The effect of team collectivism on team performance : a meta-analysis

(1)

MSc. In Business Administration – Leadership & Management Track

The Effect of Team Collectivism on Team Performance;

a Meta-Analysis

Name: Rico Habraken

Student Number: 11352779

Date of Submission: 23rd_{of June, 2017}

Version: Final Draft

Institution: Amsterdam Business School, University of Amsterdam

1st_{Thesis Supervisor:} _{Jan Luca Pletzer}

(2)

Statement of Originality

This document is written by student Rico Habraken who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Acknowledgements

This Master’s thesis, written as the final curriculum component for obtaining a Master of Science degree (MSc) in Business Administration from the University of Amsterdam, could not have been realized without the help, advice, and support of many contributors. Several acknowledgements are therefore in order.

Special thanks go out to Jan Luca Pletzer, supervisor and PhD candidate at both the University of Amsterdam and the VU (Vrije Universiteit) Amsterdam, for his brief but critical meetings, detailed feedback, and extensive meta-analytic expertise.

Many thanks go out to Peter and Cora Habraken, my parents, who have never stopped believing in me, and support me in all of my endeavours.

Many thanks go out to Darta Rozentale, fellow student of the Leadership & Management track, for always being available when stress levels rose too high and an expert opinion was needed. Thanks to Rik Populier, my good friend, for providing me with the needed relaxation and leisurely competition when thesis times got tough.

And last but certainly not least, special thanks go out to Cassandra Badiali, for her unwavering support throughout the last four years of my academic career and her unrelenting determination to make me succeed in everything I take on.

(4)

Table of Tables

Table 1. Search terms and database... 15 Table 2. Averages and range for moderator variables... 18 Table 3. Study characteristics and effect size data of studies included in the meta-analysis ... 21 Table 4. Descriptive statistics for k = 20 studies in the quantitative meta-analysis. ... 23 Table 5. Results of subgroup analysis ... 27 Table 6. Results of meta-regression analysis ... 28

Table of Figures

Figure 1. Study selection and exclusion criteria. ... 16 Figure 2. Forest plot of the relationship between team collectivism and team performance ... 24 Figure 3. Funnel plot of the estimated variability and effect size r for each study ... 25 Figure 4. Forest plot of the cumulative analysis ... 26 Figure 5. Forest plot of one study removed analysis ... 27

(7)

Abstract

Due to the changing nature of work through increasing complexity, necessity for creativity, and a need for different sets of skills and expertise, companies increasingly organize work in teams. Therefore it is incumbent to find out which factors influence team performance and to what extent can they be manipulated. Recent findings regarding the team collectivism-team performance link have been contradictory with both positive and negative effects. This study aims to meta-analytically investigate this relationship to find stronger support for a positive effect between the team collectivism and team performance variables.

By means of a systematic search strategy, 22 independent effect sizes from 20 studies have been included in the quantitative meta-analysis with a total sample size of 11,848 participants. According to the random effects model, the overall mean weighted correlation was small to medium and significant (r = .154, 95% confidence interval: .06, .25).

Additional subgroup and meta-regression moderator analyses indicated a positive significant moderating effect for the percentage of females on the team collectivism-team performance relationship. Location, age, and team size did not indicate significant moderating effects. The results of this meta-analysis reject the null hypothesis that there is no relationship between team collectivism and team performance, and provide strong support for the alternative hypothesis that higher scores of team collectivism will positively affect team performance. The study concludes with a discussion of implications, limitations and possibilities for future research.

(8)

Chapter 1. Introduction

Does national or societal culture affect team performance? That is the question being posed in this research. Seemingly obvious, yet in truth ambiguous and quite possibly influenced by many factors, team performance has become increasingly important in modern day business (McDowell, Agarwal, Miller, Okamoto, & Page, 2016; Schumpeter, 2016). In recent years, organizational design has changed to incorporate self-organized work teams in organic businesses, as opposed to the bureaucratic and hierarchical mechanistic organizations (Daft, 2015; Wilden, Gudergan, Nielsen, & Lings, 2013). Moreover, diversity, with regard to gender and culture in particular, has been a ‘hot’ topic for politicians and business executives (Kearney, Gebert, & Voelpel, 2009; Stahl, Maznevski, Voigt, & Jonsen, 2010). Therefore it is of academic and practical importance that researchers find answers to the conundrum of team performance, for example, what makes effective teams tick and which factors influence their performance? This research will examine the accumulated knowledge of the extant literature in order to attempt bridging the research gap and find support for a positive relationship between the effects of societal culture, in particular team collectivism, and team performance.

The main focus of this research lies on team level culture and its effect on team performance. More specifically, the effect of team collectivism on team performance because this cultural dimension of collectivism indicates a preference for group work and for placing higher value in the group rather than in the individual (Dierdorff, Bell, & Belohlav, 2011). Following that line of reasoning, teams with higher levels of collectivism might value working in teams more, subsequently exert additional effort on team tasks, which could likely lead to higher team performance.

(9)

The increasing emphasis on culture in organizations (Kearney et al., 2009) and an increase in self-organized work teams in modern organizational design (Daft, 2015; McDowell et al., 2016) make this research practically valuable as well. It is self-evident that practitioners and business executives will benefit from high levels of team performance. If strong evidence is found that cultural values, background and characteristics on a team level directly affect team performance, it is incumbent that managers and team leaders are well informed in order to adapt the team composition in favour of efficiency and effectiveness. It can thus be argued that this research has valuable practical implications.

Given the increasing recognition of culture playing a crucial role in the organizational context and the developments regarding self-organized work teams in businesses (McDowell et al., 2016; Stahl et al., 2010) it is of importance to discover to what extent culture has an effect on performance, and in particular team performance. Since there are contradicting findings between studies regarding the effects of culture on team performance (Chen et al., 2015; Gibson, 1999; Kirkman & Shapiro, 2005), a analytic design is chosen, since a meta-analysis gives an overview of the current body of work by combining findings from multiple studies, simultaneously increasing the statistical power of findings (Borenstein, Hedges, & Higgins, 2009). To conclude, the key question of this research is formulated as;

(10)

Chapter 2. Literature review

This chapter contains the literature review for this thesis, shedding light on the theoretical concepts of teams, team performance, culture, collectivism and their interrelations.

Teams

Teams and groups of people have been around since the birth of mankind (Harari, 2014). There is however a large difference between a group of people and a team of people. While a group is a combination of two or more people, sharing the same interests and having continuing interaction (Arnold & Silvester, 2005), a team, or work team, is an interdependent group of people with skills that complement one another, who are committed to (1) a common mission, (2) performance goals, (3) and an approach for which they feel and hold members of the team mutually accountable (Arnold & Silvester, 2005). Note that the main differences between groups and teams lie with the commitment towards a common objective and the need for interdependency. Without an objective there is no need for interaction and cooperation, and without interdependency there is merely a group working on individual tasks.

