Leadership styles and group organization

Introduction

There is growing recognition in the organizational behavior literature for the relations between societal culture, leadership, and cooperation in teams (Gelfand, Erez, & Aycan, 2007). The aim of this article is to contribute to our knowledge of these relations, by focusing on two questions: (a) what is the direct relation between societal culture and the level of organizational citizenship behavior (OCB) in teams, and (b) what is the moderating effect of societal culture on the relation between leadership and OCB in teams?

With the increased attention to groups and teamwork in organizations, the relationship between OCB and team performance has become increasingly important (Cummings & Worley, 2001; Van der Vegt, Van de Vliert, & Oosterhof, 2003). Focusing on teams, citizenship behaviors need to be defined and

* Correspondence to: Martin C. Euwema, Department of Psychology, Katholieke Universiteit Leuven, Tiensestraat 102, 3000 Leuven, Belgium. E-mail: Martin.Euwema@psy.kuleuven.be

M. C. EUWEMA ET AL.

investigated at the group level as well as the individual level (Ehrhart & Naumann, 2004; Karam & Kwantes, 2006; Naumann & Ehrhart, 2005; Schnake & Dumler, 2003).

Both individual and group organizational citizenship behaviors (GOCBs) are thought to be strongly related to leadership. Many studies have demonstrated for example the positive impact of supportive leadership on the deployment of (individual) OCB by subordinates (LePine, Erez, & Johnson, 2002; Podsakoff, MacKenzie, Paine, & Bacharach, 2000). Ehrhart and Naumann (2004) also emphasize the importance of leadership behaviors for the development of OCB norms and practices in groups. Few studies so far have empirically investigated this relation (Pearce & Herbik, 2004). Investigating the influence of leadership on OCB at the group level (referred to here as GOCB), therefore seems a natural step to take (Naumann & Ehrhart, 2005).

Although the cultural context of OCB and GOCB has increasingly been recognized as highly relevant, it has received little attention in empirical studies so far (Podsakoff et al., 2000). There are some indications that the meaning of OCB and the antecedents of OCB vary across cultures (Gelfand et al., 2007). The importance and functioning of groups and group processes on the other hand are at the core of societal differences. Particularly, the literature on IDV and collectivism (IC) gives a central place to the meaning of groups in everyday life, including the work place (Gelfand, Bhawuk, Nishii, & Bechtold, 2004; Hofstede, 2001). Therefore, exploring the impact of societal culture on GOCB is filling

a gap in our current knowledge of organizational group processes.

Societal culture is related to, and includes, norms about leadership behavior, and leadership effectiveness (House, Hanges, Javidan, Dorfman, & Gupta, 2004). The relation between leadership behaviors and group processes such as GOCB, might also be moderated by societal cultural differences. In this study we investigate the influence of societal culture on GOCB and its relationship with leadership. Three questions will be addressed: (a) what is the effect of leadership on GOCB; (b) does societal culture influence the level of GOCB, and (c) does societal culture moderate the relationship between leadership and GOCB? These questions are explored using data collected by a multinational consultancy firm in more than 60 countries around the world, with surveys among both managers and their direct reports. The measurements do have some limitations, as will be explained in the method section. The data however offer an excellent opportunity to explore and validate theoretical notions developed with other instruments in national samples. We will first elaborate the concept of GOCB, then discuss the relation between leadership and GOCB, followed by the impact of societal culture on GOCB and the relation between leadership and GOCB.

GOCB The classic definition of organizational citizenship refers to individual behavior, not directly or

explicitly recognized by the formal reward system, which promotes the effective functioning of the organization (Organ, 1988). OCB has been studied extensively over the last 20 years (Organ, Podsakoff, & MacKenzie, 2006; Podsakoff et al., 2000). The majority of OCB research has been conducted at the individual level of analysis. Recently, several authors have shifted the focus from individual OCB to OCB at the group level, using different terms, such as unit-level OCB, team citizenship behavior or team OCB, or collective citizenship behavior (Ehrhart & Naumann, 2004; Karam & Kwantes, 2006; Koys, 2001; Pearce & Herbik, 2004; Podsakoff, Ahearne, & MacKenzie, 1997; Schnake & Dumler, 2003). Chen, Lam, Schaubroeck, and Naumann (2002) introduced GOCB. This term will be used in our present study. According to Ehrhart and Naumann (2004), there is a lack in understanding the conditions under which high levels of GOCB are likely to form. The aim of this study is to explore two of those conditions: leadership style and societal culture.

LEADERSHIP AND GOCB ACROSS CULTURES

Chen et al. (2002) have elaborated the construct of Group OCB (GOCB). They conceptualize GOCB as a distinct group-level phenomenon concerning the extent to which the work group as a whole engages in OCB within the team. 1 Behaviors such as taking constructive initiatives and helping each other are voluntary, but at the same time important to the success of the group. Salam, Cox, and Sims (1996) argue that OCB is really interactive and ‘social’ in nature, and should therefore also be viewed as an element of team culture. Recognizing this value, OCB research at the group or unit level of analysis is beginning to accumulate (Ehrhart, 2004; Ehrhart & Naumann, 2004; Pearce & Herbik, 2004; Schnake & Dumler, 2003). Schnake and Dumler (2003) emphasize the need to study OCB at group levels. They argue that team and organizational effectiveness typically depends on the collective levels of OCB, or GOCB: ‘The OCB construct itself is a multi-level construct .... Theoretically, it is OCB in the aggregate, and not individual instances of OCB, which impact organizational effectiveness’ (p. 295), and they continue: ‘It is theorized, . . .that aggregate or group-level OCB contributes to organizational effectiveness . . ..Other group-level phenomena which may impact group-level OCB include group-cohesiveness, group norms, and leadership . . .’ (p. 296). GOCB is therefore defined here as a group-level variable, and should be considered separate from individual level models of OCB.

