The Role of Individual- and Contextual-Level Social Capital in Product Boycotting: A Multilevel Analysis

Abstract

Product boycotts represent an important form of sustainable consumption, as withholding purchasing can restrain firms from damaging the natural environment or breaking social rules. However, our understanding of consumer participation in these protests is limited. Most previous studies have focused on the psychological and economic determinants of product boycotting. Drawing on social capital literature, this study builds a framework that explains how individual- and contextual-level social capital affects consumer participation in boycotts of products. A multilevel logistic regression analysis of 29 country representative samples derived from the European Social Survey (N = 54221) shows that at the individual level product boycotting is associated with a person’s social ties, whereas at the country level, generalized trust and social networks positively affect consumer decisions to take part in these protests. These results suggest that to better understand differences among countries in consumer activism, it is necessary to consider the role of social capital as an important predictor of product boycotting.

1. Introduction

A product boycott, also called a consumer boycott, ‘is an attempt by one or more parties to achieve certain objectives by urging individual consumers to refrain from making selected purchases in the marketplace’ [1] (p. 97). Thus, the essence of a product boycotts is the use of consumer power to impose an effect on the firm. These protests can target a wide variety of organizations and social concerns; however, they typically aim at firms that damage the natural environment or severely break social rules [2]. In other words, consumer boycotts tend to address socially irresponsible practices that is ‘those business behaviors and actions that are: illegal; or legal but severely unsustainable and/or unethical and thus totally socially unacceptable’ [3] (p. 8). By boycotting firms deemed to use irresponsible practices, consumers express their concerns and obligations for the environment and society [4]. Thus, boycotting products is regarded as an important aspect of ethical [5] or sustainable consumption [6,7,8].

Several boycotts have demonstrated that even the largest firms in the world must treat their consumers who do not accept corporate policy or action with respect. The most renown example of such a boycott is the case of Brent Spar, which occurred when consumers severely protested against the decision by Shell to sink an oil rig in the North Sea. Although Shell changed its decision, they lost between 20–50% of their revenue during that boycott [9]. More recently, because of US and British student protests, Fruit of the Loom had to reopen a Honduran factory that closed after its employees joined a labor union. As a consequence of that boycott, 1200 workers gave their jobs back, and the firm lost $50 million [10]. These examples clearly show that coordinated consumer protests can severely affect corporate sales and profitability. This is probably the reason why from 33% to 50% of firms facing a boycott meet the demands of activists and change their policy or actions [11].

The successful stories of boycotts suggest that consumer protests can perform an important role in dealing with irresponsible firms, that is, those entities that use illegal or legal but severely unsustainable practices [3]. However, boycotts as a form of introducing prosocial changes into the business world are not always effective. To be effective, a product boycott needs to be supported by consumers. Without consumers, a product boycott cannot be successful. The latest edition of the European Social Survey (2016) shows large differences among countries in the percentage of people who participate in product boycotts [12]. On the one hand, there are countries, such as Sweden and Iceland, with over 40 percent of the population boycotting a product in the last 12 months. On the other hand, in several countries, less than 1 in 10 individuals take part in this form of protests (e.g., Hungary, Italy, Poland). The low number of people who take part in product boycotts makes these campaigns ineffective. To enhance consumer activism and increase the effectiveness of boycotts, one needs to understand why there are some regions with so few people taking part in such protests.

Previous studies says little about contextual (regional or country) variables behind boycotting products. Most of them have focused on psychological and economic predictors of protesting against firms, including the relation of personal cost and personal benefits to the consumer [2,13,14,15], the perceived egregiousness of the act by the target firm [2,16,17], consumer reciprocity [18], emotional attachment to the boycotted product [15], amounts of targeted product being purchased by the consumer [2,15,19,20] and the availability of substitutes for the boycotted product [15], or the competitiveness of the industry [21].

The present study assumes that social capital as a feature of a country facilitates civic engagement, including product boycotts [22]. Past research, surprisingly, has paid little attention to the effects of social capital on boycotting, though a product boycott is a collective action problem [2]. To fill this gap, the present paper aims to extend our understanding of these protests by examining the role of social networks and generalized trust in boycotting products as two components of social capital. Given that social capital can shape behaviors not only at the contextual level but also at the individual level [23], this study examines the effects of the social capital of a country and the effects of the social capital of an individual. More specifically, the paper aims to answer two research questions:

R1: How does individual- and contextual-level social capital affect consumer participation in boycotts of products?