In modern day organic businesses, many tasks or projects are executed by means of teams and team work. This might be due to the change in the nature of work (Johns & Gratton, 2013). In some tasks, more creativity is required to come up with solutions, therefore it is necessary to bring more perspectives to the table (West, 2002). Projects might have too large a scope for a single employee to handle and dividing the work in small teams offers a way to combat this problem. Work is becoming increasingly more complex, thus demanding a multitude of expertise and competencies which cannot be found in a single person. The literature on team work has also found that it might be more motivating to work in teams as opposed to solitary confinement in an office cubicle (Kozlowski & Ilgen, 2006). Lastly, projects organized in teams are beneficial when it comes to making difficult decisions if these decisions need to be accepted by a larger audience.

(11)

Still, not all tasks need to be managed in teams. When tasks are easily handled by an individual, when there is no possibility of creating a common goal, or when tasks have no interdependence, an individual work organization is preferred.

It is not only beneficial for companies to organize work in teams when projects and tasks have the characteristics described above. Employees also have reasons to join and be part of a team according to Kozlowski and Ilgen (2006). It allows employees to feel secure, since they are not alone in their endeavours. It gives a certain status to be part of a work team and provides the means to identify with others, subsequently raising self-esteem. Moreover, there is a sense of goal achievement upon successful execution of projects, which is perceived as stronger than individual achievement (Cohen & Levesque, 1991). Another reason to join a team is an increase in (political) power, when connections with other employees are made and finally, team work can increase a sense of affiliation or belonging (Hogg, 2016). This sense of belonging is particularly important for employees and companies alike, because increased affiliation is argued to be directly related to lower turnover rates and higher organizational commitment (Allen & Meyer, 1990). Commitment in this sense links back to the definition of what constitutes a team, and is therefore needed as the basic foundation upon which teams are built. Team performance

Team performance is defined as the organizational outcome of teamwork and the degree to which teams have accomplished their common objective (Bell, 2007). Team productivity, team effectiveness, and team efficiency are consistently used in the literature as substitutes and synonyms for team performance and are thus also relevant for this study (Bell, 2007). Theoretically it is important to open the black box of factors relating to team performance in order to uncover what the predictors of team performance are and how they are influenced. Practically, a better understanding of team performance might give business executives and

(12)

practitioners additional tools to increase the level of team performance and subsequently other organizational outcomes such as profit, growth, and market share (Salas et al., 2008).

To some extent, this black box of predictors of team performance has already been opened for some time and research on determinants of team performance has been conducted by numerous authors (De Dreu & Weingart, 2003; Gully, Incalcaterra, Joshi, & Beaubien, 2002; Salas et al., 2008; Schneid, Isidor, Li, & Kabst, 2015). According to the extant literature, team cohesiveness, or the extent to which team members are attracted and attached to each other and motivated to be part of the team, is an important predictor of team performance (Beal, Cohen, Burke, & McLendon, 2003). Higher team cohesiveness is argued to increase team productivity. However, this only holds true in combination with common norms held by the team concerning the level of performance. When both variables are high, teams will have higher performance than other combinations. However, when team cohesiveness is high, but the norms concerning the level of performance are low, team performance might sustain damage and turn out to be lower than when team cohesiveness would also be low. Therefore the aim of companies should be to have shared norms for high performance and high team cohesiveness in order to obtain the best results.

Other predictors of team performance are team size (Belbin, 2011), team conflict (De Dreu & Weingart, 2003), and team tenure (Keller, 2006). A large part of working in a team is the ability to cope with team processes. Depending on team task, such as making decisions during a project, a smaller team size would increase team performance. A well-known stage-process model for teams was created by Tuckman and Jensen (1977) explicating the inception and termination of teams by means of five stages (forming, storming, norming, performing, and adjourning). Some of the aforementioned predictors have an effect on the stages in this model, subsequently affecting team performance. Some teams need conflict to be more productive in the storming phase. Other teams might benefit from team members having more team tenure,

(13)

meaning they are familiar with the team processes and as such make the overall trajectory easier to go through, aiding team performance.

Furthermore, team-based incentives (Milne, 2007), and team competition (West, 2002) might also have an effect on team performance, team productivity, or team effectiveness. Team-based incentives, as opposed to individual incentives, give team members more motivation and support for their common goal, which links back to the main criteria for being part of a team. Team competition has a similar effect on motivation and also affects team cohesiveness positively. Increased motivation and team cohesiveness are argued to increase team performance (Beal et al., 2003).

Team composition is also deemed an important predictor of team performance (Bell, 2007). Naturally, different projects demand different sets of knowledge, skills, abilities, and other personal characteristics (Buller & McEvoy, 2012), and therefore the team composition must be in line with the task at hand. A misalignment might likely lead to lower team performance. Moreover, the extant literature indicates a distinction between surface-level and deep-level composition variables (Bell, 2007). Surface-level composition variables are based on more obvious demographic characteristics such as age, ethnicity, level of education, and organizational experience. All of these characteristics are easily identified after short interactions with team members. Deep-level composition variables however, are related to personality, values, and attitudes. These characteristics are more difficult to pinpoint at first glance. Deep-level composition variables are argued to have a higher impact on team performance than surface-level composition variables. To illustrate this with an example, cultural diversity (surface-level) is argued to have a smaller effect on team performance than cultural values (deep-level). This research will take a closer look at these cultural values on a team level and to what extent they affect team performance.

(14)

Culture

The topic of culture in itself is broad, and many misconceptions exist of what culture actually is. Culture has been defined as consisting of “commonly experienced language, ideological belief systems, ethnic heritage, and history” (House et al., 1999, p.13-14). Moreover, culture is argued to have multiple layers; on a country-level (Hofstede, 1984; House, Hanges, Javidan, Dorfman, & Gupta, 2004), within the organizational context (Schein, 1990), within groups (Johnston, 2013), and within the individual (Kristof-Brown, Zimmerman, & Johnson, 2005). Other distinctions of cultural characteristics include surface-level and deep-level differences, as briefly discussed in the previous section.

By definition, different levels of culture imply potentially overlapping, but also contradicting values or basic assumptions (Schein, 1990). For example, the definition of national culture differs from organizational culture in terms of language (jargon spoken within an organization), shared organizational values, and firm history (Schein, 1990). Given these statements, it is possible for an individual or a team to have certain societal culture characteristics which can inherently be different from the characteristics portrayed within their respective organizations or institutions.