Antecedents of GOCB Defined at the group level, GOCB is facilitated primarily by characteristics of the group and factors that impinge on the group, for example, leadership. This is an important difference with models of individual OCB. For example, in a recent study Bachrach, Powell, Collins, and Richey (2006) demonstrated the effect of task structure (a group variable) on the level of OCB in working groups. At a more general level, Chen et al. (2002) and Ehrhart and Naumann (2004) argued that GOCB is likely to

be the result of inter-individual and intra-group level characteristics, such as interdependence of team members, perceived similarity with other group members, status of group members employing OCB, and identification and attraction to the group. All these factors contribute to the development of norms in relation to OCB in the group (Chen et al., 2002; Ehrhart & Naumann, 2004).

GOCB facilitates coordination among group members, and fosters group efficiency and efficacy (George & Bettenhausen, 1990; George & Jones, 1997). GOCB is associated with group-cohesiveness, and is related to pro-social and helping behaviors in groups (Naumann & Ehrhart, 2005; Salam et al., 1996). Studies by Koys (2001), Podsakoff, MacKenzie, Moorman, & Fetter (1990), and Podsakoff et al. (1997), all support the positive relation between GOCB and performance at the team level. The level of GOCB within a team, representing collective team behavior, is most likely influencing team performance (Ehrhart & Naumann, 2004; Schnake & Dumler, 2003).

Leadership and GOCB OCB is strongly related to both, informal and formal leadership. In this study we concentrate on the

effects of formal leadership behavior on the group. In their extensive review, Podsakoff et al. (2000) present results from meta-analytic studies of different leadership styles in relation with OCB. Podsakoff’s review (2000) indicates that directive leadership is negatively related, and supportive leadership is positively related to OCB. Organ et al. (2006) report also positive relations between

1 Chen, Lam, Naumann, and Schaubroeck (2005) later transferred this concept into group citizenship behavior, defined as the behavior of the group towards other groups in the organization.

M. C. EUWEMA ET AL.

supportive leadership and different forms of OCB. In addition, transformational leadership (Podsakoff et al., 2000), charismatic leadership (Deluga, 1995), and quality of leader–member exchange (Wayne & Green, 1993) all promote OCB. Particularly transformational leadership has received a lot of attention (Podsakoff et al., 2000). In this study we focus on directive and supportive leadership styles. These styles might be seen as one of the most classical and parsimonious leadership models, a model which has already been used in several cross-cultural studies.

Many leadership models differentiate two main types of leadership behaviors: task-oriented and relationship-oriented behavior, also referred to as initiating structure and consideration (Judge, Piccolo, & Ilies, 2004), or as directive and supportive leadership (Northouse, 2004). These two leadership dimensions have received considerable attention in cross-cultural studies (Peterson & Hunt, 1997). Dorfman (2004, p. 286) concludes: ‘findings from a number of cross-cultural studies have supported the importance of the task-oriented and relationship-oriented leadership dimensions that were originally conceptualized in the United States’. Dorfman (2004, p. 287) further concludes: ‘in general, cross-cultural studies support the importance of considerate leadership in increasing subordinates’ satisfaction with supervision’. Leadership theory has developed beyond the two dimensional model of task and relationship-oriented behavior. This does not mean that using this model is outdated. Judge et al. (2004) advocate to include initiating structure (directive) and consideration (supportive) as leadership styles in contemporary research. This view is shared by other authors (Northouse, 2004; Yukl, 2002). We therefore believe that the concepts of directive and supportive behavior are among the most robust leadership concepts. They have been used in research (including cross-cultural research) for over 30 years. These dimensions therefore offer a good starting point for exploring the effects of leadership on group processes, or more specifically on GOCB across cultures.

Few studies have addressed the relation between leadership and GOCB (Ehrhart, 2004; Pearce & Herbik, 2004). In one of these studies, Ehrhart (2004) reports strong relations between supportive (called servant) leadership and GOCB. More general, the relevance of leader behaviors for many team processes is evident (Yukl, 2002). West and Hirst (2005) for example, summarizing empirical support on team leader research, show that both facilitative leadership and task leadership are important for group effectiveness. Directive and supportive leadership are important correlates to team effectiveness, productivity, and team learning (Burke, Stagl, Klein, Goodwin, Salas, & Halpin, 2006).

Directive and supportive leadership Directive leadership is usually defined as task-oriented behavior, with a strong tendency to control discussions, dominate interactions, and personally direct task completion (Cruz, Henningson, & Smith, 1999). In addition, time management, pressure to realize targets, and close supervision on details are seen as characteristic of this style (Schmidt & Yeh, 1992). In this study, we define directive leadership as task-oriented behavior, with a strong focus on targets, close supervision, and control of subordinate actions. Directive leadership seems to be negatively related to GOCB. Podsakoff et al. (2000) find in their meta-analyses that leader specification of procedures, a form of directive leadership, is generally not promoting OCB and even inhibiting helping behaviors. When managers are controlling strongly, initiative by employees is easily seen as inappropriate, risky behavior, or even insubordination (Paine & Organ, 2000). In line with these studies on OCB, Salam et al. (1996) report a negative direct impact of directive (authoritarian) leadership on GOCB. Directive leader behavior puts the team members in a dependent role, facilitating them to wait for the manager before acting, showing less initiative and fewer extra activities. Studies on team communication also demonstrate the often negative effect of directive leadership (Cruz et al., 1999). The overall conclusion is clearly that directive leadership is negatively related to GOCB.

