. Ignoring this structure would underestimate the variance in the data or misrepresent the source of this variance. Moreover, the analysis indicated that only 2.35% of the total variance is explained by sampling error, 40.98% is explained by differences in effect sizes extracted from the same study, and differences between studies explain 56.67%. Altogether, this provides evidence of the adequacy of the data sample and supports the importance of running a broad meta-analysis.
Second, the potential publication bias was also examined to check for potential issues with unpublished results, implementing Egger’s regression (Egger et al., 1997) directly by adding N to the three-level model (Suurmond et al., 2020). Results showed a small but significant positive coefficient for the sample size (b = 0.0002, p < .001), 6 indicating a potential issue with publication bias, suggesting that the results may be slightly conservative and that the data is potentially missing small sample size studies with large effect sizes This is less likely to be an issue as large effect sizes are more likely to be published than smaller ones.
Hypothesis Testing
Hypothesized Main Effects
We started by testing H1, which suggested a general positive impact of CSR on consumers’ responses toward endorsing companies. As expected, findings suggest a medium correlation effect size of r = .38 over the complete dataset, supporting our hypothesis that CSR positively influences consumers’ responses toward companies (see Table 3 for a summary of the results). The effect sizes reported in the papers range from [−.58, .98].
Summary of Results.
Note: α = intercept; b = unstandardized coefficient; r = effect size; [CI-LB, CI-UB] = lower and upper bounds of confidence intervals; MM = methodological moderators.
Effect size = intercept + coefficient for environmental CSR + coefficient for HDI × HDI value + coefficient for interaction between environmental CSR and HDI × HDI.
Effect size = intercept + coefficient for consumer staples + coefficient for social CSR + coefficient for HDI × HDI value + coefficient for the interaction between consumer staples and social CSR + coefficient for interaction between consumer staples and HDI × HDI + coefficient for the interaction of social CSR and HDI × HDI + coefficient for the three-way interaction between consumer staples, social CSR, and HDI × HDI.
Effect size = intercept + coefficient for financial industry + coefficient for ethical CSR + coefficient for HDI × HDI value + coefficient for the interaction between financial industry and ethical CSR + coefficient for interaction between financial industry and HDI × HDI + coefficient for the interaction of ethical CSR and HDI × HDI + coefficient for the three-way interaction between financial industry, ethical CSR, and HDI × HDI. Note: the output of these formulas is a fisher-z, which needs to be converted to an r.
Importantly, we tested for paper influence and found no papers that have a DFBETA greater than one or with a Cook’s distance that would cut off 50% of a chi-square distribution with one degree of freedom, indicating that there were no papers with a considerable influence and, as such, we can use the full sample. Moreover, we also tested the sensitivity of the effect size to the computation of values from the BEP by adding a moderator variable for BEP effect sizes derived from regression coefficients. Importantly, we find a significant difference in effect size (p < .001) with the effect size derived from linear models (r = .25), smaller than the effect size for all other data sources (r = .44; correlations). These results are interesting since they highlight that when studies use correlations, they may offer an enhanced result because they do not control for several variables simultaneously.
We proceeded then with the analysis of the remaining hypotheses, always testing each effect with and without controlling for BEP coefficients. We solely present results without controlling for BEP coefficients since results controlling for BEP coefficients remain unchanged.
Theoretical Moderators
When testing for hypothesis 2, which proposed that the impact of CSR initiatives would be stronger for consumer attitudes than consumer behaviors, we find a significant effect (F (1, 739) = 8.963, p = .003), such that behavioral responses lead to a lower effect (b = −.059, t (739) = −2.994, p = .003) than attitudinal responses. This provides evidence that although consumers favorably evaluate CSR initiatives, they often do not engage in the corresponding behaviors, which is in line with the attitude-behavioral gap reported in the literature (Baron & Spranca, 1997). This suggests that studies focused on the impact of CSR initiatives on behavioral consumer responses are likely to report lower CSR effects, which researchers and managers should be aware of.
Indeed, when testing for this across the different consumer response types, analyzing if the effects of each consumer response are significantly different compared with all the other responses, we find that for the image toward the endorsing company, an attitudinal response, the effect of CSR initiatives is higher (b = 0.097, t (739) = 3.721, p < .001). On the contrary, when considering the WTP (b = −.245, t (739) = −4.348, p < .001) and loyalty (b = −.047, t (739) = −2.071, p = .039), both behavioral measures, we find lower positive effects, compared with the other types of consumer responses, providing additional support to this hypothesis. Interestingly, when analyzing the impact of CSR initiatives on consumer quality perceptions about the company, we also find a lower positive effect (b = −0.133, t (739) = -3.129, p = .002). None of the remaining consumer responses were significantly different.
