1 Introduction

Perceived organizational support (POS) is an employee’s “perception of being valued and cared about by the organization” (Eisenberger et al. 1990, p. 52). Support from the organization has been shown to be consequential for a host of positive work outcomes, including engagement, task performance, and citizenship (Kwon et al. 2010; Qi 2005; Eisenberger and Stinglhamber 2011 for a recent narrative; Rhoades and Eisenberger 2002 for a meta-analytic review). With the POS research domain becoming more mature, it also becomes ripe for new theoretical insights (Kwon et al. 2010; Qi 2005). For example, in a recent narrative review, Baran et al. (2012) note that “to date only a few of the relationships explicated in organizational support theory (OST) have been tested internationally.” More importantly, these authors also note that “future research on POS should continue to explore whether the relationships hold across cultures, whether they differ in strength, whether new antecedents or outcomes are relevant, and why” (p. 139). This study responds to such calls for research by offering theoretical clarifications supported by meta-analytic evidence (Bausch and Krist 2007; Reus and Rottig 2009; Stahl and Chua 2012; Tan and Sousa 2013).

Specifically, we ask: is the relationship between perceived organizational support and employees’ citizenship behaviors modified by cultural factors? Employees’ organizational citizenship behaviors (OCB) are defined as “contributions to the maintenance and enhancement of the social and psychological context that supports task performance” (Organ 1997, p. 91; Organ et al. 2006). Within the broader theoretical landscape of social exchange theory, in which the POS–OCB relationship is typically embedded, the relationship can be viewed from at two, often competing, perspectives—social psychology versus anthropology (Ekeh 1974, p. 44–46; Parsons 1961). One perspective, presented by selected schools of social exchange thought in social psychology, is that employees act as citizens in reciprocity to perceived support from organizations for their individual interests (Blau 1964; Gouldner 1960; Homans 1958). If employees perceive that their self-interests are supported by the organization, they will reciprocate with citizenship behaviors. Much of this social exchange perspective can be traced to Homans’ (1958, 1961) emphasis on conscious self-interests in reciprocity. Blau (1968) also noted that “many aspects of social life do reflect an interest in profiting from social interaction, and these are the focus of the theory of social exchange” (p. 452). This “self-interests” perspective is based on the belief that satisfying the psychological needs of the individuals helps promote positive social behavior (Ekeh 1974) and presumes a “contribution of self-interest to social morality” (Janowitz 1967, p. 638).

Focusing on a different aspect, a number of anthropologists honed in more pointedly on cross-cultural variations in reciprocity during social exchange. They posited that cultural beliefs of the larger society (assumed to be embedded in its members) strengthen or weaken moral reciprocity in social exchanges (e.g., Hofstede and Bond 1984; Hofstede and McCrae 2004; Lévi-Strauss 1949, 1969; Mauss 1954). That is, people who engage in social exchange “do so as part of society” (Ekeh 1974, p. 33). This perspective can be traced back to Lévi-Strauss (1949), who emphasized that reciprocity is embedded in the generalized moral norms of the society. Lévi-Strauss explained that it is “the human mind, buried down deep in common humanity, unknown even to actor himself, that provides us with an explanation of social phenomena” (Ekeh 1974, p. 40). Society is seen as a “preexisting matrix of trust and moral obligation which couches all transactions” (Stolte 1975, p. 396). Social exchanges are no exception: as everything is embedded within socio-cultural beliefs, “social exchange is a supraindividual process and individual interests may be involved in it but they cannot sustain a social exchange process” (Ekeh 1974, p. 43). In such exchanges, the morality resulting from reciprocity is culturally determined (Lévi-Strauss 1969, p. 138). Furthermore, the “basis of moral action is general,” and the generic cultural beliefs of society form the “bedrock” that enables moral outcomes from reciprocity (Mauss 1954, p. 68).

The arguments we have highlighted above are not intended to suggest any major disciplinary biases. After all, both social psychologists and anthropologists, when brought together, would probably acknowledge that the POS–OCB relationship is an important relationship and that it could vary among cultures (Jackson 1988). Moreover, it would be difficult to argue that social psychologists do not believe in culture, or that anthropologists don not accept that some relationships are at least in the same direction across cultures. Yet the arguments reveal some history of theoretical tension (see also Faye 2012) and provide useful pointers on how our research question can theoretically contribute to the social exchange literature. In an attempt to further integrate these historical schools of thought, we ask: how can these perspectives be brought closer to one another?

Our integrative effort starts by accepting that self-interests play a role in social exchanges between the organization and its employees: employees are likely to act as good citizens in reciprocity to the (perceived) support provided by their organization for their individual interests. We also accept the argument that society couches and hence modifies the strength of these exchanges. The integrated perspective offered in this study captures both aspects: the relationship between perceived organizational support and employees’ organizational citizenship behaviors is embedded within—and hence moderated by—the cultural dimensions of the larger society.

To test our hypotheses empirically, we rely on country-level cultural dimensions as moderators in a meta-analysis of the relationship between organizational support and citizenship behaviors in their affiliative (prosocial) forms (Eisenberger et al. 1986; Organ et al. 2006). In the light of findings indicating that cultural dimensions influence or modify a number of phenomena of importance in work organizations (e.g., Atwater et al. 2009; Li and Cropanzano 2009; Rockstuhl et al. 2012; Shao et al. 2013; Taras et al. 2010), our objective is to extend current knowledge by exploring the extent to which the POS–OCB relationship varies across cultures. We focus on the cultural dimensions outlined by Hofstede (1980, 2001) as moderators: individualism–collectivism, masculinity–femininity, power distance, and uncertainty avoidance. A recent meta-analysis confirmed that these cultural dimensions influence a large number of organizational outcomes (Taras et al. 2010).

We extend, in several ways, the knowledge base built by prior studies. First, researchers theorized and found support for an overall positive relationship between POS and OCB (Baran et al. 2012; Riggle et al. 2009; Rhoades and Eisenberger 2002). Despite the apparent universality of this relationship, can researchers assume that its strength is invariant across cultures? Researchers have provided both meta-analytic results (Taras et al. 2010) and called for research on how cultural factors are related to organizational citizenship (Gelfand et al. 2007; Organ et al. 2006, p. 138). Practically, this is a stringent need in a world where engaging in business beyond one’s local culture is pervasive (Bartlett and Ghoshal 1998).