R2: How do social ties and generalized trust affect consumer participation in boycott of products?

The remainder of this paper includes a theoretical explanation of links between social capital and consumer participation in product boycotts. The research method is then presented together with a variable measurement, a sample description, and the statistical procedure. Next, the results of multilevel logistic regression analysis are provided. The paper ends with a discussion of the research findings, including implications for theory and practice, the study’s limitations, and directions for future studies.

2. Theoretical Framework and Hypotheses

Scholars offer multiple definitions of social capital. Putnam, in his seminal work, considers social capital as including “features of social organization, such as trust, norms, and networks, that can improve the efficiency of society by facilitating coordinated actions” [24] (p. 167). Bourdieu highlights that social capital lies in the less or more institutionalized relationships among members of the community. According to such a view, this collective asset helps individuals and groups of people to operate more effectively [25]. Whereas, Coleman while defining social capital focuses more on its functions than on its ingredients. In particular, he points out that social capital is not a single object, “but a variety of different objects having two characteristics in common: they all consist of some aspect of social structure, and they facilitate certain actions of individuals who are within the structure” [26] (p. 302). In other words, these often-cited definitions regard social capital as an intangible asset that enhances the outcomes of a society and its members. This asset is derived from the links among members of a society and its characteristics.

When considering the role of social capital in product boycotting, it is worth noting that participation in a product boycott is a complex and multidimensional phenomenon that involves the issue of collective action [2]. Neilson considers product boycotting as a form of political consumption, because by purchasing (or abstaining from purchasing), a consumer ‘votes’ for the firm and its actions [27]. Zhang [28], Copeland [29], Stolle et al [30] see in product boycotts a form of civic engagement (involvement) or prosocial behavior, as these protests typically aim at protecting the common good (e.g., natural environment, human rights). All of these features of product boycotting are associated with social capital. Hence, this study grounds its theoretical framework on social capital literature and the literature on prosocial behavior and political consumption.

When describing the effects of social capital on consumer participation in product boycotts, the study distinguishes between social ties (hereafter also called social networks) and generalized trust (hereafter also simply called trust), because they are the key components of social capital [24]. Given that social capital can be considered at different levels [23], the role of each component of social capital is explained at the individual and contextual levels.

2.1. Generalized Trust and Boycotting Products at the Individual Level

At the individual level, trust generally refers to “an expectation that the trustee is willing to keep promises and fulfil obligations” [31] (p. 6). McKnight and colleagues state that trust plays an important role in consumer decisions because it helps them to overcome risk and engage in “trust-related behaviors [...] such as sharing personal information or making purchases” [32] (p. 335). While participation in a product boycott means refraining from purchasing a certain product, making such a decision requires information. Thus, the present study assumes that information acts as the bridge between trust and boycotting products.

In a similar vein, scholars argue that trust may affect consumer behavior through its role in responding to information about products and firms [27,33,34]. According to such a view, consumers with higher levels of trust should be more likely to believe that a call for protest is motivated by the good intentions of the people who launch it, in turn enhancing boycott participation. In addition, believing that activists tell the truth about the emergence and existence of a certain boycott may make a consumer feel effective as part of a larger community. In line with this argument, Putnam showed, that people with greater trust in others are more likely to take part in civic activities than people with lower trust [35]. Altogether, one may reasonably expect that an individual who trusts more in other people will be more likely to take part in product boycotts. Hence, the following hypothesis is formulated:

Hypothesis 1 (H1).

A generalized trust positively influences consumer participation in product boycotts at the individual level.

2.2. Social Ties and Boycotting Products at the Individual Level

Participation in networks can help one gain information about an irresponsible firm and may mobilize members of a society to take part in protests. Numerous studies in the word-of-mouth field have shown that social connections are a valuable source of information in consumer decision-making. Goldenberg and colleagues demonstrate that information dissemination is dominated by word of mouth [36]. According to their research findings, weak and strong ties could be even much more effective than advertising. In other words, purchasing decisions are determined by the face-to-face communication of a consumer with his in-group and out-group colleagues. However, the power of networks is not limited to face-to-face communication. Steffes and Burgee demonstrated that the information gained from electronic word of mouth (a weak-tie information source) could also be influential [37]. The important role of online information sharing via social networks in consumer decision-making cannot be overlooked, as more and more boycotts are being launched on the Internet [10]. Considering all of the aforementioned arguments, one can expect that an individual with access to a larger social network will be more likely to take part in a product boycott than an individual with access to a small network. The following hypothesis expresses this expectation:

Hypothesis 2 (H2).