These differences in cultural characteristics are argued to have an effect on a multitude of variables such as organizational commitment (McKinnon, Harrison, Chow, & Wu, 2003), job performance (Stahl et al., 2010), and organizational citizenship behaviour (Turnipseed & Murkison, 2000). Whether these cultural characteristics also directly have an effect on team performance remains elusive, because studies researching the relationship between culture and team performance appear contradictory (Chen et al., 2015; Gibson, 1999; Kirkman & Shapiro, 2005). It is therefore theoretically and practically important to find stronger support for the effect of culture, and in particular team culture, on team performance.

(15)

One of the most famous contributions to cross-cultural research has been Geert Hofstede’s research on culture and the inception of cultural dimensions (Hofstede, 1984). Through the analysis of a large database with value scores of IBM employees, spread worldwide across approximately seventy countries, Hofstede came up with four cultural dimensions. These four cultural dimensions (Power Distance, Uncertainty Avoidance, Individualism, and Masculinity vs. Femininity) can be viewed separately as continuums, on which countries can either score higher or lower. These resulting scores can then be interpreted and light can be shed on which cultural dimensions countries differ. To date, Hofstede has added two more dimensions to his framework, consisting of Long Term Orientation and Indulgence (Hofstede, 2011). The knowledge from this research opened up a whole new field of study and sparked the interest of multinationals to develop cross-cultural competencies and allow for mutual understanding of organizations conducting business abroad.

Another well-known large-scale research platform is project GLOBE (House et al., 1999). Project GLOBE was specifically designed to uncover if leadership styles are culturally contingent and whether different cultures prefer or detest certain styles. Building on Hofstede’s research they identified nine cultural dimensions as opposed to Hofstede’s six dimensions. Comparing the two research projects’ dimensions, Individualism is split up in two separate Collectivism dimensions, (I) pertaining to the extent to which institutions reward collective distribution of resources and (II), reflecting the degree of pride and loyalty of individuals towards family and organizations (House et al., 1999). Hofstede’s Uncertainty Avoidance and Power Distance dimensions remain unchanged and Masculinity vs. Femininity is adapted to Gender Egalitarianism in project GLOBE, referring to “the extent an organization or society minimizes gender role differences” (p. 25).

(16)

The other four dimensions of project GLOBE are Assertiveness, Future Orientation, Performance Orientation, and Humane Orientation. These are however not discussed further since their relevance is deemed of lesser importance with regard to their effect on team performance and the feasibility of encountering sufficient studies measuring these dimensions. Even though it might be argued that all of these cultural dimensions influence team performance to some extent, the cultural dimension that stands out the most is Individualism, or Collectivism in many studies (Dierdorff et al., 2011; Earley, 1989; Jackson, Colquitt, Wesson, & Zapata-Phelan, 2006). From this point forward, the term Collectivism will be used to indicate this cultural dimension. Collectivism is one of the most researched cultural dimensions, which would indicate a high number of studies include this dimension when culture is involved. In addition, Collectivism has several underlying values that can logically be combined with the requirements for working in a team and which are necessary to be successful at delivering high team performance. Therefore, Collectivism is selected as the main cultural dimension to be investigated with regard to team performance for this research.

Collectivism

Collectivism has several definitions depending on the field of expertise. Politics and sociology deal with collectivism in slightly different ways than is meant in this research and therefore the cultural definition is used. In the cultural literature, collectivism, or a collectivist culture is defined as a culture where individuals will value the common goals of the group over their own goals or adapt their individual goals to be aligned with the group goals. In collectivistic cultures, a person’s self-image is defined as ‘we’ (Hofstede, 1984).

(17)

Individualism, or individualistic culture, can be seen as the opposite of Collectivism, in the sense that individuals will put their personal goals before the common goal of the group. In individualistic cultures, a person’s self-image is defined as ‘I’ and enjoyment is far more important than duty (Hofstede, 1984).

Triandis and Gelfand (1998) distinguish between two types of collectivism, namely horizontal and vertical collectivism. They argue that there is a difference between horizontal and vertical social relationships. A horizontal social relationship refers to equality, or the individual being no different than other individuals. Quite the opposite is the case with vertical relationships, where the individual is seen as different from others, specifically indicating hierarchy as the distinguishing factor. Combining these social relationships with collectivism, two types emerge. In horizontal collectivism, individuals are deemed equal and there is consensus towards the group or team in terms of common goals, importance of team cohesion and being dependent on each other. In vertical collectivism, others are viewed as being part of in-groups and groups. Individual goals can be set aside for the in-group, but not for the out-group. Concerning hierarchy, individuals will conform to in-group authority even if this is at odds with their own values.

A slightly different distinction is made between institutional collectivism and in-house collectivism coined by House et al. (2004) as part of the GLOBE project. Like horizontal collectivism, institutional collectivism also has roots in equality, but on a societal institutional level rather than the individual level, and emphasizes “the degree to which collective distribution of resources and collective action is encouraged and rewarded” (p. 25). In-house collectivism is directed at organizations and families and the extent “to which individuals express pride loyalty, and cohesiveness” (p. 25).

Due to these slight differences between definitions of collectivism, and with regard to the manner of collecting data on culture and cultural dimensions in general, there has been some

(18)

criticism on Hofstede’s work in particular (Heine, Lehman, Peng, & Greenholtz, 2002). Collectivism, together with Individualism, has been conceptualized to contain up to four distinct dimensions (Triandis & Gelfand, 1998), has been referred to as Allocentrism and Idiocentrism (Katz, 1999), and has been reported as encompassing two separate versions of the self, independent vs. interdependent (Singelis, Triandis, Bhawuk, & Gelfand, 1995). Moreover, cultural data has been collected from individual surveys, which are then mostly aggregated and generalized to an entire nation. Some authors question whether this is the appropriate way to discover cultural differences and whether more significant and relevant results would be obtained when actual cultural-level characteristics are examined, for example divorce rates and voting patterns (Vandello & Cohen, 1999), or cultural messages in magazine advertisements (Han & Shavitt, 1994).

Despite these criticisms, Hofstede’s body of work is considered to be relevant today and is still being reproduced and accepted by the majority of authors on culture as the guiding framework for cultural dimensions, albeit with slight adaptations (Earley, Gibson, & Chen, 1999; Maznevski, DiStefano, Gomez, Noorderhaven, & Wu, 1997). All variations of collectivism seem to converge on the essence of goals, as described in the first paragraph on collectivism, where individuals value the attainment of group goals more than individual goals or adapt these individual goals to coalesce with group goals. This particular notion of common goals is of critical importance when it comes to indicating the relationship between collectivism and team performance, which is further elaborated in the next section.