Second, our thinking about the POS and OCB constructs might need to change to accommodate the fact that these constructs do not exhibit a stable relationship in different cultural contexts. Meta-analytic tests are necessary to start addressing mixed findings originating from primary studies across the world (Bausch and Krist 2007; Reus and Rottig 2009; Stahl and Chua 2012; Tan and Sousa 2013). If guidance is sought in the existing literature, the likely role of the cultural context in the POS–OCB relationship is either unclear or puzzling. Power distance is one such example. As substantiated in existing work, some forms of social exchange generate increased citizenship under high power distance conditions (Begley et al. 2002; Botero and Van Dyne 2009). In other research, however, the same relationship is stronger in cultural settings with low levels of power distance (Farh et al. 2007). A test across multiple samples and contexts, as performed in the current study, can clarify existing ambiguities while stimulating more precise theory development.

While the advantages of meta-analysis are well known (Rosenthal and DiMatteo 2001), of particular importance when testing moderators is its ability to detect (otherwise unaccounted for) variation among subgroups. In doing so, we join a growing trend where meta-analyses are used to examine how important organizational outcomes are differentially influenced across cultures by organizational aspects including fairness (Li and Cropanzano 2009; Shao et al. 2013), quality of relationship with the supervisor (Rockstuhl et al. 2012), and job satisfaction (Ng et al. 2009). We also provide more actionable information than heretofore available, especially to organizations operating globally and thus subject to cultural influences.

Third, our study has implications for how cultural influences can be considered when designing organizational practices and interventions. International human resource management (HRM) interventions and practices have been seen as important drivers of perceived organizational support (e.g., Eisenberger et al. 2004; Eisenberger and Stinglhamber 2011), and their fit may vary across different cultural contexts (e.g., Ramamoorthy and Carroll 1998). Organizations across the world, driven by the need for standardization of international HRM policies, may develop similar organizational procedures to support employees. However, such an approach may not be appropriate due to variations in cultural contexts. Accounting for cultural differences can help shape better interventions, in addition to refining existing theoretical explanations.

Finally, our research can redirect the attention of perceived organizational support and citizenship behavior researchers toward a more explicit integration of cultural dimensions. In fact, numerous calls for research on the topic under investigation here have been issued, by both specialized and cross-cultural scholars. We already mentioned Baran et al.’s (2012) call for cross-cultural studies on organizational support. Cultural influences have been also posited by organizational citizenship researchers, who focused on the need to “investigate whether the antecedents and mediators of OCB differ across cultures” (Organ et al. 2006, p. 138). While cross-cultural researchers noted how critical it is to “broaden the study of cultural values as moderators” (Taras et al. 2010, p. 435, italics added)—a general recommendation, some of their other suggestions are highly specific. Gelfand et al. (2007) for example remarked that “antecedents of OCBs also vary across cultures” (p. 487). Perhaps the clearer guidance for the current study is provided by Kirkman et al. (2006) who proposed that “one might also reasonably hypothesize that culture moderates the relationship between a number of inputs and OCB” (p. 310). In addition to providing guidance for useful future investigations, such calls for research also point toward the paucity of cross-cultural research for our relationship of interest (see Baran et al. 2012, pp. 134; Gelfand et al. 2007, pp. 487–488; Kirkman et al. 2006, p. 310). Despite this quasi-absence of research on our relationship of interest in primary studies, in the light of empirical findings indicating that cultural dimensions can modify a number work outcomes (e.g., Taras et al. 2010, 2012), we examine how cultural dimensions modify the relationship between POS and OCBs, and provide hypotheses in the next section.

2 Hypotheses

2.1 Individualism–Collectivism

The literature suggests that the social system of collectivist cultures generate a well-integrated network of relationships and exchanges, where collaboration, interdependence, in-group harmony, and interpersonal relationships are expected (Triandis 1995). A society with collectivist beliefs is characterized by “a tight social framework in which people distinguish between ingroups and outgroups, they expect their ingroup to look after them, and in exchange for they feel they owe absolute loyalty to it” (Hofstede 1980, p. 45). In collectivist cultures, an individual’s responsibility toward his or her collective’s interests becomes more important than and takes precedence over the individual’s self-interest, and collective or interdependent self-construal take precedence (Thomas et al. 2003).

Due to their tendency to sacrifice for the common good and maintain harmonious relationships (Triandis 1995; Thomas et al. 2003), it is probable that employees in countries with collectivist cultures would be more sensitive to and more appreciative of organizational support. After all, according to past research findings, collectivistic employees are more likely to identify with their organization’s interests (Markus and Kitayama 1991), be extra sensitive to the support received from it (Eisenberger et al. 2004), become loyal to the organization, and reciprocate favors more readily (Hofstede and Bond 1984). In other words, employees in collectivistic cultures personify and anthropomorphize (attribute human form or personality to) the organization, making it as their own, and most likely do so to a greater extent than employees situated in individualistic cultures (Hofstede 1980). Indeed, research suggests that in collectivistic cultures, people often reciprocate with generosity that goes beyond quid-pro-quo principles (e.g., Morris and Leung 2000; Triandis 1995). Thus, when employees in collectivist countries receive organizational support, they feel that they owe much more in return to their organization (Paine and Organ 2000) than employees in individualistic cultures. As a result, they will reciprocate with citizenship behaviors with greater enthusiasm and intensity (Taras et al. 2010). Employees’ perceptions of organizational support are based on their evaluations of the organization’s human resource systems (Eisenberger et al. 2004). In collectivistic cultures, research has found that employees prefer systems of performance appraisal that are less contractual and more social (Ramamoorthy and Carroll 1998), indicating a higher propensity toward engaging in citizenship (described in many instances as a discretionary behavior) as a means of creating a positive impression in response to support received from the organization. Likewise, employees in collectivist cultures have been found to prefer loyalty-based (rather than equity-based) forms of reward from their organizations (Ramamoorthy and Carroll 1998). This can also amplify the positive relationship between support and citizenship, due to the role of loyalty in connecting support with citizenship. Overall, we suggest that in response to support from the organization, employees in collectivistic cultures will engage in citizenship to a greater extent than employees situated in individualistic cultures.

H1: The POS–OCB relationship is stronger in cultures higher on collectivism.