When testing for H3, which suggested that active CSR initiatives will have a stronger effect on consumers’ responses than passive CSR ones, results were not significant (F (2, 732) = 0.505, p = .604). However, when analyzing the type of consumer response considering the attitudinal versus behavioral classification, we find a significant interaction (F (3,727) = 4.404, p = .004), partially supporting this hypothesis. Specifically, this finding suggests that even though behavioral responses are significantly less affected by CSR initiatives than attitudinal ones (b = −0.033, p = .003), there is a marginally significant interaction effect (b = 0.096, p = .057), indicating that the exposure to active CSR initiatives mitigates this reduction leading to higher behavioral responses (r = .37) than the exposure to passive CSR initiatives (r = .31). This provides an interesting insight for managers and researchers showing that the engagement in voluntary (active) CSR initiatives has a positive effect on consumers’ willingness to behave accordingly.
Interestingly, when analyzing the moderating impact of industry sector on CSR’s initiatives on consumer responses (H4b), despite the reasoning in favor of the importance of this variable, results indicate that the industry type seems to not significantly moderate the results (F (8,735) = 0.544, p = .800). However, when testing for the individual levels of the moderator, we find that the combination of consumer discretionary spending with the multidimensional CSR type has a significant effect (t (734) = −0.198, p = .004), such that the effect is higher for both consumer discretionary (r = .45) and multidimensional CSR initiatives (r = .40), but lower for their combination (r = .32), suggesting that when consumers engage in purchases out of choice (as opposed to need), aspects related with CSR initiatives undertaken by companies are less taken into consideration in the choice process.
Furthermore, when testing for H4b, which suggested that controversial industries could lead to a weaker effect on consumers’ responses than non-controversial ones, results were not significant (F (1, 742) = 0.000, p = .989). This is an interesting finding as it contradicts the existing research on the topic (Song et al., 2020; Vieira et al., 2023; Yoon et al., 2006). One possible reason for this might be that the distribution of controversial versus non-controversial companies in the current sample is unbalanced, with the largest % of industries classified as non-controversial (93.6% vs 6.4%). This might be because academic journals may exhibit a bias against publishing studies on stigmatized industries due to concerns over legitimacy or potential criticism. Additionally, authors who are interested in studying CSR’s impact on consumers’ behavior and attitudes may have a natural tendency to focus on less controversial industries (vs gambling, oil, and army industries), due to the difficulty in obtaining reliable data (Marshall et al., 2023), which overall may affect the analysis of this phenomenon.
Regarding the moderating impact of the country development level on the hypothesized relationship, results indicated that the country development level alone is not a significant moderator (F (1, 699) = 0.614, p = .434). Although surprising from a theoretical perspective, these results are in line with findings from Santini et al. (2021) and AlJarah and Emeagwali (2017) who did not find significant results when testing for the moderating effect of country development in their meta-analysis. A possible explanation for this might be due to the existence of more studies focused on the impact of CSR on consumers from developed countries, compared with studies in developing economies, creating unbalanced samples.
Finally, when testing whether the relationship between CSR and consumers’ responses would be stronger in collectivistic than in individualistic societies (H7), results suggested that it was not the case (F (1,668) = 0.590, p = .443). Again, despite being surprising from a theoretical perspective, these results are in line with the findings from Aljarah and Ibrahim (2020) and Santini and colleagues (2021), who did not find a moderating effect of culture on the impact of CSR on different consumer responses. This result can be related to an increasing focus on CSR by large global companies (KPMG, 2020; Meier & Cassar, 2018), many of them present globally, which may have contributed to increased awareness regarding CSR initiatives, mitigating differences across cultures.
Theoretical Moderators—Interactions
When looking at the interaction between a country’s developmental level and the CSR initiative type (H5a and H5b), although the model was not significant (F(13,678) = 0.967, p = .483), when testing for the individual differences of the CSR initiative types, we found a significant interaction effect for environmental CSR, (t (691) = 2.890, p = .035), partially supporting H5a. Specifically, the interaction term with country development level was significant and positive (b = 1.095, p = .006), suggesting that in the case of environmental CSR initiatives, its impact is enhanced the higher the country development level. For example, if we consider Australia and Haiti (respectively the countries with the highest and lowest HDI in the dataset), the model predicts an effect size (r) of 0.46 in Australia and an effect size of −0.02 for Haiti, suggesting that environmental CSR initiatives are likely to have opposite effects in these countries. However, we found no significant interaction for the economic CSR type, which does not support H5b.