Focusing more specifically on teams and team performance, it is logical to also examine the same level of the independent variable; the team level of collectivism. Team collectivism refers to “the degree to which a team values loyalty, responsibility, and team cohesiveness in the team family” (He, Baruch, & Lin, 2014, p.955). Team collectivism will be the independent variable in this thesis and the focus of research.

(19)

Team collectivism and team performance

Team work, as outlined previously, requires interdependency, competency-fit with other members and commitment to (1) a common mission, (2) performance goals, (3) and an approach for which they feel and hold members of the team mutually accountable (Arnold & Silvester, 2005). Effective and successful team work leads to the desired outcome, team performance. Team collectivism, or a collectivistic team culture, indicates the preference of the team for team work, cooperation and valuing team goals more than individual goals or that the individual goals are aligned with the team goals (He et al., 2014). When working towards a common goal and when one is dependent on other’s to reach this goal, it can logically be stated that a better score or higher level of team collectivism will improve team performance.

Moreover, high team cohesiveness, the extent to which team members are attracted and attached to each other and motivated to be part of the team, has been argued to influence team performance positively (Beal et al., 2003). Additionally, the effect of team cohesiveness depends on the shared common norms regarding team performance. These norms can theoretically be extended to include values and attitudes (deep-level composition variables) concerning teamwork and the importance of the team, which in turn might have a positive effect on team performance when these values are higher (Bell, 2007). Therefore, multiple theoretical arguments are in favour of a positive effect regarding the team collectivism-team performance relationship. Consequently, the null and alternative hypothesis can be formulated as;

H0: High scores for team collectivism will not have a positive effect on team performance. H1: High scores for team collectivism will have a positive effect on team performance.

(20)

Chapter 3. Methodology

As outlined in the introduction section, the research design proposed for this research is meta-analytical. Meta-analysis is a research method for systematically combining qualitative or quantitative data from several primary studies in order to develop a conclusion that is of higher statistical power than a single study (Borenstein, Hedges, Higgins, & Rothstein, 2009). The statistical power of the conclusion is increased due to larger numbers or quantities of subjects. Usually, meta-analysis is used as a method when studies have conflicting results or to develop a better estimate of effect sizes between the examined variables. In this case, the findings of culture’s effect on team performance are ambiguous and previous meta-analyses on the topics either do not focus on team performance or measure culture and cultural dimensions differently than this research intends (Stahl et al., 2010; Taras, Kirkman, & Steel, 2010).

Systematic search strategy

On April 11, 2017, the systematic search was conducted in Web of Science. The search strategy used is outlined in table 1. Only English studies were included in the search, which yielded 153 sources. Moreover, the first author of Taras et al. (2010) was contacted for additional sources, since his meta-analysis made mention of nine studies that also measured the effect of collectivism on team or group performance, which were however not discussed in that study. Dr. Taras provided these nine studies and attached an additional file containing a list of 551 studies which could also be of interest for the research question. These 560 studies were brought down to 52 studies of interest, based on their title and after duplicates had been deleted. The total amount of studies examined added up to 713 in total (Web of Science and the lists provided by Dr. Taras).

(21)

Table 1. Search terms and database (only English studies included)

Number of

sources (k) Search terms and limits

* _{Database (1975 – April 2017)}

153 [TS=(Hofstede OR Globe OR “Cultural dimension” OR individualism OR collectivism OR

individualism-collectivism)] AND [TS=(“team performance” OR “group performance” OR “work group performance” OR

teamwork OR “team work” OR “collective performance” OR “group work” OR groupwork)]

Web of Science

52 Title indicating culture AND/OR (team) performance Dr. Vas Taras *_{TS = Topic}

Study selection

The title and abstract of the combined 205 studies in table 1, were first screened. Studies were included if they showed clear signs of measuring a combination of culture, collectivism, or individualism with team performance or performance on a group task. When it was not completely clear from the abstracts if a study measured the relationship of interest, but showed indications of these variables, they were included to be examined in full in order to not miss any useful study. This resulted in 71 studies, which were examined in full for their appropriate measure of collectivism or individualism, team performance (or an equivalent of team performance), and the inclusion of effect sizes. Additionally, a quantitative design had to be used and for effect sizes the Pearson product moment correlation r had to be reported. More importantly, studies were excluded if either collectivism, individualism or performance were not reported on a team level. 51 studies were excluded based on the various criteria indicated in the PRISMA flowchart (Moher, Liberati, Tetzlaff, & Altman, 2010) in figure 1, leaving 20 studies to be included in the quantitative meta-analysis. The data of these 20 studies were extracted by two authors. Initially the authors agreed on 14 out of 22 effect sizes (63.6%), but all disagreements in the coding were resolved by re-examining the articles. The study characteristics and effect sizes are represented in table 3.

(22)

Figure 1. Study selection and exclusion criteria. k = number of studies. Adapted from the PRISMA statement

(Moher et al., 2010). Definition of variables

Team collectivism

The cultural dimension of team collectivism was measured by various instruments (Alavi & McCormick, 2007; Donthu & Yoo, 1998; Dorfman & Howell, 1988; Earley, 1993; Erez & Earley, 1987; Hofstede, 1984; Jackson et al., 2006; Maznevski et al., 1997; Robert & Wasti, 2002; Singelis et al., 1995; Triandis & Gelfand, 1998; Triandis, 1995; Wagner, 1995; Wagner, Meyer, Humphrey, & Hollenbeck, 2005; Wagner & Moch, 1986). Many studies used different modifications or adaptations of Hofstede’s original cultural dimension ‘Individualism’. Nevertheless, all studies included in the quantitative meta-analysis regarded team collectivism as a continuum, where low scores indicate Individualism and high scores indicate Collectivism.

(23)

In addition, some studies differed between collectivistic tendencies at home or at work (Alavi & McCormick, 2007; Singelis et al., 1995; Triandis & Gelfand, 1998; Wagner et al., 2005). In these cases the work-related score was included in the analysis. The level of measurement was calculated similarly in all studies by aggregating the scores for individually taken surveys, which then provided a team score for Collectivism.

An exception to most studies was Dierdorff et al. (2011). This study measured psychological collectivism, which is the same concept as Collectivism, by means of five subcategories and did not report an aggregated score for Collectivism. One other exception was Katz (1999), who tested for both Allocentrism (Collectivism) and Idiocentrism (Individualism). If a study measured different versions of Individualism or Collectivism, these versions were combined in the overall analysis to guarantee the independence of effect sizes.

Team performance

Team performance can have multiple variations in the literature, defined as team productivity, team effectiveness, or even team reputation. These variations are also regarded as forms of team performance and deemed similar for this meta-analysis. A higher score for team performance, or any of its variations, indicated better performance on the team tasks. Of the studies included in the quantitative meta-analysis, team performance was measured in three different ways. Either by supervisor-rated scores on a group task, by member-rated scores on a group task or by an objective team task. Similar measures were used in some studies to examine team productivity and team effectiveness. As indicated above, these dependent variables are also regarded as forms of team performance and are included for this meta-analysis.