Social ties positively influence consumer participation in product boycotts at the individual level.

2.3. Generalized Trust and Boycotting Products at the Contextual Level

As mentioned earlier, social capital, including trust, can influence behaviors not only at the individual level but also at the contextual level, as a feature of community, region or country. When considering this type of effects, it is useful to note that Fukuyama understands trust as “the expectation that arises within a community of regular, honest, and cooperative behavior, based on commonly shared norms, on the part of other members of that community” [38] (p. 26). Trust is thus a characteristic of a community with its roots in the norms shared by members of that community. Several studies have proved that regions (or countries) with high levels of trust outperform regions with low trust in economic and societal outputs [39].

When explaining the role of trust in boycotting at the contextual level, it is also worth noting that organizing a consumer boycott involves a collective action problem [2]. In other words, a boycott may take place when an activist organization is able to persuade a group of consumers to abstain from using or buying a product of a particular firm at a certain point in time. Thus, to launch a boycott, a noncommercial organization must coordinate activities of those consumers and other entities involved in such a protest. While describing the various functions of trust in society, Putnam notes that “trust lubricates cooperation. The greater the level of trust within a community, the greater the likelihood of cooperation” [24] (p. 171). Therefore, in the context of product boycotts, one may reasonably expect that in regions with a high level of trust, more consumers are likely to take part in boycotts than in regions with a low level of trust.

As mentioned above, a consumer boycott could be considered a specific form of prosocial behavior. Such a protest typically aims at reducing the negative impacts of business operations on society and the natural environment. Thus, a consumer who abstains from buying the certain product contributes to a positive change in society. Several studies have proved the positive effects of trust on various forms of prosociality. For example, Glanville and colleagues showed that regions in which people highly trust each other are characterized by a much higher level of generosity than regions with low levels of trust [23].

A product boycott could also be seen as a specific form of civic engagement. For example, Friedman points out that boycotts are typically the result of actions taken by noncommercial organizations which initiate such protests to protect the common good [40]. Thus, by taking part in such protests, consumers fulfill their civic commitments. Past studies on civic engagement clearly demonstrate that the higher the level of trust is in a region, the higher the engagement of citizens in community activities and political activities [41]. Considering the collective nature of a product boycott, its links with prosocial behavior and civic engagement, the present study proposes the following hypothesis:

Hypothesis 3 (H3).

Generalized trust positively influences consumer participation in product boycotts at the contextual level.

2.4. Social Ties and Boycotting Products at the Contextual Level

As mentioned previously, though trust is an indispensable ingredient of social capital, it is not its only component. Without social ties, trust cannot be as effective as it is when joined to them. Putnam describes this catalytic role of ties as following: “social networks allow trust to become transitive and spread: I trust you, because I trust her and she assures me that she trusts you” [24] (p. 169). In other words, social ties enhance the effects of trust in the community. Given this relationship, one can expect social networks to positively influence boycotting products at the contextual level.

However, the positive effects of social ties on product boycotting do not stem only from its catalytic role in strengthening the effects of trust. Glanville and colleagues suggested that individuals in regions high in social ties would be more likely to engage in prosocial behaviors because they are exposed to more information and requests to donate time or money [23]. These authors noted that the positive impacts of networks at the regional level are not only due to the aggregation of networks at the individual level. In a highly integrated community, even an individual with a small network could be provided with a relatively high amount of information about prosocial campaigns and requests for support, as his network is better connected with those of other people. Taking into account that consumer participation in a product boycott is a type of prosocial behavior, and the catalytic role of networks on the effects of trust, the following hypothesis is proposed:

Hypothesis 4 (H4).

Social ties positively influence consumer participation in product boycotts at the contextual level.