2.2 Power Distance

Central to our argument, researchers in the perceived organizational support domain note the following: “employees think of their relationship with their organization in terms similar to a relationship between themselves and a more powerful individual” (Eisenberger et al. 2004, p. 207, italics added). Power distance is thus relevant in an organizational support theory context. Power distance is defined as “the extent to which a society accepts the fact that power in institutions and organizations is distributed unequally” (Hofstede 1980, p. 45). In cultures high in power distance, employees are deferential to those in higher positions, show respect for their position, and comply with their orders and directives. On the contrary, cultures lower in power distance see lesser of a hierarchical difference on power, status, and authority. The literature offers mixed results on the role of power-distance in exchanges between organizations and employees, which highlights the need for a meta-analysis. On one hand, there is a view that power distance may weaken the influence of POS on OCB (e.g., Farh et al. 1997; Lam et al. 2002). Employees who accept the existence of power distance differentials may have a strong deference to authority and tend to focus on their formal assignments as delegated by the organizational hierarchy. They might focus on task performance, rather than engage in citizenship (Farh et al. 2007; Hui et al. 2004). That is, in a high power distance setting, employees may feel obligated to conscientiously focus on the formally allocated job assignments, rather than divert their attention toward what can be considered discretionary, such as citizenship (Farh et al. 2007).

On the other hand, there is a view that power distance strengthens the influence of POS on OCB. There are several important reasons for supported employees to increase their citizenship behaviors when they are situated in high power distance cultures: (a) genuine gratitude and (b) showing gratitude for impression management. First, genuine feelings of gratitude to benevolent authorities tend to be higher in cultures with high power distance. For employees to reciprocate with behaviors that are discretionary (not formally required by the organization), employees should regard the supportive actions originating from the organization as also being discretionary—“rather than the result of such external constraints as government regulations, union contracts, or competitive wages paid by alternative employers” (Eisenberger et al. 2004, p. 207).

In high power distance cultures, the status quo is manifested in widely prevalent social inequalities, which are the norm (Hofstede 2001). Employees perform their jobs as per the stringent hierarchical structures, and get accustomed to a life where their needs are not necessarily a concern to the organizational authorities (Schwartz 1992). Nevertheless, organizational authorities do retain the discretion to extend support to employees. When this happens, employees in high power distance cultures are expecting it less than their counterparts in low power distance cultures. This will increase employees’ citizenship to return the favor to the more powerful party, the organization. Hence, when employees receive support beyond their expectations in high power distance cultures, they express their gratitude by positively reciprocating through citizenship behaviors that go beyond their specified job roles.

Second, pressures to impress powerful authorities tend to be higher in cultures with high power distance (e.g., the concept of ‘face’ in Eastern cultures). High power distance cultures tend to be less democratic when it comes to distribution of resources to the less powerful (Chakrabarty 2009, p. 36). The powerful can deprive the less powerful with hardly anyone to monitor, question, or challenge them (La Porta et al. 1999). Often, a class system exists, which prevents the upward mobility of the powerless—unless they have approval, endorsement, or support from the powerful (Markus and Kitayama 1991). Hence, the powerless may go overboard to impress the powerful when the ones in power provide support, and as a result, engage in citizenship behavior. Discretionary support by the organization—construed by employees in unequal societies as exceeding the formal provisions from their employer—will therefore be reciprocated with overwhelming gratitude or excessive behavioral dedication.

H2: The POS–OCB relationship is stronger in cultures higher on power distance.

2.3 Uncertainty Avoidance

Organizational support can be interpreted as a form of organizational action directed toward decreasing employees’ uncertainty. In high uncertainty avoidance cultures, defined as “the extent to which a society feels threatened by uncertain and ambiguous situations and tries to avoid these situations” (Hofstede 1980, p. 45), individuals guard against future uncertainty (Hofstede 2001). Individuals feel threatened (by uncertain situations) and attempt to devise strategies to avoid such perceived threats. In these cultures, members are susceptible to stressful events and to the resulting anxiety (Hofstede 2001). Signals of support—as well as concrete forms of support originating from the organization—will thus be interpreted by organizational members more positively, due to their high priority given to actions that diminish uncertainty. While all employees may look up to organizations for support (Eisenberger et al. 2004), those in high uncertainty avoidance cultures may do so to a greater extent, given their feeling of powerlessness toward external forces (Hofstede 2001). When receiving support, employees in high uncertainly cultures will be more likely to engage in behaviors aimed at reciprocating for what they got from the organization. Thus, in addition to possible increases in their task performance (Taras et al. 2010), employees will also engage in citizenship. An additional reason to increase citizenship by supported employees situated in high uncertainty avoidance cultures is that positive moral acts such as ‘giving’ or ‘doing good’ (in the form of altruism, courtesy, caring for and helping others) tend to further reduce individuals’ fear of uncertainty and create an uplifting feeling that all is well.

Members of cultures high in uncertainty avoidance have increased susceptibility to stressful events and anxiety (Hofstede 1980, 2001). Hofstede (2006) explains uncertainty avoidance culture in a country as a “collective anxiety level in view of the unknown and the unfamiliar” and is “correlated across countries with measures of anxiety symptoms, neuroticism, and lower subjective well-being” (p. 888). Thus, support from the organization, such as taking care of the employees’ well-being, not ignoring their complaints, and caring about employees’ satisfaction, will be perceived by employees situated in cultures high in uncertainty avoidance as reducing anxiety and providing requisite security. These supportive actions by the organization in cultures high in uncertainty avoidance will be appreciated to a much greater extent than in cultures where uncertainty avoidance is less of a problem (Hofstede 1980, 2001). This is precisely because such provisions of support provide a necessary counterpart to these employees’ desire for security and their need to avoid uncertainty. When the organization provides support, employee worries are diminished, and the possibility to reciprocate with citizenship is increased. By signaling that it will absorb some of the risks inherent in any employment relationship, which are seen as more prominent by members of high uncertainty avoidance cultures, the organization will create a safe haven for reciprocation and citizenship.

H3: The POS–OCB relationship is stronger in cultures higher on uncertainty avoidance.

2.4 Masculinity–Femininity

The masculinity versus femininity distinction of culture refers to “dominant gender role patterns in the vast majority of both traditional and modern societies” in terms of the extent to which “patterns of male assertiveness and female nurturance” are prevalent in society (Hofstede 2001, p. 284). Masculinity stands for a “society in which social gender roles are clearly distinct: men are supposed to be assertive, tough, and focused on material success,” whereas “women are supposed to be more modest, tender, and concerned with the quality of life” in tune with the expectations of men (Hofstede 2001, p. 297). Accordingly, masculine societies have features such as: “challenge and recognition in jobs important, advancement of earnings important, values of women and men very different, higher job stress, belief in individual decisions” (Hofstede 2001, p. 298). Femininity, stands for a “society in which gender roles overlap: both men and women are supposed to be modest, tender, and concerned with the quality of life” (Hofstede 2001, p. 297). Accordingly, feminine societies have features such as: “cooperation at work and relationship with boss important, living area and employment security important, values of women and men hardly different, lower job stress, belief in group decisions” (p. 298).