We then tested for H6, where we suggested an overall interaction of the three proposed moderators of CSR initiatives on consumer responses. Even though the general analysis did not yield a significant interaction (F (88, 603) = 1.177, p = .143), two significant three-way interaction effects partially supported this hypothesis. The first significant interaction was for social CSR initiatives, country development level, and consumer staples industry (t (691) = 2.26, p = .024). In this case, the interaction term combining consumer staples and social CSR was negative (b = −2.016, p = .017), but their interaction with HDI was positive (b = 2.242, p = .024). This means that CSR initiatives focused on social aspects reduced the effect on consumer response for consumer staples goods, but this effect is moderated by the country’s development level, such that it is positive in more developed countries, and the higher a country’s development level, the stronger it is. For example, this model estimates an effect size (r) for Australia of 0.52, and an effect size of −0.46 for Haiti.
A similar effect was found for ethical CSR initiatives, country development level, and the financial sector (t(691) = 2.46, p = .014). Again, the interaction term combining the financial sector and ethical CSR was negative (b = −0.940, p = .018) but the interaction with the country development level was positive (b = 1.256* the country’s HDI, p = .014). This means that CSR initiatives focused on ethical aspects reduced the effect on consumer response for CSR actions from the financial sector, with this effect being moderated by a country’s development level, such that the effect is enhanced the higher the country’s development level. For instance, this model would estimate an effect size (r) of 0.57 for Australia and an effect size of 0.34 for Haiti.
Methodological Moderators
Finally, we further tested whether specific characteristics of the studies’ methodology affect the overall impact of CSR initiatives on consumer responses. Experimental data, by controlling for the effect of external variables, can result in smaller effect sizes than correlational studies, which do not control for all the variables that can contribute to an effect size. The difference between these two methodological approaches can, therefore, impact the size of the effect of the relationship being studied. Importantly, results suggest that the type of study design (experimental vs non-experimental) does not affect the overall results (F (1, 742) = 0.464, p = .496).
Similarly, we also tested for differences between studies using students and non-students since the two groups of consumers could differ in terms of attitudes toward CSR, finding a significant effect (F (1,740) = 75.650, p < .001), such that the effect size of student samples (r = −.052) was lower than that of non-student samples (r = .420). Building on this evidence, we deepened the analysis and tested the interaction between sample type and CSR type, finding a significant interaction with the initiative environmental type (F(3, 372) = 41.2773, p < .0001). While for non-student samples, the type of CSR does to not seem to play a large role in determining their response to the CSR actions, for student samples, it does. Interestingly, results indicate that students’ samples respond much more to CSR initiatives focused on the environment than to other CSR initiatives, even if they generally seem to respond less to CSR than non-student samples. Specifically, the effect size for non-student samples varies between .400 for environmental CSR initiatives and .472 for non-environmental CSR initiatives, whereas for student samples and environmental CSR initiatives it is −.194 and for non-environmental CSR initiatives it is −.461.
Discussion
This meta-analysis aimed to offer a better understanding of how different CSR strategies can impact consumers’ attitudes and behaviors. By synthesizing existing research, our findings reinforce the importance of CSR to consumers, suggesting that companies benefit from incorporating CSR initiatives in their business practices. However, our results also add to prior meta-analyses on this phenomenon (see Table 1), highlighting that the generally positive effect proposed in previous studies is not uniform but instead shaped by several moderators, which can help explain the diversity of findings (both in valence and magnitude) across the literature.
Main Contributions
The current work provides several relevant contributions to the CSR literature. First, it offers a new quantitative synthesis of the CSR effects on consumers’ responses as reported in prior studies. By systematically organizing these findings it provides a comprehensive perspective on the varied impacts of CSR, underscoring the complexity of drawing overarching conclusions. Findings indicate that CSR initiatives have a positive and medium effect (r = .38), which suggests that companies engaging in CSR are likely to evoke a favorable consumer response. However, results also indicate that this effect differs greatly depending on the methodology used. Interestingly, when considering only the studies that use correlations, the effect is substantially higher (r = .44) than when considering papers that use beta coefficients (r = .23). This divergence underscores the importance of methodological choices in shaping conclusions about CSR’s impact—an insight valuable for both researchers and policymakers.