Two studies were slightly different. Tyran and Gibson (2008) used team reputation as the dependent variable, and argue in their text that team reputation is a direct indicator of team performance. Therefore, this study is included in the analysis. Wagner et al. (2005), used ‘speed’ and ‘accuracy’ as performance outcomes of a group task. Both outcomes have been included as well.

(24)

Moderator variables

Some moderator variables were coded to conduct additional analyses. Table 2 provides the averages and range for three of these moderator variables.

1. Percentage of females. Measured on a scale from 1-100. k = 12 studies reported this data. 2. Team Size. This variable was a scale measure of the mean number of team members per team. All studies reported this data.

3. Age. This variable was a scale measure of the average age of participants per study. k = 12 studies reported this data.

4. Continent. This variable was a dichotomous variable based on whether the country of origin was located in a Western society (United States, UK, and Europe) or an Eastern society (Asia-Pacific). West: k = 6 studies; East: k = 5 studies.

Table 2. Averages and range for moderator variables

Moderator Average Range k studies (% of 20)

% Female 64.6% 1 - 80 12 (60%)

Team Size 6.7 2.61 – 19.85 20 (100%)

Age 30.9 19 – 40.7 12 (60%)

Data analysis (meta-analysis)

The meta-analysis was conducted using a program called Comprehensive Meta-Analysis 2.0 (CMA; Biostat; USA). Meta-analyses are conducted using either a fixed-effects model or a random-effects model. The fixed-effects model assumes that for every study there is only one true effect size and all differences can be attributed to sampling error (Borenstein et al., 2009). The random-effects model actually assumes that differences in effect sizes can be attributed to differences in study characteristics such as age and education of participants or differences in research design and measures used. This meta-analysis makes use of the random-effects model to calculate the overall weighted effect size of the selected studies, since there are various measures used and study characteristics differ across studies. Additionally, the authors did not assume to have sampled all studies out of the population of studies (Borenstein et al., 2009).

(25)

The effect size is represented by means of the Pearson product moment correlation coefficient r. According to Cohen (1988), Pearson’s r can vary between -1 (perfect negative correlation) and 1 (perfect positive correlation). An effect size of 0 would mean no relation at all. Other effect sizes can be interpreted as: .1 small effect, .3 medium effect, and .5 or higher, large effect. The effect size r, the sample size N and the number of teams are reported in table 3 for all studies included in the analyses. Since effect sizes can vary between studies and larger effect sizes might bias the variance of effect sizes, the raw data will be converted to Fisher’s z-scores (DerSimonian & Laird, 1986). The program CMA will calculate these z-z-scores to conduct the subsequent analyses and convert Fisher’s z back to Pearson’s r upon completion.

Additionally, the differences or heterogeneity between effect sizes was computed using the Q statistic and an I2_{index. The function of the I}2_{index is to examine whether differences in}

effect sizes are due to real differences as opposed to chance differences. The I2_{index can be}

interpreted as low (25%), moderate (50%), or high (75%) heterogeneity because of real differences.

Publication bias analysis

Studies with highly significant results or effect sizes are more likely be published in journals and are therefore overrepresented (Borenstein et al., 2009). This might cause an overestimation of effect sizes (Begg & Mazumdar, 1994). Hence, publication bias is examined by using the following tests in CMA;

1. Duval and Tweedie’s trim and fill method. A rank-based augmentation technique to examine the sensitivity of the results to the selection mechanism. The method plots the standard error with Fisher’s z, draws a funnel plot and attempts to make the funnel plot of studies more symmetrical by imputing theoretically missing studies (Duval & Tweedie, 2000).

(26)

2. Begg & Mazumdar’s rank correlation. This correlation tests whether study effect sizes differ systematically. When the correlation is statistically significant, this is a cause for concern (Begg & Mazumdar, 1994).

3. Egger’s regression. Like Begg and Mazumdar’s rank correlation, this regression tests for systematic differences in study effect sizes. A cause for concern also occurs when this test is statistically significant (Egger, Smith, Schneider, & Minder, 1997).

Sensitivity and moderator analyses

Sensitivity analysis is used to measure how much influence one study has on the overall weighted effect size r. Therefore, a one-study removed analysis was conducted to investigate this influence. Additionally, a cumulative analysis was conducted, portraying the influence of one study added on the total weighted effect size. Moderator analyses have been conducted using the available data of percentage of female participants, average team size, average age of participants, and continent (either West or East). The moderator analyses test the systematic differences between studies on the overall effect size.

(27)

Table 3. Study characteristics and effect size data of studies included in the meta-analysis

Study Year r N IND-COL Measure Team

Performance Measure Country N Teams ATS % Female Mean Age

Chen et al. 2015 -0,30 216 Triandis and Gelfand (1998) TP-SR China 47 4,60 - 30,3

Colquitt & Jackson 2002 0,10 1747 Wagner (1995) TP-SR 88 19,85 - -

Dierdorff, Bell & Belohlav 2011 0,23* ₂₆₄ _{Jackson et al. (2006)} _OTT _{United States} ₆₆ _4,00 _- _-

2011 0,32 264 PC-Preference OTT United States 66 4,00 - -

2011 0,38 264 PC-Concern OTT United States 66 4,00 - -

2011 -0,09 264 PC-Reliance OTT United States 66 4,00 - -

2011 0,08 264 PC-Norm Acceptance OTT United States 66 4,00 - -

2011 0,41 264 PC-Goal Priority OTT United States 66 4,00 - -

Earley 1993 0,15 165 Erez and Earley (1987) TP-SR Multiple 20 8,25 - -

Eby & Dobbins 1997 0,13 148 Wagner & Moch (1986) PTA 33 4,48 40 19

Gibson - Study 1 1999 -0,06 294 Earley (1993) TP-MR Multiple 60 4,90 50 25

Gibson - Study 2 1999 -0,17 185 Earley (1993) TP-MR Multiple 71 2,61 1 35

Hiller, Day & Vance 2006 0,24 277 Dorfman & Howell (1988) TE-SR United States 52 5,33 - -

Katz 1999 0,15 156 Singelis (1995) Allocentrism TP-SR United States 16 9,75 72,4 28,1

1999 -0,13 156 Singelis (1995) Idiocentrism TP-SR United States 16 9,75 72,4 28,1

Kirkman & Shapiro 2005 0,18 34 Maznevski & DiStefano (1995) TPR-SR 34 1,00 59 33,1

(28)

Kirkman & Shapiro 2001 0,25 81 Maznevski & DiStefano (1995) TPR-SR Phillipines 25 3,24 38,3 35,7