LEADERSHIP AND GOCB ACROSS CULTURES

Supportive leadership has been defined and used since the 50s of the 20th century. Definitions usually include sensitivity to team member needs (House, 1971). In this study supportive leadership includes sensitivity to individual and group needs, care for group tensions, and focus on harmonic working relations. In their meta-analysis, Podsakoff and colleagues (2000) report consistent positive relations between supportive leadership and different aspects of OCB. This is in line with findings from Organ et al. (2006). Looking more specifically at the team level, Chen et al. (2002) find a strong positive relation between supportive leadership and GOCB. Supportive leadership creates a team climate in which members feel empowered to act and collaborate with each other. Trust and support by the manager helps team members to take initiative, reciprocate the behavior by supporting team members, and overcome fears of criticism.

Hypothesis 1 . Directive leadership is negatively related to GOCB. Hypothesis 2 . Supportive leadership is positively related to GOCB.

Culture and GOCB Though recognized as highly relevant, surprisingly little attention has been given so far to the cultural

context of OCB (Organ et al., 2006; Podsakoff et al., 2000). There are some indications that culture influences the meaning of the construct of OCB. Farh, Earley, and Lin (1997) for example developed a specific Chinese OCB scale. Civic virtue, altruism, and conscientiousness are present in both western and Chinese OCB scales, whereas protecting company resources seems a typical Chinese form of OCB. Paine and Organ (2000) argue that both the perception of OCB and the likelihood of demonstrating OCB are influenced by cultural dimensions, such as IDV–collectivism and power distance (PD), and in recent years the academic interest in the influence of culture on OCB is growing (Gelfand et al., 2007; Karam & Kwantes, 2006).

Cross-cultural research has long been dominated by the constructs of IC. IC has been studied widely in organizational research (Dickson, Den Hartog, & Mitchel, 2003). Different definitions, operationalizations, and measurements of IC exist; underscoring the relevance and importance of this cultural dimension (Gelfand et al., 2004; Oyserman, Coon & Kemmermeier, 2002). Gelfand, Erez and Aycan (2007) conclude in their recent review that current cross-cultural research is focused too much on IC, and future research needs to explore other constructs and cultural dimensions as well.

In this study we use two models dominating the field of cross-cultural organizational studies; Hofstede’s (1980, 2001) work and the GLOBE project (House et al., 2004). Recognized as mile stone projects, the pros and cons of both approaches are still under debate (Earley, 2006; Hofstede, 2006; Javidan, House, Dorfman, Hanges, & Sully de Luque, 2006; Smith, 2006). We explore the effects of two cultural dimensions which are present in both theoretical models, that is IC and PD. These dimensions seem particularly relevant for the study of leadership in relation with GOCB. We decided not to take an explorative approach, including all dimensions of Hofstede and GLOBE. Though one might find arguments to relate each different dimension to group processes and leadership, we prefer a parsimonious approach, dealing with these two dimensions. The most obvious culture dimension for studying the importance of groups seems collectivism, and the most obvious culture variable for hierarchy-related issues is PD. These two dimensions are not independent. In Hofstede’s study as well as in the GLOBE project, IDVand PD are rather strongly related; more collectivistic societies also have more PD. Therefore, we thought it relevant to include both dimensions in this study, to prevent biased conclusions based on one cultural dimension (Smith, 2006). (An overview of the definitions of all

M. C. EUWEMA ET AL.

dimensions by Hofstede and GLOBE is available from the authors). Before exploring the effects of IC and PD more in depth, we briefly discuss the Hofstede and GLOBE models here, as we use the country level data collected by Hofstede (2001) and GLOBE (House et al., 2004).

Hofstede collected his data mostly in the late 1960s and early 1970s. Peterson (2003) concludes that these data will continue to be controversial, given the measures used, given that the data collection took place in one company only, and the time since data collection. Hofstede (2001) argues that national cultures are stable over time, though even Hofstede recognizes cultural changes. The 2001 scores of Hofstede were cross-validated with other world wide databases and theories such as Schwartz (1994), Trompenaars (1993), and Inglehart (1997). Though criticized, Hofstede’s model is still recognized as highly influential in the field of cross-cultural leadership (Peterson, 2003, 2004; Smith, 2006). IC is probably the most documented and recognized dimension of Hofstede (Smith et al., 2002; Triandis, 2001). In addition to IC, PD is widely recognized as a cultural dimension with important implications for leadership, and group processes (Hiller, Day, & Vance, 2006; Hofstede, 2001).

In the GLOBE study nine cultural dimensions are distinguished, based on an extensive data collection in 64 countries. As Peterson (2004, p. 641) states: ‘The project seeks to refine Hofstede’s societal culture dimensions and link the projects’ new measures to organizational culture and leadership’. The GLOBE project offers an important contribution to our understanding of cultural influences on leadership and organizational practices. A limitation of the GLOBE study however is that both leadership and organizational practices are primarily assessed with self reported surveys among managers, asking managers what is effective in their society (Earley, 2006; Peterson, 2004; Smith, 2006). A more reliable picture of actual leader behavior and descriptive group practices might be received when subordinates are asked to describe the actual behavior of their managers.