3. Materials and Methods

The aforementioned hypotheses assume that the participation in a product boycott depends both on the social capital of an individual and on the social capital of society in which such an individual lives in. Thus, the research model includes two levels of analysis (Figure 1). The individual level consists of variables describing a person’s social capital components and the dependent variable that is consumer participation in product boycotts, whereas the contextual level focuses on the social capital of a country. Following past studies on the effects of social capital [23,39,42], the research model includes several control variables. The inclusion of these variables enables to see whether the effects of social capital also hold when controlling for other variables that may influence consumer decision to join a product boycott. More specifically, their impacts on the relation between social capital and consumer participation in product boycotts are removed (i.e. the effects of control variables are kept constant). Individual-level control variables were chosen in line with the literature, suggesting that age, gender, education level, place of residence and household income are predictors of prosocial behaviors [43]. With regard to control variables at the contextual level, the present study follows scholars who note that the availability of substitutes for the boycotted product positively affects consumer activism [15]. Given that consumers in more developed countries have easier access to substitutes of a boycotted product than in less developed ones, the study controls for GDP per capita as a proxy for economic development. To have a better control for the economic conditions under which a consumer lives, the study also takes into account a country’s unemployment level. In addition, the study controls for education level, as past research suggested that higher education may enhance civic engagement [44]. Thus, the gross enrolment ratio was used as a proxy for education level. Lastly, the study controls for a country’s individualism (-collectivism). This dimension of Hofstede’s model of national culture expresses the degree of interdependence among members of a particular society [45]. Given that past studies have shown a significant association between in-group individualism and the risk of boycott occurrence [46], the inclusion of that variable may help reduce model errors.

3.1. Data and Sample

In order to test hypotheses, the present study uses data from the European Social Survey (ESS). The ESS is funded by the European Commission, the European Science, and the National Science Foundations. Its aim is to measure and monitor the changing attitudes, beliefs and behaviors of various European populations. This large international project has been repeated every two years since its creation in 2001. The ESS draws on large samples of subjects aged 15 and over from the noninstitutionalized population. The carefully translated questionnaire into national languages and the equivalent sampling methods used in all participating countries ensure multinational comparability [47]. The sampling is conducted in two steps. In the first stage, the scientific team of ESS provides national partners with the general principals and requirements with regard to the sampling methods that have to be probabilistic [48]. Then, national ESS teams select samples separately. Detailed sampling procedures, including designs, stratifications and the number of sampling stages, for all the participating countries in the sixth round of ESS, can be found in ESS6—2012 Documentation Report [49].

It also worth noting that the ESS Core Scientific Team places a high importance on methodological standards. They carry out a wide range of activities to ensure high quality of the collected data. For example, in the sixth wave of the ESS that is used in this study as a source of data, they conducted four Split-Ballot Multitrait-Multimethod (SB-MTMM) experiments to evaluate the quality of survey questions. The detailed results of that assessment can be found in a report by DeCastellarnau and colleagues [50]. More information on validity and reliability is provided in the following website: https://www.europeansocialsurvey.org/methodology/.

The data for the current study were downloaded from the ESS website [51], which provides open and free access to collected data in the European Social Survey. More specifically, the study used multilevel data from the sixth wave of the ESS. That round was selected because it covered the largest number of countries from all ESS editions. The relatively large number of countries taking part in this survey enabled the testing of hypotheses on more culturally diversified data. The joint sample included 54,221 participants from 29 countries: Lithuania, the United Kingdom, Kosovo, Norway, Hungary, Bulgaria, France, Belgium, Germany, Finland, Denmark, Ireland, the Czech Republic, Cyprus, Sweden, Israel, Slovenia, Slovakia, Iceland, Italy, Poland, Portugal, Ukraine, the Russian Federation, Switzerland, Spain, the Netherlands, Albania, and Estonia.

With regard to collecting data in the sixth wave of ESS, the fieldwork period lasted four months (from September to December 2012). The data in that round were gathered by using face-to-face interviews. The national samples ranged from 752 participants (Iceland) to 2958 participants (Germany). Respondents were on average 48.31 years of age (SD  = 18.59). Females constituted 52.6% of all participants. When comparing these statistics with the EU population, one can see a slightly higher percentage of women (52.6% vs. 51.3%) [52] and a greater median age of respondents in the sample than in the EU population (48.2 vs. 41.6) [53].

Table 1. Descriptive statistics of the sample and measures.