Second, it contributes to identifying the role of CSR conceptualization and consumer responses in shaping CSR’s effectiveness, distinguishing between attitudinal and behavioral responses, active versus passive CSR initiatives, and analyzing in detail the specific lower-level effects of each of these variables. Specifically, findings indicate that the overall relationship between CSR initiatives and consumer responses varies significantly, helping explain the mixed results in prior research. For example, when categorizing consumer responses into attitudinal and behavioral, our findings show that attitudinal responses tend to be more sensitive to CSR initiatives than behavioral ones, with more substantial consumer commitment. This result is consistent with the aforementioned attitude-behavior gap (Baron & Spranca, 1997; Nguyen et al., 2019). Interestingly, our findings indicate that this gap is impaired when consumers are exposed to active versus passive CSR initiatives, suggesting that consumers seem sensitive and reactive to companies’ voluntary efforts to foster social, environmental, and ethical developments. Furthermore, we found that different types of consumer responses react differently to CSR initiatives. Specifically, brand/company image appears to be more permeable to CSR initiatives, whereas quality perception, loyalty, and WTP are less affected by CSR actions. The results obtained for the brand/company image align with those from prior works, suggesting that brand image is directly and highly impacted by companies’ actions (He & Lai, 2014), whereas quality expectations and loyalty behaviors depend on many characteristics, such as WOM and satisfaction (Carden & Wood, 2018; Grewal, 1995), which might contribute to a lower susceptibility to change through CSR efforts alone. Similarly, the insight that WTP is less affected by CSR actions than other types of consumer response is also in line with previous research on the attitude-behavior gap, where consumers express strong CSR preferences, but struggle to translate them into values-aligned decisions (Baron & Spranca, 1997; Nguyen et al., 2019).
The third main contribution is the proposal and analysis of potential moderators (country development level, industry type, and cultural characteristics) that shape the relationship between CSR initiatives and consumer responses. Regarding country development, prior evidence suggests that in developing countries, consumer awareness of CSR activities is lower (Ramasamy & Yeung, 2009; Sen et al., 2006). However, our results suggest that, surprisingly, country development does not significantly moderate this relationship. Interestingly, when examining its interaction with CSR type, we find that environmental CSR actions have a weaker impact on consumer responses in lower-income countries. This aligns with previous studies showing that developing countries tend to place a higher value on CSR initiatives that benefit them financially (Ramasamy & Yeung, 2009), compared with other societal benefits. This finding is particularly relevant, as much of the literature suggests a strong impact of environmental CSR (Liu et al., 2014) without fully accounting for variations across economic contexts, which our work shows to be relevant to account for.
Concerning industry type, we found a significant interaction with CSR initiative type. When considering consumer discretionary industries, multidimensional CSR initiatives have a weaker effect on consumer responses, indicating that for discretionary purchases (i.e., non-essential goods), consumers are less influenced by CSR initiatives when multiple CSR dimensions are involved. This suggests that when companies invest in more than one dimension, it may hinder consumer justification of indulgent purchases (De Witt Huberts et al., 2014). Moreover, existing research suggests that CSR in controversial industries (e.g., Song et al., 2020; Yoon et al., 2006) tends to yield weaker consumer responses. However, our meta-analysis found no significant moderating effect of industry controversy. This finding challenges the dominant narrative that CSR efforts in controversial industries are less effective. One possible explanation is that the real-world impact of CSR initiatives may be influenced by factors beyond industry perception alone, such as the authenticity of CSR efforts (Palazzo & Richter, 2005) or the extent to which they align with stakeholder expectations (De Roeck & Delobbe, 2012). Likewise, the cultural characteristics of the countries where the studies were run, operationalized by their level of individualism, were also found not to moderate the hypothesized relationship between CSR and consumer responses. Despite some prior works suggesting that collectivistic cultures tend to value CSR actions and behave more positively toward brands engaging in them than individualistic ones (Moon et al., 2015), findings from more recent works (Aljarah & Ibrahim, 2020; Santini et al., 2021) also suggest a non-significant effect, supporting our results.