IC is generally defined as the degree to which people in a country prefer to act as individuals rather than as members of groups. As Oyserman et al. (2002, p. 5) summarizes: ‘These definitions all conceptualize IDV as a worldview that centralizes the personal—personal goals, personal uniqueness, and personal control, and peripheralizes the social’. In collectivist cultures individuals define their needs more in terms of relations within groups and sacrifice personal needs for the sake of the group. IC is in the GLOBE project divided into two (sub) dimensions, institutional and in-group collectivism (IGC). IGC shows most commonalities with Hofstede’s IC (Smith, 2006). The GLOBE project defines IGC as ‘the degree to which individuals express pride, loyalty, and cohesiveness in their organizations or families’ (House et al., 2004, p. 12).

Given the central place of group values, IC is relevant to relate to GOCB. A number of studies have shown that people in collectivistic cultures tend to have a stronger attachment to their organizations and tend to subordinate their individual goals to group goals, compared with employees in individualistic cultures (Jung & Avolio, 1999; Triandis, 1995). Paine and Organ (2000) therefore expect a positive relation between collectivism and OCB. This corresponds with findings by Moorman and Blakely (1995). Gelfand et al. (2004, p. 456) conclude, based on their literature review, that ‘all in all, these findings illustrate that there is an emphasis on cooperative team processes in collectivistic cultures’. Based on these theoretical and empirical arguments we expect a negative relation between IDV at societal level, and GOCB in organizations within these societies.

PD is defined as ‘the extent to which the less powerful members of institutions and organizations within a country expect and accept that power is distributed unequally’ (Hofstede, 2001, p. 98). The GLOBE definition is almost identical: ‘the degree to which members of an organization of society expect and agree that power should be shared unequally’ (Carl, Gupta, & Javidan, 2004; p. 517). Hofstede (2001) emphasizes PD as an important factor influencing team relationships. In societies with high PD, organizations are more centrally organized, team organization is less common, and employees are not expected to take initiative, or participate in decision making. Hiller et al. (2006) studied the effects of PD on team relations, and demonstrate that shared leadership occurs more under

LEADERSHIP AND GOCB ACROSS CULTURES

low PD conditions. Shared leadership in the team can be seen as an expression of GOCB, in which team members take responsibility for the group processes. As Carl et al. (2004, p. 560) conclude: ‘ . . .people in these societies do not expect the leaders to allow for participation or to be accountable for results’. This might influence GOCB, as team members in high PD cultures are less encouraged to take initiative, are more expected to be obedient and are less independent, therefore will be acting within their formal role. Carl et al. (2004, p. 599) see here a possible limitation to innovation: ‘One element of high PD is clearly dysfunctional as it preempts the society from questioning, learning, and adapting as there is little opportunity for debate and voicing of divergent views. Asking questions may be interpreted or regarded as criticizing and blaming, and therefore may be prohibited’. GOCB as typical extra-role behavior therefore also might be inhibited in cultures with high PD.

Hypothesis 3 . Societal culture is related to GOCB, more specifically GOCB will be higher in (a) more collectivistic societies, (b) societies with lower PD.

Culture, leadership, and GOCB Previously, we argued that directive leadership has a negative relation with GOCB (H1), whereas

supportive leadership has a positive relation with GOCB (H2). We now explore the possible moderating role of culture on this relationship. So far, we are not aware of empirical studies addressing this moderating role. There is ample evidence however for the impact of culture on leadership behaviors and leadership effectiveness (Dorfman, 2004; House et al., 2004). Also, culture moderates the relationship between leadership and different employees’ outcomes, such as satisfaction and turnover, as well as work unit productivity (Gelfand et al., 2007). Therefore, we presume culture also moderates the relation between leadership and GOCB.

The moderating role of culture on the relation between leadership and group behavior can work through different processes. First, some cultures might perceive and evaluate leadership and its effects as more important than other cultures. Dorfman (2004) argues that in the US the role of leadership is overvalued, and most likely is seen as more important than in other societies. This might imply that in the US, and possibly in other cultures where leadership is similarly highly valued, the impact of leader behavior on team processes is stronger, than in cultures that do focus less on the importance of leaders.

When employees are strongly oriented towards leader behavior, the evaluation of this behavior will influence employees’ attitudes and behaviors more than it will foremployees who are not so much focused on their leader. This holds also for the development of group norms, and group behaviors. In general one might expect that cultural norms emphasizing team efforts, might moderate the influence of leadership behaviors. Such norms are represented in IC. If the societal norm is more in favor of team work, one might expect that the specific leadership behaviors are relatively less important in facilitating GOCB. Employees act according to the norm, even when the managers’ behavior is not facilitating this behavior.

In individualistic cultures, employees will regulate their investments more in relation to their direct benefits. In these cultures, employees are supposed to act as ‘economic men’, whereas in collectivist cultures they are more likely to act in the interest of their in-group rather than their self-interest (Hofstede, 2001, p. 244). In individualistic cultures, employees focus probably more on the leader. If the leader invests in the individual employee, and in the team, paying attention to the team climate and individual needs, helping and supporting team members, employees will feel rewarded. In effect, they will reciprocate towards other team members in the same way. The implication is that GOCB in individualistic cultures is more related to managerial directive or supportive behavior, than in collectivistic cultures.

M. C. EUWEMA ET AL.

A second process through which culture might moderate is the evaluation of the leader behavior in line with specific cultural norms (House et al., 2004). If a leader acts as expected from a leader, this behavior will evoke more positive responses by team members, including GOCB behaviors. We explore this for directive and supportive leadership.