Hofstede (2006) explains that masculinity-femininity and individualism-collectivism represent “orthogonal factors” because “Mas-Fem separates countries in an entirely different way from Ind-Col” (p. 894). For example, “in Europe it separates Austria (masculine) from Sweden (feminine); in Asia, Japan (masculine) from Thailand (feminine); and in Latin America Venezuela (masculine) from Costa Rica (feminine)” (Hofstede 2006, p. 894). Men and women in countries with higher levels of cultural masculinity can be expected to fight aggressively, in their own distinct ways, for obtaining greater ego-boosting achievements and recognitions (Hofstede 2001; Kim et al. 1990). In contrast, men and women in countries with cultural femininity can be expected to emphasize compassion, empathy, and caring for the needs of others, rather than focus narrowly on personal satisfaction (French and Weis 2000; Hofstede 2001). Individuals in countries with cultural femininity expect a cooperative style in solving conflict, whereas those in a culture of masculinity are more inclined toward confrontational styles to reach a resolution (Leung et al. 1990).

As a result, the extent of masculinity versus femininity in a society can influence perceptions and exchanges taking place between the employee and the organization, including support-related evaluations originating from one’s organization. Specifically, support from one’s organization is less important or expected to a less extent in countries with cultural masculinity. In such cultural settings, employees will be more self-reliant and expect less from their employer. In contrast, in countries with higher levels of cultural femininity, individuals are more appreciative of the support received from their organization.

Just as important as how support is perceived, are the beliefs regarding the need for maintaining warm personal relationships and solidarity in countries with cultural femininity (Hofstede 2001), which encourages reciprocity in social exchanges to a greater extent than in countries with cultural masculinity. Employees in countries with higher levels of cultural femininity will reciprocate the care, compassion, and cooperation that they receive from their organization. They can thus use citizenship behaviors to reciprocate, given their focus on aspects beyond material success and a caring attitude toward a supportive organization. Therefore, to give back to a supportive organization, employees in feminine cultures will reciprocate to a greater extent with OCBs.

H4: The POS–OCB relationship is stronger in cultures with higher femininity.

3 Methods

3.1 Selection and Coding of Primary Studies

3.1.1 Literature Search

We used several methods to identity studies examining the POS–OCB relationship. Our target was to identify published articles, chapters, unpublished doctoral dissertations, conference papers, and unpublished studies. First, we conducted a computerized search of four databases, including Business Source Complete (EBSCO), Management and Organization Studies, PsychINFO, and ProQuest Dissertations and Theses using the keywords perceived organizational support, POS, organizational support combined with OCB, organizational citizenship behavior, citizenship behavior, contextual performance, extra-role behavior, and helping behavior. To check if the combination of search terms was too stringent and could lead to missing studies that should be in fact included, we conducted a supplemental search using only the first three terms (describing POS). An inspection of twenty additional articles found by this method revealed that none of these studies included our criterion of interest, citizenship behavior. Second, we examined the reference sections of prior meta-analyses on the topic (e.g., Rhoades and Eisenberger 2002; Riggle et al. 2009). Finally, we manually researched recent (last three years) conference programs of the Academy of International Business, Academy of Management, and Society for Industrial and Organizational Psychology annual meetings to identity potential unpublished studies. The searches resulted in the identification of 272 papers.

3.1.2 Inclusion and Exclusion Criteria

We used three criteria to determine whether to include studies in this meta-analysis. For a study to be included, it had to (a) report a Pearson product–moment correlation coefficient between organizational support and citizenship or other statistics that could lead to a computation of a correlation coefficient; (b) use actual employees as participants (i.e., studies using undergraduate students or being conducted in laboratory settings were excluded); and (c) have at least one sample in which participants were from the same country. A total of 70 studies, which provide 78 independent samples, met the inclusion criteria. From the 70 studies included in the final analysis, 40 were journal articles, 29 were dissertations, and one was a conference proceeding. These studies provided a total of 165 unique correlations. Two authors coded the articles for relevant information.

3.1.3 Coding Scheme and Study Characteristics

Consistent with existing theory outlining the form of citizenship (Organ et al. 2006) and with prior meta-analyses examining citizenship (Podsakoff et al. 2009), we focused on affiliative (i.e., prosocial) forms of citizenship behavior such as altruism, conscientiousness, compliance courtesy, helping, interpersonal facilitation, personal support, loyalty, obedience, and sportsmanship. These dimensions were classified as affiliative in prior meta-analytic work (e.g., Chiaburu et al. 2011; Ilies et al. 2007; Podsakoff et al. 2009).

Consistent with previous studies, we grouped citizenship by target (Coleman and Borman 2000): directed toward the organization (OCBO; e.g., compliance, conscientiousness, loyalty, obedience, sportsmanship) and toward other individuals (OCBI; e.g., altruism, helping, courtesy, helping, personal support, interpersonal facilitation). For primary studies reporting OCB using a global measure and without identifying the target, we coded as overall OCB. We captured whether citizenship ratings were provided by the same (self-report) or others (supervisors, coworkers; non-self-report), respectively. Perceived organizational support did not require specific coding since the construct is isomorphic with its operationalization.

3.1.4 Hofstede’s Dimensions of National Culture

To originally measure these national culture dimensions, Hofstede had calculated numerical scores “based on the re-analysis of an existing database of employee attitude survey scores assembled by one single multinational: the IBM Corporation, from its subsidiaries in 72 countries, between 1967 and 1973, and later expanded through replications to 75 countries and/or regions” (Hofstede 2006, p. 883). Hofstede (2006, p. 884) explains that the “the IBM attitude survey questionnaires had been designed as a management tool and developed through open-ended pilot interviews with personnel in nine countries” and that the “surveys were action-driven and dealt with issues that employees from different categories and/or their management considered relevant in their work situation”. Hofstede (2006, p. 884) believes that a key feature of this data was its “decentering” (i.e., “conscious attempts at avoiding ethnocentric bias”), whereby “in the IBM project, locally recruited company researchers with local degrees conducted the pilot interviews and contributed substantially to the questionnaires and the interpretation of the results.” Hofstede’s own “cross-national analysis came years later and developed its concepts from the database on file” and was “focused solely on societal cultures (differences between IBM respondents from different countries)” (Hofstede 2006, p. 884).