Variable	Mean	Percent	SD	Range
Generalized trust	5.10		2.01	0–10
Social ties	4.79		1.63	1–7
Age	48.31		18.59	15–103
Education (years of full-time education completed)	12.54		4.03	0–51
Gender (female %)		52.6
Domicile
Big city		23.7
Suburbs or outskirts of big city		10.4
Town or small city		31.0
Country village		29.8
Farm or home in countryside		5.1
Income (feeling about household’s current income)
Living comfortably on present income		23.7
Coping on current income		42.8
Difficult on current income		22.5
Very difficult on current income		11.0
Average trust	5.09		0.84	3.54–6.82
Average social ties	4.79		0.53	3.43–5.74
GDP per capita (in thousand US dollars)	32.8		21.00	2.86–99.25
Gross enrolment ratio	68.08		10.60	43.56–94.05
Unemployment rates	9.34		3.94	3.20–21.60
Country individualism	57.27		18.68	20.00–89.00

Notes: SD—standard deviation.

provides more information on the descriptive statistics of a sample and measures.

3.2. Measures

3.2.1. Dependent Variable

Among a variety of variables, the ESS provides data on consumer participation in product boycotts that is a dependent variable in subsequent analysis. These data, by means of a question, ask whether respondents had taken part in activities to ‘improve things’ or ‘prevent things from going wrong’ in their country in the last 12 months. Among several options, a respondent was to indicate if he or she ‘boycotted certain products’ in the last 12 months on a yes (1)–no (0) scale [54].

3.2.2. Independent Variables at the Individual Level

Independent variables comprise of social capital components and a few control variables. As for social capital, generalized trust and social ties were selected to be included in the analysis, as academics typically consider them the two main pillars of social capital. Generalized trust was measured by an index derived from the respondent’s degree of agreement to the following statements: (a) “Most people can be trusted, or you can’t be too careful in dealing with people”; (b) “Most people would try to take advantage of you if they got the chance or try to be fair”; (c) “Most of the time people try to be helpful or they are mostly looking out for themselves.” The responses to these statements were measured on a 0 to 10 scale [54]. The internal consistency of that index was satisfactory (α = 0.782).

Social ties were measured using the question “How often do you socially meet with friends, relatives or colleagues?” In answering this question, respondents chose from the following seven response categories: ‘never—1’, ‘less than once a month—2’, ‘once a month—3’, ‘several times a month—4’, ‘once a week—5’, ‘several times a week—6’, and ‘every day—7’ [54].

Aside from the variables found in the research hypotheses, the subsequent analysis comprises of several control variables such as age, gender, education level, place of residence and household income, which are predictors of prosocial behaviors [43]. In terms of the operationalization of the aforementioned variables, age was measured in years. Similarly, education was also measured by asking a respondent to provide the number years of completed full-time education. Place of residence was measured using the following five categories: ‘Big city’, ‘suburbs or outskirts of a big city’, ‘town or small city’, ‘country village’, and ‘farm or home in the countryside’. Lastly, income was assessed by asking respondents about their feelings on household income. To answer that question, one had to choose among four categories: ‘living comfortably on present income’, ‘coping on present income’, ‘difficult on present income’, and ‘very difficult on present income’ [54].

3.2.3. Independent Variables at the Contextual Level

According to the main objective of the study, independent variables also comprise of exploratory variables at the country level. To measure social capital at a higher level, the study averaged individual indices of general trust and social ties separately for each country. Given that contextual-level indicators of social capital were derived from individual-level indicators, the study group mean centered individual components of social capital by subtracting country means from generalized trust and social ties indices. This procedure allowed for the separation of the effects of individual-level social capital from the effects of country-level social capital on product boycotting [55].

As in the individual-level analysis, the study also comprises of control variables at the contextual level including GDP per capita, unemployment, education level and individualism of a country. GDP per capita is measured at current prices in US dollars. The data on this indicator came from the United Nations Statistics Division. Unemployment was operationalized by the indicator of unemployment rates by all ages in a country. Data on that indicator came from Eurostat, with the exception of Albania, Israel, the Russian Federation and Ukraine. Information on unemployment in these four countries was downloaded from the World Bank Web Site [56]. The education level was operationalized by the indicator of gross enrollment ratio for the second stage of tertiary education. The indicator mentioned is provided by the United Nations Educational, Scientific and Cultural Organization [57]. Lastly, the data on country individualism were downloaded from the personal website of Geert Hofstede [58]. Appendix A provides specific data on contextual variables at the country level.