Directive behavior is more accepted in collectivistic societies, and particularly—almost by definition—in societies with high PD, compared with individualistic or low PD societies. In high PD societies directive leadership is seen more as appropriate and effective behavior (2004), and the same is true for collectivistic societies (Gelfand et al., 2004; Hofstede, 2001). Directive behavior is also used considerably more in collectivistic societies, and societies with high PD (Wendt, Euwema, & Van de Vliert, 2003). This is not surprising, as collectivism and high PD do correlate positively (Hofstede, 2001). When directive leader behavior is accepted and seen as effective, the expected negative effects on team processes, such as GOCB, might be less prevalent. Employees in individualistic cultures might not evaluate directive behavior as effective, nor personally stimulating, and therefore reduce their own efforts and investments to achieve team goals. In other words, exhibit less GOCB. Even more so, in low PD societies, directive leadership might be evaluated as dominating, and limiting learning and own initiatives by employees.

In conclusion, we expect a moderating role of IC and PD on the relationship between leadership and GOCB.

Hypothesis 4a . The negative relation between directive leadership and GOCB is stronger in individualistic societies.

Hypothesis 4b . The negative relation between directive leadership and GOCB is stronger in societies with low PD.

Hypothesis 5a . The positive relation between supportive leadership and GOCB is stronger in individualistic societies.

Hypothesis 5b . The positive relation between supportive leadership and GOCB is stronger in societies with low power distance.

Method

Population and sample This study used data from the database of a worldwide operating consulting firm (Hay Group). The

original dataset contains multi-actor data of managers and their subordinates within 473 organizations for a wide range of industries and services, both public and private. Data collection was part of the assessment of management training programs within each of the organizations and this guaranteed a response rate of approximately 100 per cent. For the present study we selected only those countries that were rated by the Hofstede (2001) and the GLOBE (House et al., 2004) studies, and all respondents with missing data were excluded. After these selections, the information concerning the leadership styles of 20 336 managers as rated by 95 893 of their direct reports (the team members) was available. This implies an average of five assessments of leadership styles by team members directly supervised by each manager in this study. For the measurement of GOCB we aggregated scores from each manager

LEADERSHIP AND GOCB ACROSS CULTURES

Table 1. Sample size and mean scores on the main variables per country # of managers

Supportive leadership GOCB Argentina

# team members

Directive leadership

3.50 3.95 4.59 Hong Kong

2.94 3.73 4.65 New Zealand

3.17 3.89 4.43 South Africa

3.04 4.12 4.61 South Korea

and his/her team members (see also data analysis). Seventy percent of these managers were male and

30 per cent female. Average age of the participating managers was 41.5 years, and seniority differs over all levels of management. The final dataset used in the present study included 33 countries. The number of managers in each country varied largely (see Table 1). 2 Table 1 presents the number of managers and the number of team members that rated the managers per country. This table also shows the means for leadership styles and GOCB per country.

2 In the present study we used multilevel analyses, which controls for large differences in sample size at country level (Hox, 2002). However, the minimum number of cases at the country level is under discussion (Maas & Hox, 2001; Muthen, 2005). Therefore,

as an extra check, we also ran the analyses without countries with less than 100 measurement points. The results remained the same.

M. C. EUWEMA ET AL.

Measurements GOCB

GOCB was measured with five items that were developed by the consultancy firm especially for this study. Each item consists of a six-point scale, ranging from A to F, with contrasting statements on the poles of the scale. The first item is shown with both alternatives: ‘The people in my workgroup are always willing to provide help in getting the work done’ (A), versus ‘The people in my workgroup are seldom willing to provide help in getting the work done’ (F). The other items, constructed in the same bi-polar way, are: ’People in my work group . . ..(2) ‘are willing to put in extra time on the job’; (3) ‘will usually go out of their way to make the work group successful’; (4) ‘.are willing to make sacrifices to get the job done; (5) ‘will gladly take on other’s responsibilities in an emergency’. Because the items were developed by the consultancy firm, they do not directly align with the multidimensional measures of OCB used in other studies (Organ et al., 2006). Instead, the items form a general measure of GOCB that includes items that are consistent with the interpersonal facilitation dimension (items 1 and 5) and the job dedication dimension (items 2, 3, and 4) of Van Scotter and Motowidlo’s (1996) measure of contextual performance. Organ (1997) suggests that contextual performance captures many of the helping and cooperating elements of OCB, such as volunteering to carry out task activities that are not formally part of the job; persisting with extra enthusiasm when necessary to complete own task activities successfully, and helping and cooperating with others (Van Scotter, Motowidlo, & Cross, 2000).

Cronbach’s alpha of the GOCB scale was .83. The alpha coefficients varied across countries, as can

be expected, however only in three countries the alpha was below .70. These countries were excluded, also based on the analysis mentioned in note 2 (where we repeated the analysis with countries where we had at least 100 cases).

Note that the content of the GOCB items refer to behaviors at the group level. According to Ehrhart and Naumann (2004) there are no clear guidelines nor consensus on the best way of measuring OCB at the group level. We followed the suggestion by Kozlowski and Klein (2000) to employ measures consistent with the conceptualization of the construct. GOCB is defined as a group phenomenon, and therefore should preferably be measured at that level. Furthermore, the items were descriptive more than evaluative. According to Klein, Conn, Smith, and Sorra (2001) the use of group descriptive items increases within-group agreement. The alternative would be, to use individual or self-referenced items, and aggregate those to the team level. Measuring at the group level offers a more direct measure of the behavior in the group as a whole. When team members show an acceptable level of agreement on GOCB this indeed is a valid measure. A well established procedure to assess the consistency within groups is the intraclass correlation coefficient (Peterson & Castro, 2006). The ICC for GOCB was .69, which is acceptable indeed (Bliese, 2000).