2001 0,30 81 Maznevski & DiStefano (1995) TPR-MR Phillipines 25 3,24 38,3 35,7

Lin et al. 2015 0,29 476 Donthu and Yoo (1998) TP-SR Taiwan 120 3,97 - -

Liu et al. 2015 0,75 565 Robert and Wasti (2002) TP-SR Taiwan 121 4,67 - -

Mach & Baruch 2015 0,01 323 Alavi and McCormick (2007) OTT United Kingdom 73 4,42 48,3 23,1

Man & Lam 2003 0,19 2357 Erez & Earley (1987) TP-SR United States 381 6,19 75 32,2

Pillai & Meindl 1998 0,25 101 Wagner & Moch (1986) TP-MR 28 3,61 58,2 40,7

Schaubroeck, Lam & Cha 2007 0,14 1090 Erez & Earley (1987) TP-SR Multiple 218 5,00 - -

Small & Rentsch 2010 -0,02 280 Wagner (1995) TP-SR 60 4,67 - -

2010 -0,05 280 Wagner (1995) OTT 60 4,67 - -

Sosik & Jung – Study 1 2002 0,14 154 Hofstede (1984) OTT United States 43 3,58 - -

Sosik & Jung – Study 2 2002 0,17 148 Hofstede (1984) OTT Korea 40 3,70 - -

Tyran & Gibson 2008 0,20 428 Triandis (1995) TR United States 57 7,51 80 38,6

Wagner et al. 2005 -0,10 206 Wagner et al. (2005) PO-S 52 3,96 38,3 21,7

2005 0,06 206 Wagner et al. (2005) PO-A 52 3,96 38,3 21,7

Abbreviations: IND-COL = Individualism-Collectivism; ATS = Average Team Size; PC = Psychological Collectivism; TP-SR = Team Performance – Supervisor Rated; TP-MR = Team Performance – Member-Rated; OTT = Objective Team Task; PTA = Perceived Teamwork Ability; TE-SR = Team Effectiveness – Rated; TPR-SR = Team Productivity – Supervisor-Rated; TPR-MR = Team Productivity – Member-Supervisor-Rated; TR = Team Reputation; PO-S = Performance Outcome – Speed; PO-A = Performance Outcome – Accuracy

(29)

Chapter 4. Results

Study characteristics

For the meta-analysis, a total of 20 studies were included. Of these 20 studies, 22 independent effect sizes were reported. These 22 effect sizes were based on a total sample size N of 11,848. The total amount of teams (N teams) equalled 1757. Of the studies that reported locational data (k = 11), k = 6 studies were conducted in Western societies (United States, UK, and Europe) with a sample size of 3531 participants. Studies conducted in Eastern societies (k = 5 studies) contained a sample size of 1486 participants. Additional descriptive statistics are reported in table 4.

Table 4. Descriptive statistics for k = 20 studies in the quantitative meta-analysis.

Average k studies (% of 20)

Total N 11,848 20 (100%)

Total N Teams 1757 20 (100%)

Total N per Continent West 3531 6 (30%)

East 1486 5 (20%)

Relationship between team collectivism and team performance

The quantitative meta-analysis using the random-effects model found a statistically significant small to medium positive relationship between team collectivism and team performance; overall mean weighted r = .154, 95% confidence interval, CI [.06,.25], p = .001, k = 22. Among these 22 studies, there existed high variability of effect sizes (I2 _{= 96%, Q = 529.93, df = 21, p}

< .001). The findings therefore reject the null hypothesis and show support for the alternative hypothesis. The corresponding forest plot is displayed in figure 2.

(30)

Figure 2. Forest plot of the relationship between team collectivism and team performance

Note. ‘Correlation’ indicates the Pearson product moment correlation coefficient r. ‘Combined’ indicates the mean effect size in studies using multiple measures for team performance. ‘Blank’ means that there was only one measure of performance. ‘Total’ refers to the sample size in each study. The black diamond shows the overall weighted effect size r for k = 22 studies. Moreover, the automatic aggregation of subgroups or subcategories did not affect the overall mean weighted r. Without aggregation, the overall mean weighted r = .160, 95% confidence interval, CI [.08,.24], p < .001, k = 31. Among these 31 effect sizes, there also existed high variability of effect sizes (I2 _{= 95%, Q = 593.67, df = 30, p < .001).}

Publication bias analysis

The funnel plot in figure 3 indicates that there exists publication bias in the current meta-analysis. This is shown by the black filled circles on the right side of the mean. These filled circles represent imputed studies, or studies that are theoretically missing from the meta-analysis. There are no imputed studies on the left side of the mean. This is an indication of underestimating the measured overall mean weighted effect size (unfilled diamond), which thus could be larger than was currently estimated. This does not negatively impact the stated alternative hypothesis that team collectivism will positively relate to team performance. In fact, it reinforces support for this hypothesis, indicating an overall estimated effect size r = .271, Q = 961.46.

(31)

Figure 3. Funnel plot of the estimated variability and effect size r (expressed as Fisher’s z) for each study.

Note. The unfilled circles represent the actual studies used in the meta-analysis. The filled circles represent the imputed studies that were needed to make the funnel plot symmetrical, indicating publication bias to the right side of the mean. The newly overall estimated weighted mean (vertical line) indicates a slightly larger effect size (r = .271).

The asymmetry of the funnel plot is clearly shown at the top of the funnel plot, since there is a high concentration of studies on the left side of the solid vertical line (representing the overall mean weighted r with imputed studies). Studies at the top of the funnel plot have less variance than the studies on lower levels and weigh heavier on the mean weighted effect size. The concentration of unfilled circles (studies) are mostly positive, which corroborates with the result of the meta-analysis. The study effect sizes and their precision did not significantly differ due to systematic errors. Even though the Duval and Tweedie trim-and-fill method showed indications of publication bias, the interpretation of the newly estimated effect size did not differ significantly from the weighted effect size found in the meta-analysis. Additionally, Begg and Mazumdar’s rank correlation coefficient (p = .714) and Egger’s regression intercept (p = .648) were non-significant. Overall, it can be concluded that publication bias did not substantially influence the results of the meta-analysis.

(32)

Sensitivity analysis

The cumulative analysis indicated no significant change in the overall mean weighted effect size over time. Each study was added one at a time based on time point (1993 to 2015) and the effect size consistently indicated a positive small to medium sized effect (lowest estimate = .052, highest estimate = .167) as shown in figure 4.