3.3. Method

To test the particular hypotheses, the present study used a multilevel logistic regression analysis because this statistical technique allowed for the simultaneous estimation of individual and contextual-level effects of the social capital. Accordingly, the data were hierarchically organized with respondents nested within countries. Given that boycotting product was operationalized by a dichotomous variable, the multilevel logistic regression was applied to estimate models used to test the research hypotheses. In order to take into account the national contexts, the study assumed intercepts to be random at country level in all models. In addition, models 2 and 3 include random effects for both ingredients of social capital. All the regression models were run in SPSS, by using the mixed command. The results are presented as regression coefficients along with odds ratios (ORs) and their 95% confidence intervals (CIs).

4. Results

Table 2. Summary of multilevel regression analysis.

Independent Variables	Model 1		Model 2		Model 3
	b	OR [95% CI]	b	OR [95% CI]	b	OR [95% CI]
Intercept	−2.036 ***	0.131 [0.087, 0.196]	−3.593 ***	0.028 [0.016, 0.049]	−9.070 ***	0.000 [0.000, 0.003]
Individual-level variables
Generalized trust			−0.004	0.996 [0.973, 1.020]	−0.007	0.993 [0.969, 1.018]
Social ties			0.043 **	1.044 [1.016, 1.073]	0.048 ***	1.049 [1.022, 1.076]
Age			0.005 ***	1.005 [1.001, 1.008]	0.005 *	1.005 [1.001, 1.008]
Years of education			0.111 ***	1.117 [1.096, 1.139]	0.111 ***	1.117 [1.095, 1.139]
Female (reference - men)			0.113 **	1.119 [1.034, 1.212]	0.127 **	1.136 [1.052, 1.227]
Domicile (reference—big city)
Suburbs or outskirts of big city			−0.053	0.949 [0.866, 1.04]	−0.075	0.928 [0.858, 1.002]
Town or small city			−0.145 *	0.865 [0.767, 0.975]	−0.154 *	0.857 [0.763, 0.964]
Country village			−0.231 ***	0.794 [0.697, 0.904]	−0.251 ***	0.778 [0.683, 0.887]
Farm or home in countryside			−0.087	0.916 [0.794, 1.057]	−0.100	0.905 [0.785, 1.042]
Income (reference—living comfortably)
Coping on current income			−0.061	0.941 [0.870, 1.018]	−0.055	0.947 [0.872, 1.027]
Difficult on current income			−0.023	0.977 [0.871, 1.096]	−0.048	0.954 [0.853, 1.066]
Very difficult on current income			−0.180	0.835 [0.684, 1.020]	−0.162	0.851 [0.686, 1.055]
Country-level variables
Average trust					0.602 *	1.825 [1.059, 3.148]
Average social ties					0.797 **	2.219 [1.236, 3.984]
GDP per capita					0.000	1.000 [1.000, 1.000]
Gross enrolment ratio					−0.033 *	0.968 [0.942, 0.994]
Unemployment rates					0.025	1.026 [0.968, 1.087]
Country individualism					0.006	1.006 [0.988, 1.025]
Random effects
Intercept $({\mathsf{\sigma }}_{\mathrm{u}0}^2$)	1.233 ***		1.148 ***		0.545 **
Generalized trust (${\mathsf{\sigma }}_{\mathrm{u}1}^2$)			0.003 *		0.003 *
Social ties (${\mathsf{\sigma }}_{\mathrm{u}2}^2$)			0.003		0.002
Adjusted Akaike	285,516.7		275,317.9		263,800.1

Notes: * p < 0.05, ** p < 0.01, *** p < 0.001; OR—odds ratio; 95% confidence intervals in square brackets. The number of observations in Model 1 is 54.221; the number of observations in Model 2 is 51.979 and is 49.696 in Model 3.

presents the results of multilevel logistic regression analysis. Model 1 was estimated on 54,221 respondents nested within 29 countries. This is an ‘empty model’, as it includes only the constant term in both the fixed and random parts. The odds ratio in this model for the fixed effect is 0.131 (p < 0.001), which means that the average predicted probability across countries that an individual takes part in a product boycott is approximately 12% [0.131/(1 + 0.131)]. The random effect in this model shows a significant between-country variance in individual participation in product boycotts (${\sigma }_{u0}^2$ = 1.233, p < 0.001).