Directive and supportive leadership The two leadership styles were measured with the original scales of Litwin and Stringer (1968), documented in Hay/McBer (2000). Exploratory factor analysis of these items resulted in two factors (see Table 2).

Directive leadership was measured with 7 items (a ¼ .80), and also supportive leadership was measured with 7 items (a ¼ .87). For the directive scale there were nine countries with an alpha reliability coefficient less than .7. These countries were excluded in our previously discussed analysis performed on the countries with a substantial sample size, except Spain (n ¼ 540), with an alpha of .68. For the supportive scale, alphas for all countries were above .7. The items of both scales were measured using bi-polar (six-point) scales, with two opposing responses on both ends in the same format as GOCB (see Table 2 for the items of both styles). Since this is a measurement of leadership style, the

LEADERSHIP AND GOCB ACROSS CULTURES

Table 2. Results of explorative factor analysis of leadership items Factor 1

Factor 2 Directive leadership (a ¼ .80)

Expects employees to follow his/her instructions precisely 0.26 0.67 Motivates employees by letting them know what will happen to them if

0.04 0.57 their work is unsatisfactory Requires employees to submit detailed reports of their activities

0.08 0.69 Makes most decisions for employees

0.26 0.68 Supervises employees very closely

0.14 0.76 Supervisor has to lay out goals and guidelines, otherwise subordinates

0.11 0.68 will be passive and get nothing accomplished Expects employees to carry out instructions immediately

0.14 0.66 Supportive leadership (a ¼ .87)

Works hard to ease tensions whenever they arise in work group 0.69 0.02 Encourages employees to talk to him/her about personal problems

0.72 0.07 Devotes a great deal of time to employees’ job security and fringe benefits

0.68 0.06 Works to develop close personal relationships with employees

0.77 0.05 Relies on what he/she learns through personal contact with employees to

0.79 0.07 use each person’s talent most effectively Frequently demonstrates concern for employees

0.79 0.21 ‘Believes subordinates’ feelings are as important as the task at hand

0.71 0.27 Eigenvalue

4.59 2.73 % variance explained

scores from the managers themselves were omitted, and only the scores of the employees were used, to avoid self-serving biases. The ICC for directive leadership was .70, and for supportive leadership .63, both acceptable scores.

Language issues are always a major concern in cross-cultural studies. Accordingly, the items were all translated from English into the languages of the participating countries by native speakers, using the so-called application mode of translation (Van de Vijver & Tanzer, 2004). With this method, it is implicitly assumed that the underlying construct is appropriate for each cultural group and that a simple, straightforward translation will suffice to get an instrument that adequately measures the same construct in the target group. The translators (consultants) were trained in the concepts, and familiar with the societies’ culture, which should improve the chance that their translations do represent the concepts indeed.

Measurement equivalence In cross-national studies, measurement equivalence is both important and difficult to achieve. There

are different techniques to check for measurement equivalence (Peterson, et al., 1995; Scandura, Williams, & Hamilton, 2001; Van de Vijver & Leung, 1997). A structural analysis was conducted to prove that the instruments have a stable structure over cultures. This procedure is recommended by several authors (Peterson et al., 1995; Reise, Widaman, & Pugh, 1993; Scandura et al., 2001). We used Confirmatory Factor Analysis to verify that the leadership model is equivalent across countries. We used AMOS 7 (Arbuckle, 2006) to contrast the two leadership styles in a multi-group analysis. The instrument was designed in the US, and therefore we contrasted the US data with countries where we had at least 100 cases (Byrne, 2001). The goal of this test is to check if the same factor model

M. C. EUWEMA ET AL.

holds for both populations, by constraining parameters to be equal for the groups. A model where the structure was constrained to be equal could meet the threshold; we found evidence for equivalence, indicating that the construct has the same meaning in both countries (Browne & Cudeck, 1993). The average RMSEA was .053, CFI .902, and NFI .899, which is acceptable. Arbuckle (2006) indicates an RMSEA lower than .08 is acceptable, and Browne and Cudeck (1993) even use as a rule of thumb an RMSEA lower that .1. The weakest RMSEA was in a multi-group analysis with US and Malaysia: .060 (CFI .870, NFI .867).

The multi-group factor analysis for GOCB also gave evidence that we can use and can compare results from this instrument across cultures. The average RMSEA was .044, CFI .975, and NFI .974. Only the multi-group model with Malaysia again was relatively poor, with an RMSEA .090, CFI .905,

and NFI .904. 3 Overall, we conclude that almost all countries meet the criteria for measurement equivalence. Malaysia has the weakest score in this respect, but is still acceptable for the current research purposes, as we analyze cultural dimensions, and not individual countries.

Cultural dimensions Two types of operationalization of the cultural dimensions were used; Hofstede’s (2001) and GLOBE

(House et al., 2004). Smith (2006, p. 917), commenting on the debate between Hofstede and the GLOBE researchers, concludes ‘if our focus is upon the most basic and normative aspects of culture, then the Hofstede and GLOBE procedures are equally appropriate’. We used the original country means for individualism (IDV) and the PD index (PDI), reported by Hofstede (2001), and IGC and PD, as reported by House et al. (2004). GLOBE offers dimension scores for values (‘should be’) and practices (‘as is’). In this study the practices are used, as these reflect more the theoretical approach of Hofstede (2001), aiming to describe actual cultural differences (Hofstede, 2006; Smith, 2006).