Figure 4. Forest plot of the cumulative analysis

Note. ‘Combined’ indicates the mean effect size in studies using multiple measures for either team collectivism or team performance. ‘Blank’ means that there were no subgroups in that particular study. ‘Total’ refers to the total sample size, added up cumulatively. The plot shows the effect size when adding each individual study chronologically. The effect size is indicated as ‘Point’. The overall mean weighted effect size r of all k = 22 studies is depicted as the diamond on the bottom of the plot.

The one-study removed analysis also supports non-sensitive findings of the meta-analysis and indicates that no single study influenced the results to a large extent. By removing one study at a time the overall weighted effect size consistently kept indicating a significant positive small to medium effect (lowest estimate = .115, highest estimate = .175). The corresponding forest plot is represented in figure 5.

(33)

Figure 5. Forest plot of one study removed analysis

Note. ‘Combined’ indicates the mean effect size in studies using multiple measures for either team collectivism or team performance. ‘Blank’ means that there were no subgroups in that particular study. The plot shows the effect size of all studies combined except for the study in that row, indicating the effect that particular study if deleted. The effect size is indicated as ‘Point’. The overall mean weighted effect size r of all k = 22 studies is depicted as the diamond on the bottom of the plot. Subgroup analysis and meta-regression

The subgroup analysis in table 5 indicated a statistically significant positive small to medium weighted effect size of team collectivism with team performance when studies were conducted in a Western society (United States, UK, and Europe). Even though the estimated weighted effect size for studies conducted in Eastern societies was slightly larger, it was not significant. The heterogeneity statistics for both subgroups indicated a non-significant effect (Q = .244, p = .622). Therefore no support was found that location moderates the effect between team collectivism and team performance in the current study.

Table 5. Results of subgroup analysis

Subgroups k studies (% of 20) Mean weighted r (95% CI) ptwo-tailed

Continent West 6 (30%) .168 (.111,.225) .000

(34)

The output of the meta-regression analyses are reported in table 6. The only predictor that was statistically significant was the percentage of females in teams. According to the meta-regression results, having a higher percentage of females per team will positively moderate the relationship between team collectivism and team performance. This means that the relationship between team collectivism and team performance is stronger when the percentage of females per team is higher. The average age and average team size do not show any significant results and thus no support is found that either age or team size influence the relationship between team collectivism and team performance.

Table 6. Results of meta-regression analysis

Meta-regression predictor k studies (% of 20) R2 _Slope _{Slope p}_two-tailed

% Female 12 (60%) .63 .0056 .000

Age 12 (60%) .16 .0098 .163

(35)

Chapter 5. Discussion

In this quantitative meta-analysis the aim was to find stronger support for a positive relationship between team collectivism and team performance. The main findings of the current study reject the null hypothesis that there is no relationship between the two variables and support the alternative hypothesis. Based on 20 studies with 22 independent effect sizes, the meta-analysis found a statistically significant positive small to medium overall weighted effect size.

Moreover, a statistically significant positive small moderating effect was found concerning the percentage of females. A higher percentage of females per team positively moderates the relationship between team collectivism and team performance. Other additional analyses did not find significant effects. Neither location nor age or team size indicated significant moderating effects.

Theoretical implications

The findings of the meta-analysis resemble the findings of Bell (2007), who also concluded that the relationship between collectivism and team performance is slightly positive. Nevertheless, Bell (2007) goes in-depth concerning deep-level composition variables including personality and attitudes, and makes a distinction between lab and field results in the meta-analysis. Moreover, Bell (2007) does not define team collectivism in particular, but examines collectivism. To conclude, the current study is inherently different from her meta-analysis and contributes to the research gap concerning cultural predictors of team performance.

Interestingly, the findings contradict the meta-analysis by Taras et al. (2010), since a statistically significant negative small to medium weighted effect size was reported. However, their meta-analysis did not focus on this particular relationship since a large scale research design was used, spanning three decades and a multitude of organizational outcomes. The results of Taras et al. (2010) on team performance were also not discussed and not included in the body of the research. Moreover, the amount of studies used in their study was limited to

(36)

nine independent effect sizes, which is limited for meta-analytic purposes (Borenstein et al., 2009). Therefore it can be argued that the current study is a more accurate representation of the relationship of interest than the study of Taras et al. (2010) and has higher statistical power.

Concerning the additional analyses, it was logically expected that location (teams operating in either Western or Eastern societies) would have either a positive or negative moderating effect on the relationship between team collectivism and team performance. Even though no distinct hypothesis was formulated before conducting the literature search, one might expect teams located in Eastern societies to have, on average, higher levels of team collectivism due to their national culture and Western societies to have, on average, lower levels of team collectivism, since previous cultural measures have indicated Western societies to be more individualistic as opposed to Eastern societies (Hofstede, 2011; House et al., 2004).

Moreover, age could have had either a positive or a negative moderating effect on the relationship between team collectivism and team performance. On the one hand, teams with a higher average age could have been working longer with and in teams (team tenure) and could therefore facilitate team processes, which in turn could benefit team performance (Keller, 2006). On the other hand, teams with a higher average age could be perceived as having less flexibility than their younger peers, making it more likely that team processes will be hindered, thus affecting team performance negatively (Chiu, Chan, Snape, & Redman, 2001).

Possible explanations for these non-findings of the moderating variables location, age, and team size, might be that too few studies have been included, or too few studies have reported these moderator variables. The theoretical constructs might be of influence as well. For example, age is likely theoretically different from experience with teams and team work (team tenure), therefore not measuring the same theoretical construct. Since these moderator analyses were additional to the main analysis, they are not derivative of the statistically significant findings of the main research question.

(37)

The finding concerning the positive moderating effect of having a higher percentage of females per team has been unexpected. The result directly contradicts the meta-analysis of Schneid et al. (2015), who have come to the conclusion that gender diversity is detrimental to team performance. However, their meta-analysis has focused specifically on gender diversity in relation to team performance, which was not examined to that extent in this study, possibly explaining the discrepancy in outcomes.

There has been theoretical understanding of collectivism leading to better team performance because it makes sense at face value. Having a higher preference for working in teams, or having stronger values towards the common team goals compared to individual goals, logically leads to performing better in teams and on team tasks, which would imply better team performance. However, these results were ambiguous according to several studies (Chen et al., 2015; Gibson, 1999; Kirkman & Shapiro, 2005). Through the clear findings of this research it can be concluded that there is sufficient supporting empirical evidence for the alternative hypothesis that team collectivism is positively related to team performance.

The findings of the current study relate positively to the original problem as formulated in the introduction because the thesis set out to find stronger empirical evidence for a positive relationship between team collectivism and team performance. Reflecting on the initial research question to what extent team collectivism has an effect on team performance, it is clear that an adequate answer has been provided. This leads to the next section, where the practical implications of these findings are discussed.