We obtained country-level data for the dimensions of interest on individualism–collectivism, power distance, uncertainty avoidance, and masculinity–femininity from Hofstede (1980, 2001, 2010). The countries represented are those for which we could find both Hofstede country-level cultural data and POS–OCB correlation data (sample-weighted correlations sourced from primary studies conducted in these countries): Australia, Belgium, Canada, China, India, Korea, Malaysia, Nigeria, Thailand, UK, and United States. Country-level cultural data were added to the POS to OCB correlation data sourced from primary studies.

3.2 Meta-Analytic Procedures

We used the Hunter and Schmidt (1990, 2004) method of meta-analysis. To calculate true population correlations between POS and OCB, we sample-weighted correlations from primary studies and accounted for sampling and measurement error in the predictor and criterion. When estimated reliabilities were not reported in the focal study, we used an imputation method based on the mean reliability estimated from the rest of the studies examining this relationship. Further, we ensured that included effect sizes were independent. When a primary study provided multiple estimates of the (within sample) correlation between a predictor and outcomes, we combined them into one correlation using Hunter and Schmidt (2004) composites formula. In doing so, we avoided double-counting and kept samples from contributing more than one effect size to the overall meta-analytic estimates.

Since our core research question is to examine to what extent cultural dimension moderate the POS–OCB relationship, we first explored if moderators are expected. Hunter and Schmidt (1990) suggested that if less than 75 % of the variability in correlations across studies is accounted for by statistical artifacts, moderators are likely. The absolute magnitude of corrected variability (SD ρ ) is another indication of moderated relationships. In addition to reporting true-score correlations, we examined the variability of the corrected correlations across studies by calculating 80 % credibility intervals around the true-score correlations and variability (SD ρ ) (Whitener 1990). Credibility intervals provide information regarding the generalizability of the focal relationships across situations. 95 % confidence intervals were generated through computing the standard error of the uncorrected mean effect size. We further calculated the percentage of the variability in correlations across studies accounted for by statistical artifacts (% s.e) to estimate whether potential moderators exist.

3.2.1 Testing Moderating Effects

Given the nature of our sample, some studies are nested within the same countries. However, some countries (e.g., United Kingdom, Thailand) are based on only one observation per country. Multilevel techniques such as hierarchical linear modeling (HLM) were inappropriate because of small sample sizes (N < 5) at the second level of analysis (see Raudenbush and Bryk 2002). Due to these constraints and consistent with prior literature (e.g., Atwater et al. 2009; Steel and Kammeyer-Mueller 2002), we included each of the four culture dimensions as moderators of the POS–OCB relationship to test our interaction hypotheses. We estimated the correlation coefficients between the culture scores and the effect size of organizational support on citizenship in modified weighted-least-squares (WLS) regression equations (Hedges and Olkin 1985). WLS methods are preferable to other available moderator estimation techniques (e.g., ordinary least squares regression, sub-group analyses; Lipsey and Wilson 2001; Steel and Kammeyer-Mueller 2002). In WLS each study is weighted based on sample size and studies with more participants have greater impact. A positive regression coefficient between a culture score and the outcome (β cultural dimension) indicates that the cultural dimension strengthens the POS–OCB relationship; a negative coefficient weakens it. In addition to WLS, we provide an alternative form of analysis—subgroup analysis—to assess the robustness of our results.

4 Results

4.1 Main Effect Size Estimates

We describe our results for the direct relationships in Table 1, which summarizes our findings across all studies, organized by the organizational citizenship behavior criteria. Specifically, we list effect sizes for the POS–OCB relationship, followed by OCB organized by target (individual-directed: OCBI and organization-directed: OCBO), and as a function of the rating source (same source vs. different source). Because prior meta-analyses exist for the relationship between organizational support and citizenship, we compared our results with prior meta-analytic studies reporting POS–OCB correlations (Rhoades and Eisenberger 2002; Riggle et al. 2009).

Table 1 Meta-analytic results for the relationship between perceived organizational support and organizational citizenship behavior
Full size table

Cumulative evidence is consistent with the positive correlation between organizational support and citizenship; the mean corrected correlation is r c  = 0.275 (N = 21,838 and k = 79), and both 95 % confidence and 80 % credibility intervals exclude zero. This means that the mean true-score correlation is statistically meaningful and distinguishable from zero and the individual effect size is generalizable. Compared to previous meta-analytic correlations (r c  = 0.22, Rhoades and Eisenberger 2002, and r c  = 0.26, Riggle et al. 2009), our effect size is slightly higher. Regarding the two sub-dimensions of overall citizenship, the estimated correlations are r c  = 0.264 for individual-directed (OCBI) and r c  = 0.288 for organization-directed (OCBO) citizenship, with 95 % confidence intervals and 80 % credibility intervals excluded zero. The correlation between support and organization-directed citizenship is slightly higher than the one with individual-directed citizenship. However, the confidence intervals overlapped to a great extent, which suggests that organizational support influences both individual- and organization-directed citizenship to a comparable extent. We also compared the results based on the source of rating (same vs. different source of rating). As expected, the correlation between organizational support and citizenship is larger when they are computed using the same rating source (self-reported data) than when using different sources. Perceived organizational support is more strongly correlated with self-reported citizenship (r c  = 0.335) than citizenship rated by others (r c  = 0.221). 95 % confidence intervals and 80 % credibility intervals excluded zero. Overall, the relationships are as expected.

4.2 Tests of Moderating Hypotheses Using Hofstede’s Cultural Dimensions

Table 2 presents descriptive results for the cultural dimension scores included in our meta-analysis (mean, standard deviation, range). Definitions for each dimension are also provided in this table. We conducted tests of moderation using the overall organizational citizenship behavior dimension because the number of studies for the sub-dimensions was smaller and this reduction in number of studies can lead to instability in the results. As noted previously, the correlation of POS with OCB was greater when data were reported by from the same source. To control for rating source, we coded the source reporting citizenship (0 = self report; 1 = non-self report) as a dummy variable and included it as a control variable when testing our hypotheses.

Table 2 Descriptive data and definitions for cultural dimensions
Full size table

4.2.1 Weighted Least Squares Analysis: Hofstede Cultural Dimensions as Moderators

Results are presented in Table 3. For each hypothesis, we entered the dummy variable into the WLS equation in the first step, with the POS–OCB relationship serving as the dependent variable. Then we entered each of the culture dimensions into the equation in the next step. We also estimated the amount of additional variance explained by adding the cultural dimensions.