Data analysis To test our hypotheses, multi-level analysis is most appropriate, given the different levels in this study:

team level and country level (Bliese & Jex, 1999; Kozlowski & Klein, 2000). At team level (that is level 1), leadership styles (aggregated scores from the team members) and GOCB (aggregated scores from manager and his/her team members) were measured. At the country level (that is level 2), Hofstede’s and GLOBE dimension scores were used. Data analysis was performed using HLM 6.0 (Raudenbush, Bryk, Cheong, & Congdon, 2005).

Results

Table 3 presents descriptive statistics and correlations for the measures at team level. Directive and

3 We skipped a check for measurement invariance using item response theory techniques (Scandura et al., 2001), because the scales of our instruments are six point-scales, and in this case item response theory and structural equation modeling are

equivalent (Glo¨ckner-Rist & Hoijting, 2003).

LEADERSHIP AND GOCB ACROSS CULTURES

Table 3. Descriptive statistics and correlations among the study variables at team level

2 Directive leadership 2.94 0.63 .25 3 Supportive leadership

3.97 0.67 .29 .22 Notes: N ¼ 20.336 for the measures at team level.

p< .01.

supportive behavior (r ¼ .29).

Table 4 presents the correlations among the study variables at country level. There are no significant correlations between GOCB and the cultural dimensions. Directive leadership is related with all four cultural dimensions, and supportive leadership is related with IGC and GLOBE’s PD. Further, there is a conspicuously high negative association between Hofstede’s IDV

GLOBE’s PD (r ¼ .66, p < .01). These correlations are in line with the reported strong correlation between IC and PD by Hofstede (2001). IDV and PD are by no means independent cultural dimensions; individualistic societies tend to have lower PD, and collectivistic societies tend to have high PD. This has implications for our hypothesis testing as well. Please note that Hofstede’s IDV is highly negatively

Hypotheses testing Analysis of variance for GOCB by country showed that 9 per cent of the variance is at country level

( p < .001); that is, 9 per cent of the variance is explained by the grouping structure in the population, and therefore, multi-level analysis is required. At the managerial or team level (i.e., level 1), the directive and supportive leadership styles and GOCB were used. At the country level (i.e., level 2), the measures of Hofstede and GLOBE were used. To test the hypotheses, we first centered the level 2 variables (Hofmann & Gavin, 1998) and built interaction terms (see Aiken & West, 1991). The

Table 4. Correlations among the study variables at country level 1 2 3 4 5 6

1 GOCB 2 Directive leadership

3 Supportive leadership

4 Hofstede’s IDV

.77 6 GLOBE’s IGC

5 Hofstede’s PD

.82 .82 7 GLOBE’s PD

.47 .46 .66 Notes: N ¼

33 for measures at country level. The team level measures are aggregated at the country level. p< 0.05; p< 0.01.

M. C. EUWEMA ET AL.

Table 5. Multi-level estimates for models predicting GOCB with Hofstede’s measures

Model 1

Model 2

Model 3 g g g

Fixed part Intercept

4.634 Directive leadership

.047 Supportive leadership

.150 Hofstede’s IDV (H-IC)

.025 Hofstede’s PD (H-PD)

Random part Variance cons (SE)

.021 Variance residual (SE)

.237 Log likelihood

28 521.070 Notes: After the estimation of the intercept-only model (i.e., Model 1), the variables were entered in two steps. First, the main

effects were entered. Next, the interaction terms were included. The level 1 variables were grand mean centered (Raudenbush & Bryk, 2002).

p< 0.01.

predicted two-way interaction effects were then tested in two separate analyses for the Hofstede and GLOBE dimensions. After the estimation of the empty model, variables were entered in two steps. First, the main effects were entered. Next, the interaction terms were included. The results of the two analyses are presented in Tables 5 and 6.

Table 6. Multi-level estimates for models predicting GOCB with the GLOBE measures

Model 1

Model 2

Model 3 g g g

Fixed part Intercept

4.640 Directive leadership

.056 Supportive leadership

.148 GLOBE’s IGC (G-IGC)

.056 GLOBE’s PD (G-PD)

Random part Variance cons (SE)

.022 Variance Residual (SE)

.237 Log likelihood

28 526.880 Notes: After the estimation of the intercept-only model (i.e., Model 1), the variables were entered in two steps. First, the main

effects were entered. Next, the interaction terms were included. The level 1 variables were grand mean centered (Raudenbush & Bryk, 2002).

p< 0.01.

LEADERSHIP AND GOCB ACROSS CULTURES

Leadership styles and GOCB We expected a negative relation between directive leadership and GOCB (H1), and a positive relation

for supportive leadership and GOCB (H2), and the correlations in Table 2 confirm this. Model 2 in Table 5 shows the results of the multi-level analysis. Directive leadership has a negative relation with

(g ¼ .138, SE ¼ .004, p < .01), offering support for both hypothesis 1 and hypothesis 2.

Cultural dimensions and GOCB The relationships between IC, PD, and GOCB were separately estimated for the Hofstede and the

GLOBE dimensions. As can be seen from Table 5 Model 2 (Hofstede), and Table 6 Model 2 (GLOBE) neither IC nor PD is associated with GOCB. Consequently, there is no support for our third hypothesis.