Model 2 assessed the effect of individual-level predictors on the probability that an individual would take part in a product boycott. After excluding observations with missing variables, this model included 51,979 respondents nested within 29 countries. The results showed that, contrary to hypothesis H1, trust in other people was not a significant predictor of consumer participation in a product boycott (OR = 0.996, p > 0.05). It is worth noting that the random effect of individual trust was statistically significant, denoting the existence of the between-country variance in the impacts of trust (${\sigma }_{u1}^2$ = 0.003, p < 0.05). With regard to the other component of social capital, the study found that the more social ties an individual has, the more he or she is likely to boycott products (OR = 1.044, p < 0.01). This finding supports hypothesis H2 predicting that social ties positively influence consumer participation in product boycotts at the individual level. The random effect of social ties was insignificant, suggesting that the effects of social ties on boycotting remain largely stable across countries (${\sigma }_{u2}^2$ = 0.003, p > 0.05).

In addition to trust and social ties, Model 2 also comprised of five sociodemographic variables as controls, including age, gender, the number of years of completed education, domicile, and feelings about household income. Most of them turned out to be statistically significant. More specifically, women had an approximately 12% increase in the odds of boycotting than men (OR = 1.119, p < 0.01). Increasing age was associated with a greater likelihood of joining boycotts (OR = 1.005, p < 0.01). In the same vein, education positively influenced consumer participation in a product boycott (OR = 1.117, p < 0.001). More specifically, a single year of completed education increased the odds of boycotting by 11.7%. In terms of place of residence, people who lived in a small town or city (OR = 0.865, p < 0.05) and in a country village (OR = 0.794, p < 0.001) had lower odds of boycotting than respondents who lived in a big city. Surprisingly, the regression coefficient for income was statistically not significant. This insignificant effect could occur because income was measured in a subjective way that is by feelings of the respondent.

The purpose of Model 3 was to examine whether social capital variables affect boycotting products also at the country level. Thus, in addition to individual-level predictors, this model includes aggregated trust and social ties as components of social capital of a country. Given that data on national individualism for Cyprus and Kosovo were unavailable, Model 3 was estimated on 49,696 observations nested within 27 countries. The results showed that the effects of individual level predictors still hold after adding country-level variables (Table 2). With regard to the effects of contextual variables, the study found that both components of social capital significantly affected an individual’s decision to participate in a product boycott. In line with hypothesis H3, trust positively influenced consumer participation in product boycotts at the contextual level (OR = 1.825, p < 0.05). That is, a one-point gain in the trust index resulted in an 82.5% increase in the odds of boycotting. In regard to social ties, a single category in the scale of social ties increased the odds of boycotting by 120% (OR = 2.219, p < 0.01). Thus, Model 3 provided support for our hypothesis (H4) that social ties positively influence boycotting products at the contextual level. It is also worth noting here that the effects of country-level social capital were much greater than the effects of individual-level social capital. This applies to both components of social capital, that is, social trust (OR country = 1.825 vs. OR individual = 0.993) and social ties (OR country = 2.219 vs. OR individual = 1.049).

To control the effects of social capital at the contextual level, the model included GDP per capita, rate of unemployment, the gross enrollment ratio of the second stage of tertiary education, and the level of country individualism. Three in four of these variables were statistically insignificant. Surprisingly, gross enrollment ratio was negatively associated with boycotting (OR = 0.968, p < 0.05). This small negative effect was probably caused by high rates of gross enrolment in East European countries, which simultaneously have low percentage of people who boycott products (e.g., Russia, Ukraine, Poland).

With regard to random effects, the inclusion of national-level variables in Model 3 has a substantially reduced variation in intercepts between countries. More specifically, the variance dropped by more than 50% (${\sigma }_{u0}^2$ = 0.545 vs. ${\sigma }_{u0}^2$ = 1.233) when comparing Model 2 with Model 3. However, the random effect of the intercept remained significant (${\sigma }_{u0}^2$ = 0.545, p < 0.01), suggesting that social capital is not the only phenomenon linked to product boycotting at the country level. In regard to other random effects, Model 3 showed significant between-country variance in the trust effects (${\sigma }_{u1}^2$= 0.003, p > 0.05) and stable effects of social ties across countries (${\sigma }_{u2}^2$ = 0.002, p > 0.05).