Table 3 Hofstede cultural dimensions as moderators: weighted least squares (WLS) tests
Full size table

Concerning Hypothesis 1, collectivism positively and significantly moderated the relationship between organizational support and citizenship (β = 0.396, t (75) = 3.964, p < 0.001), indicating that the POS–OCB relationship is stronger in cultures high in collectivism. Collectivism explained 15.3 % additional variance in the POS–OCB correlation. And the overall model was significant: F (2,75) = 13.684, p < 0.001, R 2 = 0.267. The semi-partial correlation indicating the unique contribution of collectivism had a value of 0.392. Hypothesis 1 is therefore supported. In support of Hypothesis 2, power distance positively moderated the POS–OCB relationship (β = 0.317, t (75) = 3.070, p < 0.01), indicating that the relationship between support and citizenship is stronger in higher power distance cultures. The semi-partial correlation of power distance was 0.315. Additional variance (10.0 %) is significantly explained by power distance, with ΔF (1,75) = 9.428, p < 0.01. The model was significant, with F (2,75) = 10.137, p < 0.001, R 2 = 0.213.

In line with Hypothesis 3, uncertainty avoidance positively moderated the POS–OCB relationship (β = 0.353, t (75) = 3.397, p < 0.01; 11 % additional variance explained): support was more influential for citizenship in cultures with high levels of uncertainty avoidance. The semi-partial correlation of uncertainty avoidance was 0.344. We also found a significant overall model, with F (2,75) = 11.328, p < 0.001, R 2 = 0.232. Finally, supporting Hypothesis 4, femininity positively moderated the relationship between support and citizenship (β = 0.519, t (75) = 5.292, p < 0.001). Thus, the POS–OCB relationship is stronger in cultures with high levels of femininity. The masculinity–femininity dimension explained a significant amount of additional variance (24 %) and the semi-partial correlation was 0.491.

4.2.2 Subgroup Analysis

In addition to WLS, we provide an alternative form of analysis, through subgroup analysis. Subgroup analysis provides the advantage of presenting effect sizes for high vs. low levels of the cultural dimensions. First, we did a mean split (based on low and high cultural dimension scores) to create low versus high cultural score groups. Then, we meta-analyzed each subgroup and obtained the corrected correlations (effect sizes) and the respective 95 % confidence intervals (95 % CIs). As shown in Fig. 1, consistent with our prediction, the POS–OCB effect size is greater for the group with high collectivism than the group with low collectivism (high individualism) (r c  = 0.39 vs. r c  = 0.23). The 95 % confidence intervals from the two subgroups are not overlapping (0.36 to 0.42 vs. 0.22 to 0.25). Similarly, the effect size is greater for the group with high rather than low femininity (r c  = 0.41 vs. r c  = 0.23; confidence intervals from 0.38 to 0.44 vs. 0.21 to 0.24), with high rather than low power distance (r c  = 0.34 vs. r c  = 0.24; confidence intervals from 0.32 to 0.37 vs. 0.22 to 0.26), and for high rather than low uncertainty avoidance (r c  = 0.38 vs. r c  = 0.23; confidence intervals from 0.36 to 0.41 vs. 0.22 to 0.25). These results indicate that our findings are robust across methods.

Fig. 1
figure 1

Subgroup analysis: cultural dimensions as moderators (bars indicate 95 % confidence intervals)

Full size image

4.2.3 Post-Hoc Analysis for Long-Term Orientation

Apart from the four dimensions of national culture measured in the paper, Hofstede had also put forth a fifth dimension—long term orientation. Long-term orientation is the extent to which a nation culturally emphasizes a focus on the future (long term) vs. the present or past (short term). This dimension of national culture is not formally hypothesized in our study because of (1) the lack of fit into our theoretical framework, (2) missing data for three countries (Malaysia, Belgium, and Canada), and (3) the neglect of this construct in empirical research resulting in a low availability of studies for inclusion in meta-analysis (Kirkman et al. 2006). As suggested by an anonymous reviewer, we nevertheless report the moderating effect of this cultural dimension as following for comprehensiveness. After controlling for rating sources, there was a non-significant moderating effect detected and the overall model was non-significant as well. Sub-group analysis further confirmed the non-significant moderating effect: the POS–OCB effect size was greater for the group with long term orientation than the group with short term orientation (r c  = 0.38 vs. r c  = 0.25) and the 95 % confidence intervals from the two subgroups overlapped (0.26 to 0.51 vs. 0.21 to 0.28). The result should be interpreted with caution, given the gap between the number of studies under long term vs. short term categories (k = 7 vs. k = 71).

4.3 Post-Hoc Analyses for GLOBE Measure

We also carried out moderator analyses based on recently developed cultural dimension measures—GLOBE (House et al. 2004). These supplementary meta-analysis were based on only two of GLOBE’s dimensions—in-group collectivism and power distance—because they were similar to Hofstede’s culture dimensions of collectivism and power distance. The results for these two GLOBE measures were consistent with the hypotheses and consistent with the corresponding Hofstede measures. Consistent with Hofstede’s measure of collectivism, in-group collectivism positively moderated POS–OCB relationship (β = 0.358, p < 0.001). GLOBE’s measure of collectivism explained 13 % additional variance in the POS–OCB correlation. Further, consistent with Hofstede’s measure of power distance, GLOBE’s measure of power distance positively moderated the POS–OCB relationship (β = 0.443, p < 0.001). GLOBE’s measure of power distance explained 19 % additional variance in the POS–OCB correlation. The detailed results are available from the authors upon request.

We, however, did not consider GLOBE’s cultural practice measures of uncertainty avoidance and gender egalitarianism (or of any other GLOBE measures) as potential replacements for Hofstede’s uncertainty avoidance and masculinity–femininity, respectively. This is because Hofstede (2006, p. 886) has indicated strong concerns in that the two measures of uncertainty avoidance (UA)—his measure versus the GLOBE measure of UA—actually measure very different aspects of UA. Hofstede (2006, p. 894) also rejected any notion of equivalence between his masculinity–femininity and GLOBE’s gender egalitarianism measures.

5 Discussion

Researchers have argued that employee-organization relationships are based on “relationship, reciprocity, and exchange” that satisfies the self-interests of the employee and the interests of the organization (Coyle-Shapiro and Shore 2007, p. 166). Such employee-organization social exchanges are nevertheless embedded in a cultural environment (e.g., Ekeh 1974; Hofstede 1986; Hofstede and McCrae 2004; Lévi-Strauss 1949, 1969), and cultural aspects may shape the nature of the relationship between what one party (the organization, in the form of organizational support) provides and what the other party responds with (employees, in the form of organizational citizenship behaviors). Our research presented an integrated argument: that the relationship between POS–OCB is embedded within—and hence moderated by—the cultural aspects of the larger society. Results are compelling and consistent with our integrated perspective: national culture matters and modifies the strength of social exchanges.