Moderating effects of culture on the relationship between leadership and GOCB Next, our hypotheses four and five predicted a moderating effect of IC and PD on the relation between

directive and supportive leadership and GOCB. The results for the Hofstede dimensions are presented in Table 5, Model 3. The results for the GLOBE dimensions are presented in Table 6, Model 3. The significant interaction for directive behavior is plotted in Figure 1.

Figure 1 shows the moderating role of Hofstede’s IDV–collectivism on the relationship between directive leadership and GOCB. In collectivistic countries, directive leadership hardly influences the extent of GOCB, whereas in individualistic countries, more directive leadership signifies a strong

found for GLOBE’s IGC (Table 6, Model 3: g ¼ .028, SE ¼ .003, p < .01). The results are in line with our expectation in hypothesis 4a. We expected also that societal PD moderates the relation between

IC and PD, are entered simultaneously. In this combination with IC, PD does not have an additional explaining value. When analyzed separately, PD does have a significant moderating effect, as expected.

Figure 1. Moderating role of Hofstede’s IDV on the relationship between directive leadership and GOCB

M. C. EUWEMA ET AL.

These two cultural dimensions, IC and PD, can however not be seen as independent. Consequences are further elaborated in our discussion.

Hypotheses 5a and 5b predict a moderating effect of IC and PD on supportive behavior. As Table 5

p<.01). The relation between supportive leadership and GOCB is slightly stronger in collectivistic societies. This is not in line with our Hypothesis 5a. GLOBE’s PD does not moderate the relation

hypotheses.

Discussion

To the best of our knowledge, this is the first large scale empirical study that focuses on GOCB in an international context. Several authors emphasize both the need to study OCB at group or team level (Ehrhart & Naumann, 2004; Schnake & Dumler, 2003) and to relate OCB to culture (Gelfand et al., 2007; Organ et al., 2006; Podsakoff et al., 2000). OCBs have been studied mostly at the individual level, showing the relevance for organizational outcomes, including productivity and well being (Organ et al., 2006). With an increasing focus on teamwork in current organizations, the importance of OCB at unit or team level is growing, and there is clearly a need to relate our knowledge at individual levels, to group phenomena. GOCB might offer a bridge between these two domains (Chen et al., 2002; Ehrhart & Naumann, 2004; Schnake and Dumler, 2003). A survey assessing both managerial behavior and GOCB as perceived by a large sample of managers and their subordinates from 33 countries across the world shows the moderating role of culture on the relationship between managerial behavior and GOCB, as well as the negative relation between directive leadership and GOCB, and a positive relation between supportive leadership and GOCB.

Culture and GOCB

A first surprising result of our study is that there appears no direct relationship between GOCB in organizations and cultural dimensions at the societal level. We expected to find relations between arguably the two most relevant cultural dimensions, IC and PD, and the level of GOCB. Evidently, the direct influence of societal cultural differences on GOCB can be neglected. Several authors (e.g. Gelfand et al., 2004) emphasize the group orientation at work in collectivist cultures, whereas individualistic cultures almost by definition emphasize the importance of the individual employee. This clearly is not reflected in differences in GOCB. Several explanations are possible.

First, several cultural biases and methodological limitations might play a role. For example, the meaning of the construct of GOCB differs between cultures (Gelfand et al., 2007; Organ et al., 2006). We measured GOCB with a short, five item scale. This does not permit differentiating between important dimensions of OCB, such as helping behaviors, or working overtime. These elements though, can vary across cultures. Further studies therefore require a more differentiated measurement of GOCB.

A second explanation might be that GOCB has different causes in different cultures. For example, whereas GOCB in collectivist cultures is facilitated by the cultural group orientation, it might be inhibited by a lower commitment to the organization as a whole (Hofstede, 2001). These different

LEADERSHIP AND GOCB ACROSS CULTURES

processes balance each other, with the result we find in the present study. A final methodological explanation might be that there is limited variance in GOCB in our study. The mean score worldwide of

4.7 on a scale from 1 to 6. This implies a high level of GOCB worldwide, together with a SD of 0.54, results might be partly explained by the method used.

Dokumen yang terkait

Analisis Komparasi Internet Financial Local Government Reporting Pada Website Resmi Kabupaten dan Kota di Jawa Timur The Comparison Analysis of Internet Financial Local Government Reporting on Official Website of Regency and City in East Java

19 819 7

ANTARA IDEALISME DAN KENYATAAN: KEBIJAKAN PENDIDIKAN TIONGHOA PERANAKAN DI SURABAYA PADA MASA PENDUDUKAN JEPANG TAHUN 1942-1945 Between Idealism and Reality: Education Policy of Chinese in Surabaya in the Japanese Era at 1942-1945)

1 29 9

Improving the Eighth Year Students' Tense Achievement and Active Participation by Giving Positive Reinforcement at SMPN 1 Silo in the 2013/2014 Academic Year

7 202 3

Improving the VIII-B Students' listening comprehension ability through note taking and partial dictation techniques at SMPN 3 Jember in the 2006/2007 Academic Year -

0 63 87

Analisi pesan dakwah pada lirik lagu kebesaranmu group band sti 2

2 155 101

The Correlation between students vocabulary master and reading comprehension

16 145 49

An analysis of moral values through the rewards and punishments on the script of The chronicles of Narnia : The Lion, the witch, and the wardrobe

1 59 47

Improping student's reading comprehension of descriptive text through textual teaching and learning (CTL)

8 140 133

The correlation between listening skill and pronunciation accuracy : a case study in the firt year of smk vocation higt school pupita bangsa ciputat school year 2005-2006

9 128 37

Transmission of Greek and Arabic Veteri

0 1 22