5. Discussion

The present study aimed at examining how individual- and contextual-level social capital affects consumer participation in boycotts of products. Thus, the discussion of research findings addresses the effects of social capital at both individual and contextual levels. In addition, the limitations are presented with some directions for future research.

With regard to the effects of social capital at the individual level, past studies suggest that several components of a person’s social capital may be linked to boycotting. For example. Newman and Bartels showed that active engagement of a person in civic initiatives for the community in which one lives in, and donating to non-commercial organizations, are significant predictors of boycotting and/or buycotting by such a person [59]. Similarly, Neilson and Paxton found that generalized trust, frequency of social meetings, and association involvement are sources of political consumerism [60]. However, it is not clear whether individual social capital truly affects boycotting, as these studies have adopted the consumer voting metaphor [61], that is, political consumerism, and operationalized boycotting and buycotting together using one variable. Following the latest research, this paper focuses solely on boycotting, as boycotting and buycotting differ in their motivational underpinning [62]. The results of this study replicate the findings of Neilson and Paxton by showing the positive effects of social ties on boycotting [60]. In addition, the present study proved that the effects of social ties are stable across countries, as the random effect of social ties was insignificant.

Contrary to the predictions in hypothesis H1, the results of this study suggest that trust of an individual in other people cannot be regarded as a significant predictor of boycotting everywhere. This result seems to be caused by significant between-country variation in the trust effects. The existence of such variation is supported not only by the present research, but also by other studies. For example, using pooled data from the 2004 International Social Survey Program, Zhang found that the level of political rights in a country moderates relationship between trust and boycotting, that is in countries with high level of political rights, trust positively affects the participation in boycotts, whereas in countries with a relatively lower level of political rights, the effects of trust are insignificant [28].

Concerning the effects of social capital at the contextual level, the present study provides new insights into the relationship between the social capital of a country and participation of citizens in product boycotts. More specifically, the results of this paper showed that both components of social capital of a country, that is, generalized social trust and social ties, positively affected an individual’s participation in product boycotts. In other words, this study shows that individuals are more likely to join product boycotts in countries with a greater rather than smaller level of social capital. To the best of our knowledge, this is the first study to address the effects of social capital on product boycotting at the country level. Past studies have examined the role of social capital in political consumption (i.e., boycotting and buycotting) at the regional level [60].

Although the current paper provides theoretical and practical contributions, it is not free from limitations. For example, to test the hypotheses, the study used data from the ESS survey. Indeed, ESS is a project that generates high-quality data; however, this survey is not dedicated to measuring sustainable consumption. The individual’s participation in product boycotts was measured using a 0-1 scale. Thus, the study did not take into account how frequently a respondent boycotted products. In addition, there are some concerns about social ties’ measurement in ESS. For example, Yang argued that national social surveys actually measure only some features of a respondent’s social ties, as the precise assessment of this component of social capital requires a small-scale survey on a limited number of respondents [63]. To move beyond these limitations, future studies may carry out primary research to collect data on sustainable consumption and social capital.

6. Conclusions

The present study provides several insights into the literature on sustainable consumption, as it extends our understanding of consumer participation in product boycotts. More specifically, its theoretical contribution is twofold: first, the study showed that both components of social capital of a country, that is, generalized trust and social ties, positively affected an individual’s participation in product boycotts; second, the study found positive effects of individual-level social ties on boycotting; whereas the individual’s trust in people was neutral to the participation in product boycotts.

In addition to theoretical contributions, the present paper offers some practical implications. First, the positive effects of social ties on consumer participation in product boycotts suggest that word-of-mouth communication can be an effective channel to mobilize consumers to take part in protests against irresponsible firms. Second, the significant effects of social capital at the country level suggest that an activist organization wishing to launch a product boycott aimed at an international company may find greater support among citizens in a country with a high capacity of social capital. Thus, consumers in countries like Sweden, Denmark or Norway would be more likely to participate in a boycott than consumers in countries with low levels of social capital such as Albania, Slovakia, Poland or Ukraine. Third, the research findings of this paper may also be of some interest to policymakers as they suggest that investments in the development of social capital may also bring positive effects on sustainable consumption by increasing an individual’s likelihood to take part in protests against socially unacceptable business activities.