5.1 Theoretical Implications

An important finding of this study is that the strength of POS–OCB relationship varies across cultures. The mere existence of a positive POS–OCB relationship (as demonstrated in prior meta-analyses; Rhoades and Eisenberger 2002) does little to explain variations in magnitude across cultural contexts. Even though the boundary conditions represented by our hypothesized cultural dimensions do not reverse the positive relationship between support and citizenship, they significantly weaken (or strengthen) it. Researchers have cautioned against the possibility of the “universalistic attribution bias,” in the form of downplaying or not accounting for cultural differences, despite their importance (Leung et al. 2001, p. 165). Referring to social exchange research, de Lauwe (1966) had also warned about lack of integration noting that, “In the United States, the emphasis given from the start to the idea that society is a body of individuals and that it has no existence of its own contrasts to a certain extent with the French conception that first considers society taken as a whole” (p. 248, italics added). Our integrated view captures both individual and societal aspects and has to potential to draw researchers’ attention toward cultural factors.

In addition to showing that there is systematic variation in the strength of the POS to OCB relationship, we develop theoretical arguments—supported by our meta-analytic results—and outline specific patterns: the relationship is accentuated by higher collectivism, power distance, uncertainty avoidance, and femininity. First, these findings qualify the general recommendation that to elicit higher levels of citizenship from employees, the organization has to correspondingly increase its levels of support. Second, the specific pattern we uncovered is thought-provoking and informative for future research. Consider for example Hofstede’s cultural dimensions for the United States: low collectivism (individualism: 91), low power distance (40), low uncertainty avoidance (46), and low femininity (masculinity: 62). In the light of our results, US emerges as a prototypical illustration of a cultural setting displaying an attenuated POS–OCB relationship. This finding suggests several possibilities. One is that social exchanges and the norm of reciprocity are more important in cultures where the pattern of cultural dimensions is opposite to the one found in the US. Consistent with this explanation, Taras et al. (2010) found that exchange ideology is weaker (inclining toward social exchanges) in collectivistic cultures. Due to lack of primary studies, Taras et al. did not report how exchange ideology co-varies with the other Hofstede dimensions. Additional research is therefore needed, and our meta-analysis can inform its direction: if social exchanges and reciprocity underlie the POS–OCB relationship, reciprocation should be found to be more influential in connecting the two in the cultures we highlighted as accentuating this relationship.

Another possibility is for explanations other than social exchange to be operating in conjunction with, or even overshadowing, social exchange. To tease apart specific mechanisms, we join researchers who argued for the need to design studies where the reciprocation is explicitly measured (Coyle-Shapiro and Shore 2007; Lambert 2000). In fact, Coyle-Shapiro and Shore (2007) provide a number of alternatives to social exchange explanations, including the presence or perceptions of communal relationships, economic (rather than social) exchanges (Shore et al. 2006) and more general non-reciprocation mechanisms, including obedience to a more powerful authority (the organization). Both our findings and theoretical explanations can stimulate future effort directed toward clarifying what accounts for variations in the POS- (and other social exchange constructs; Tekleab and Chiaburu 2011) and OCB.

Further, our meta-analysis can shed new light on results and recommendations based exclusively on primary studies. An over-reliance on primary research may lead to the possibly premature conclusion that high power distance inhibits the influence of support on citizenship behaviors (Farh et al. 1997). According to such arguments, employees who accept power distance defer to authority (rather than engaging in reciprocation) and focus on formal assignments (rather than on discretionary citizenship). In a primary study, Farh et al. (2007) found that power distance weakens the influence of POS on OCB altruism (but not OCB conscientiousness and voice, for which there were no significant differences in simple slopes).

On one hand, the contrasting results between what we found meta-analytically and the primary study of Farh et al. (2007) can be attributed to methodological differences. While we assessed power distance as a cultural phenomenon, Farh et al. conceptualized it at the level of dyadic individual self-interests (e.g., “managers should seldom ask for the opinions of employees”). On the other hand, based on our results obtained across several studies, we propose that when the organization provides support, employees in high power distance cultures will focus on the fact that the organization is not mandated to provide such high levels of support, be more appreciative, and increase citizenship. Theoretically, this explanation is consistent with findings on how high power distance employees interpret fairness. Because such employees are less likely to expect favorable rewards, perhaps also due to a lack of social exchange with their supervisors, they focus on procedural aspects. As Begley et al. (2002) have shown, citizenship is boosted by procedural fairness to a greater extent for employees high in power distance. Results consistent with this explanation have also been obtained meta-analytically by Shao et al. (2013) and are discussed below. Overall, because differences in findings might be attributed to differences in measurement (power distance assessed at a dyadic individual-level vs. generic country-level), one logical extension is to determine the extent to which power distance situated at different levels of analysis (e.g., individual, team, organization, country) has differing effects on the POS–OCB relationship.

From a different direction, researchers and practitioners recognize that globalization requires some shift of organization’s HRM practices across settings (e.g., home country vs. overseas). As research has demonstrated, perceived support and HRM practices are very much related (Eisenberger and Stinglhamber 2011; Tremblay et al. 2010; Whitener 2006). The issue is relevant especially in a context of standardization–localization of HRM practices, where companies deal with two tendencies: ‘push’—parent companies pushing policies toward the subsidiaries vs. ‘pull’—local influences avoiding standardization (Farley et al. 2004). Beyond the current debate contrasting standardization versus localization in HR practices (e.g., Pudelko and Harzing 2007; Khilji 2003), our results inform the extent to which an intensified focus on HR practices—or on other aspects signaling organizational support—is necessary. Based on our findings, such intensification of efforts (or of signals) is warranted in some settings (e.g., high power distance) more than in others (low power distance contexts). In the absence of such information, decision-makers are prone to suboptimal decisions in resource allocation.

Our findings complement and—at the same time—are consistent with results reported in other meta-analyses (e.g., Shao et al. 2013). Shao et al. examined the relationship between fairness dimensions and outcomes, including citizenship behaviors, moderated Hofstede’s cultural dimensions (individualism, power distance, uncertainty avoidance, and masculinity). In a meta-analysis, their reported weights for the relationship between supervisor justice and citizenship behaviors directed at the supervisor, uncovering all four Hofstede’s dimensions as moderators: collectivism (β = −0.41 for individualism), power distance (β = 0.41), uncertainty avoidance (β = 0.41), and femininity (β = −0.41 for masculinity) (see the Appendix of Shao et al. 2013, p. 33). Their findings are consistent with the results reported in our Table 3. Even though the theoretical underpinnings of the relationship between fairness (originating from the supervisor) and citizenship (directed back toward the same source) (Shao et al. 2013) are not exactly the same as the POS–OCB relationship, the commonalities at theoretical level (e.g., both relationships can be based on reciprocation) could account for the convergence of findings.