Fourth, this meta-analysis offers an integrated perspective on how CSR type, industry type, and country development level shape consumer responses. Our findings suggest that social CSR initiatives lead to a more positive response in the staples industry sector, the higher the country’s development level. As consumer staples tend to be psychologically closer to consumers since they encompass products of daily consumption, and social CSR actions focus on social and local issues, this association may be perceived by consumers as having a particularly good fit. However, in developing countries, this relationship weakens, possibly due to lower exposure to CSR initiatives from informal vendors, a typical means of acquisition of such goods in such countries (International Labor Office [ILO], 2017). Similarly, ethical CSR initiatives appear particularly effective in the financial sector, especially when the country’s level of development is higher. This might constitute a similar case of fit, in which the financial sector’s association with corporate scandals and dishonest behavior makes companies more likely to be well perceived when acting ethically (Soltani, 2014), especially in developed countries.
The fifth and last contribution is the recognition that methodological choices shape CSR effects. Specifically, sample type plays an important moderating role, with student samples responding less positively to CSR initiatives than non-student samples. This may seem surprising, especially considering prior findings that suggest that the younger generations, such as millennials or Generation Z, are very concerned with societal and environmental issues (Chatzopoulou & Kiewiet, 2020). However, younger generations are also more attuned to inconsistencies between brand messaging and actual behavior (Ahmad, 2019), making them particularly skeptical of CSR initiatives they perceive as disingenuous or as greenwashing (Bulut et al., 2021). This suggests that companies targeting younger consumers may need to go beyond traditional CSR efforts to foster engagement and trust.
Taken together, these contributions offer a more comprehensive and integrative understanding of the factors shaping consumer responses to CSR. By systematically distinguishing between different CSR types, response categories, and contextual moderators, our findings provide valuable insights for academic research and managerial decision-making, adding to the existing meta-analyses.
Managerial Implications
First, our results show that although CSR initiatives have a general positive effect on consumers’ responses, the magnitude of this effect can differ significantly, depending on the multiple characteristics of their implementation. This emphasizes the need for managers to move beyond a one-size-fits-all approach and carefully tailor their CSR policies to maximize effectiveness.
Second, our analysis highlights that the success of CSR initiatives is highly context-dependent. Factors such as the country’s level of development can shape consumer perceptions, meaning that managers must take a nuanced approach when selecting and designing CSR programs. Indeed, the effectiveness of social or environmental CSR actions may differ based on local market expectations and societal priorities.
Third, our results suggest that merely engaging in CSR is not enough—companies must ensure that their CSR efforts resonate with consumers. In some cases, CSR initiatives that lack perceived relevance or authenticity may fail to generate positive responses or, worse, backfire. This is, for example, the case of environmental CSR initiatives in less developed countries, where our findings suggest they may not always lead to favorable consumer reactions.
Fourth, current findings also stress the importance of considering the target group companies want to attract or involve when designing their CSR initiatives. The fact that the younger generations seem less influenced by the CSR actions employed by companies highlights the need to develop other engagement activities that go beyond or complement the CSR actions to strengthen their impact.
Finally, our findings also confirm the attitude-behavior gap reported in the literature, offering evidence that although consumers favorably evaluate CSR initiatives, they often do not engage in the corresponding behaviors. This is something critical that managers should be aware of, making an effort to design initiatives that enhance the likelihood of converting consumers’ attitudes into behaviors. Our results shed some light on this issue, showing that although exhibiting lower effects, there is space to grow regarding WTP and loyalty behaviors. Possible strategies to enhance such behavioral responses could be to put more emphasis on communicating the impact of CSR activities effectively to increase trust and knowledge regarding the company’s impact. Companies can also implement actions that directly request consumer engagement (e.g., co-participation in CSR actions), to foster confidence, raise awareness, and generate loyalty toward the companies.
Altogether, our findings contribute to a more profound knowledge regarding CSR strategies, which we hope contributes positively to managerial decision-making advancement and drives meaningful business outcomes.
Limitations and Future Research Directions
Despite the comprehensive scope of this meta-analysis, some limitations and future research directions are worth outlining, paving the way for future research opportunities. The first limitation is related to data constraints. As with any meta-analysis, the data used was extracted from studies, combining the correlation between variables related to CSR initiatives and consumer responses. While this approach provides valuable insights, it limits the ability to incorporate additional explanatory variables, such as the firms’ nature, which can be of interest when trying to understand the companies’ engagement in CSR (Matten & Moon, 2008). In addition, the COVID pandemic has changed consumer behavior (Crane & Matten, 2021; He & Harris, 2020), potentially altering responses to CSR initiatives in ways not yet fully captured in the literature. For instance, during lockdowns, consumers switched to online shopping (Ngoh & Groening, 2022), which might have altered their response to some CSR initiatives. As the current meta-analysis included articles up to 2021, which, in principle, due to writing-up and publication lag, does not include papers that study the hypothesized relationship in the post-COVID period, future research should explore the study of CSR impact further, comparing consumer responses pre- and post-COVID.
A second limitation concerns the choice of moderators used in the current analysis. The moderators included in the meta-analysis (CSR type, country development level, industry type, and cultural characteristics) were identified after a careful literature review. However, some other constructs, often not reported in primary studies, could also moderate the relationship between CSR initiatives and consumer responses, eventually providing additional insights. It could then be interesting to analyze, for example, the impact of individual differences on the evaluation of CSR initiatives (Gond et al., 2017) as socioeconomic status (Atalik & Eratik, 2015), gender (Jones et al., 2017), environmental consciousness (Pham et al., 2024), CSR skepticism (Nguyen et al., 2023), or even religious values (Ramasamy et al., 2010), since these may shape consumers’ responses to CSR initiatives. Therefore, we encourage future studies to report individual-level variables to foster richer meta-analytic work.
Besides exploring the diversity of consumers’ or firms’ characteristics on the CSR impact on consumer responses, expanding the scope of analysis to include further stakeholders’ responses (e.g., media and regulatory entities) to CSR initiatives and their impact on corporate branding (Maon et al., 2021) could be valuable. In addition, regulatory changes concerning CSR actions and the idea that businesses should serve society, advocated by several authors (Aguilera & Ruiz Castillo, 2025; Mayer, 2018; Schwab, 2021), may change the landscape and future CSR practices, calling for future research that attempts at understanding how these changes can impact consumer expectations regarding companies’ CSR actions.
In addition, recent works have highlighted the importance of encompassing ethical, environmental, and social dimensions as outcomes of good corporate governance (Aguilera & Ruiz Castillo, 2025). While in the past, corporate governance mainly focused on maximizing financial results for shareholders (Gillan et al., 2021), recent theories have highlighted the importance of adopting a broader stakeholder approach, incorporating both societal and environmental concerns, and developing specific governance metrics (Aguilera & Ruiz Castillo, 2025). Other authors have even emphasized that incorporating mandatory ESG and CSR practices in companies’ directives can spill over among peers, enhancing companies’ concerns for societal issues (Chen et al., 2025). Taking into consideration these evolving and dynamic corporate demands, future research is then needed to assess the critical role that the pursuit of purposeful CSR practices can have on consumers’ perceptions toward companies, since including ethical and unselfish CSR practices in firm strategies can contribute to enhancing consumers’ trust in companies and their initiatives. In a similar vein, also regarding controversial versus noncontroversial industries, future research may benefit from examining more nuanced mechanisms, such as the role of CSR communication strategies or stakeholder engagement, to clarify when and how industry controversy shapes consumer responses to CSR.
Conclusion
Altogether, our findings underscore the importance of CSR initiatives while emphasizing that their effectiveness is contingent on multiple factors. By identifying key moderators and methodological considerations, this meta-analysis advances both theoretical and managerial understanding of CSR’s impact, offering valuable insights for researchers, practitioners, and public policy makers.
Supplemental Material
sj-docx-1-brq-10.1177_23409444251368391 – Supplemental material for Meta-Analysis on the impact of Corporate Social Responsibility Initiatives on Consumer Attitudes and Behavior
Supplemental material, sj-docx-1-brq-10.1177_23409444251368391 for Meta-Analysis on the impact of Corporate Social Responsibility Initiatives on Consumer Attitudes and Behavior by Rita Coelho do Vale, Pedro Verga Matos and Filipa de Almeida in BRQ Business Research Quarterly
Footnotes
Acknowledgements
We also thank all the research assistants and post-docs who helped this paper come true, both in terms of searching and coding papers, as well as helping with the literature revision and insights: Anna Bernard, Saleh Shuqair, Vera Herédia Colaço, Eduardo Xavier, and Ian Scott. They all contributed in their own way to the current version of the paper.
Authors’ note
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We acknowledge FCT funding support with project PTDC/EGE-OGE/30345/2017, multi-year research funding UIDB/00407/2020 (CUBE), and UIDB/04521/2020 (ADVANCE/CSG).
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Research involving human participants
No experiments were run using human participants.
Informed consent
Not applicable.
Supplemental material
Supplemental material for this article is available online.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