Besides using cultural dimensions at a national level as boundary conditions of the relationship between POS and OCB, it is also worth considering OCB measures that directly reflect the complexities of the social environment in which citizenship behaviors are embedded—a procedure recommended in cross-cultural psychology (Hui and Triandis 1985). In their study considering the social context of OCB and its correlates, Farh and colleagues (1997) found that OCB shared universal meanings of civic virtue, altruism, and conscientiousness across US and Chinese contexts. At the same time, two dimensions—“interpersonal harmony” and “protecting company resources”—were unique to the Chinese context. Further, Farh et al. (2004) found that not only were OCB formulated in China distinct from their form in the West, but also certain items under seemingly identical dimensions were different, due to the difference in how citizenship behavior can be interpreted. For example, they found that although altruism in China was found to be similar to Western altruism at a definition level, non-work-related helping captured under altruism in China was not typically considered in US contexts.

Following this logic, it is not hard to imagine that organizational support may elicit interpersonal harmony in the particular context of our example, because Chinese tend to engage in social exchange driven by “their self-derived obligations to their companies” and to coworkers in the workplace (Farh et al. 1997, p. 424). Stated differently, indigenous measures are most likely to reflect the influence of POS because a general OCB measure misses some important facets of citizenship valued in a different cultural context. Despite the arguments outlined above, we were not able to examine type of measure as a moderator due to the lack of primary studies using culturally-specific measures of citizenship. Indigenous measures of OCB and the connection with correlates merit further attention in future studies.

5.2 Practical Implications

Organizations doing business globally attempt to bridge the global and local discrepancies (Bass and Chakrabarty 2014; Chakrabarty and Wang 2012, 2013) and optimize their human resource practices (Allen et al. 2003; Eisenberger et al. 2004; Whitener 2006). Indeed, prior conceptual work emphasized that “researchers who broaden our awareness of the role of the social exchange ideology across cultures will facilitate our understanding of how to craft and modify human resource practices” (Whitener 2006, p. 498). Because the organizations’ human resource policies and employment practices represent anchoring points used by employees to form their impression of support, our study has implications for organizations conducting business in multiple national and cultural settings, and need to create or adapt human resource policies to reflect such cultural differences (Kraimer and Wayne 2004). As our results show, support leads to greater engagement in citizenship in countries with high levels of collectivism, power distance, uncertainty avoidance, and femininity. In such settings, when matched with employees’ perceptions of support, increasing support provided by the employer are more likely to accentuate employees’ prosocial citizenship.

Organizations need however to conduct business in cultural contexts where the connection between support and citizenship is not the strongest. As illustrated by our findings, in conditions characterized by low levels of collectivism, power distance, femininity, and uncertainty avoidance, enhancements in organizational support will still lead to some amount of citizenship. In such cases, the issue of what else contributes to employees’ citizenship (in addition to, and separately from, organizational support) becomes important. Organizations in such cultural contexts should therefore expand their search for what else could work to improve OCB. Specifically, organizations should determine if other factors (e.g., organizational climate, workgroup support) could be more effective as drivers of employees’ citizenship behaviors.

5.3 Limitations and Future Research

The meta-analysis is not without limitations. First, although we investigated the role of cultural dimensions, these may not be the only meaningful moderators. The strength of social exchanges may also vary, for example, as a function of the extent to which countries’ institutions have embraced modernity (e.g., Chakrabarty 2009; Chakrabarty and Bass 2013, 2014a, b, 2015; Inglehart and Baker 2000; Parboteeah et al. 2004). Furthermore, because cultural dimensions exert their influence in “bundles” rather than isolated from one another, their simultaneous influence needs to be assessed. While we were unable to do so due to difficulties in creating and testing higher-order interaction terms through meta-analysis, such tests may be possible in the future. Second, we recognize that Hofstede’s conceptualization and measurement of cultural dimensions is not without critics (e.g., McSweeney 2002). While we have attempted to mitigate such concerns by including post hoc analyses using GLOBE dimensions, our results should nevertheless be interpreted with caution. Third, similar to prior work (e.g., Atwater et al. 2009; Taras et al. 2010), cultural dimensions were measured using country scores. We therefore assumed—rather than measured—homogeneity in employees’ responses within a country. As more primary-studies accumulate in the future, other hypotheses can be tested, where country-level culture dimensions can be combined with data obtained on cultural dimensions situated at organizational, workgroup, and individual levels so that potential differences within a country can be captured. Fourth, in addition to organizational support, social exchanges can be conceptualized in other ways, including quality of the relationship with one’s leader and with coworkers (Chiaburu and Harrison 2008; Ilies et al. 2007) and these may also influence organizational citizenship behaviors. Examining more than one such relationship is however prohibitively large for the scope of one meta-analysis and we therefore opted to focus only on the POS–OCB relationship. Future studies may investigate other forms of social exchange. Finally, this study was based on primary studies investigating the POS–OCB relationship originating from countries where primary studies were conducted. Future research can investigate whether the findings replicate when data from more countries become available. Despite the limitations, the study has a number of strengths, including examining our hypotheses across a large number of respondents (more than 20,000), controlling for the influence of the source reporting citizenship (self- versus non-self-report), and positioning cultural dimensions as moderators and thus extending prior meta-analyses focused on direct relationships (Riggle et al. 2009; Rhoades and Eisenberger 2002).

6 Conclusion

More than a decade ago, highlighting the importance of culture, Hofstede (1993) asked an incisive question: is the management theorist or practitioner who ventures with a theoretical idea “outside his or her country into other parts of the world [like] Alice in Wonderland” (p. 93)? In a more limited (POS–OCB) context, we posed a related question: is the social exchange researcher to assume that a well-established relationship applies across cultures without major modifications and its corresponding effect size remains invariant? If social exchanges are as universal as often held to be, this (and other similar) relationship(s) should remain robust across cultural contexts. Responding to calls for cross-cultural research that investigate the foundations of social exchanges and citizenship (e.g., Baran et al. 2012; Farh et al. 2008; Kirkman et al. 2006; Shore et al. 2009), we demonstrated that the POS–OCB relationship is not impervious to cultural influences, which have to be accounted for. We hope that our investigation stimulates additional research in national settings not included in our sample, and encourages corresponding theory building.