Practical implications

Practical implications of the current study are directed at business executives and practitioners. Using the findings of this study they might keep cultural characteristics in mind when forming teams or when working with teams. The information can be used before teams are formed, while project team are already in play, or even after teams have disbanded altogether. Mainly,

(38)

cultural characteristics with regard to team collectivism could be an added factor with regard to influencing team performance and attempting to increase organizational results.

Firstly, business executives can influence the composition of project teams to make sure there is an adequate level of team collectivism. Lower levels of team collectivism would demand a change in the team composition. When problems arise with an ongoing project, team collectivism can be investigated besides regular quality management, to ensure that the team is cooperating in accordance with the company’s standards. In addition, team collectivism and values can be brought into the feedback loop and implemented in performance appraisals in order to evaluate the team performance outcome with team collectivism levels. This information can then be used and taken into account for future projects.

Secondly, corresponding with the outcome of the meta-regression concerning the percentage of females in a work team, business executives can make sure that there is a percentage of females in the team. The meta-regression has indicated that this positively moderates the team collectivism-team performance relationship, thus increasing this percentage might give team performance a small boost even when team collectivism levels were already adequate.

Lastly, the meta-regressions regarding age and team size did not show significant effects and the practical implications of these findings might be useful in the sense that neither age nor team size make a difference regarding the team collectivism-team performance relationship. Managers can thus rule out these factors when assessing team performance. It is however, recommended to use caution since other predictors like team cohesiveness can be influenced by for example team size.

(39)

Limitations and future research

Team collectivism is only one predictor of team performance. This does not mean that variance explained in the dependent variable of team performance is unique and only attributable to team collectivism. Team tenure, personality and motivation of team members might all theoretically cause variance in team performance as has been argued in research by Bell (2007) regarding surface-level and deep-level composition variables. Additionally, the team processes and interrelation between these processes might be too complex to measure on a univariate level, as has been done in this study. More intricate research designs are necessary to shed light on these interactions in future research.

A further limitation of the study is causality, which cannot be explained by this meta-analysis. In general, correlations only indicate relationships and do not imply causal direction (Sassower, 2017). Theoretically it can be argued that performing in teams leads to an increase in liking team work, which might over time change cultural values for collectivism slightly. This could then lead to higher levels of team collectivism. Causality can thus not be proven.

Moreover, the strength of meta-analyses comes from the higher sample size and large quantities of aggregated data. The use of Web of Science and the reference list of Dr. Taras alone does not cover the full scope of studies available, which is why a random-effects model was employed because it was assumed there are more studies examining the relation of interest in the entire population of studies. Perhaps broadening the scope of databases would have had an impact on the amount of studies included in the quantitative meta-analysis, subsequently increasing the statistical power of the meta-analytic design even more. However, this is not deemed to have influenced the outcome to a large extent. The current study has found an adequate amount of independent studies to conduct a meta-analysis.

(40)

Finally, the examined moderator variables were not present for all included studies, which might have influenced the results of the additional analyses. The study could be expanded to include additional moderator variables such as education or sector/industry in which the teams operate. Similarly, other cultural dimensions that have not been examined in this study might have an influence on team performance, which offers new ideas for potentially interesting future research as well.

New studies could benefit from a universal measure and definition of collectivism and team collectivism. Multiple measures have been used in this meta-analysis which were deemed to be commensurable but it would be more useful to have exactly similar measures. This can also be the case for the dependent variable, team performance, by either making an objective general measure or having the same specific team performance characteristics on which supervisors can base their assessment. One study criticising the current state of cultural measures urged future research to utilize cultural-level measurements, such as divorce rates, voting patterns, and culture-specific laws (Heine et al., 2002; Vandello & Cohen, 1999). These types of measurements would not be biased by aggregated individual responses, but according to them reflect national culture more appropriately.

Briefly mentioned before, team performance can be explained by a multitude of factors, and it is not clear whether these factors are interdependent as well. This study focused only on the direct relationship between team collectivism and team performance. The knowledge can be expanded by coming up with more complex models researching interdependencies of other predictors of team performance.

(41)

Conclusion

The aim of the current study was to provide stronger support for the alternative hypothesis that team collectivism has a positive relationship with team performance because the results of individual studies have been ambiguous. Through the means of a meta-analysis utilizing the random effects model, this study has actually found stronger support for this relationship. This study contributes to the body of knowledge on culture and cultural predictors with regard to team performance.

(42)

References

References marked with an asterisk (*) indicate studies included in the quantitative meta-analysis.

Alavi, S. B., & McCormick, J. (2007). Measurement of vertical and horizontal idiocentrism and allocentrism in small groups. Small Group Research, 38(4), 556-564. doi: https://doi.org/10.1177/1046496407304919

Allen, N. J., & Meyer, J. P. (1990). The measurement and antecedents of affective,

continuance and normative commitment to the organization. Journal of occupational

and organizational psychology, 63(1), 1-18. doi: 10.1111/j.2044-8325.1990.tb00506.x

Arnold, J., & Silvester, J. (2005). Work psychology: Understanding human behaviour in the workplace: Pearson Education.

Beal, D. J., Cohen, R. R., Burke, M. J., & McLendon, C. L. (2003). Cohesion and

performance in groups: a meta-analytic clarification of construct relations. Journal of

Applied Psychology, 88(6), 989-1004. doi:

http://dx.doi.org/10.1037/0021-9010.88.6.989

Begg, C. B., & Mazumdar, M. (1994). Operating characteristics of a rank correlation test for publication bias. Biometrics, 1088-1101. doi: 10.2307/2533446

Belbin, M. R. (2011). Management teams: Why they succeed or fail. Human Resource

Management International Digest, 19(3). doi:

https://doi.org/10.1108/hrmid.2011.04419cae.002

Bell, S. T. (2007). Deep-level composition variables as predictors of team performance: a meta-analysis. Journal of applied psychology, 92(3), 595. doi: 10.1037/0021-9010.92.3.595

Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2009). Introduction to Meta-Analysis: Chichester: John Wiley & Sons, Ltd.

(43)

Buller, P. F., & McEvoy, G. M. (2012). Strategy, human resource management and

performance: Sharpening line of sight. Human Resource Management Review, 22(1), 43-56. doi: https://doi.org/10.1016/j.hrmr.2011.11.002

*_{Chen, Y., Chen, Z. X., Zhong, L., Son, J., Zhang, X., & Liu, Z. (2015). Social exchange} spillover in leader–member relations: A multilevel model. Journal of organizational

Behavior, 36(5), 673-697. doi: 10.1002/job.2030

Chiu, W. C., Chan, A. W., Snape, E., & Redman, T. (2001). Age stereotypes and

discriminatory attitudes towards older workers: An East-West comparison. Human

relations, 54(5), 629-661. doi: https://doi.org/10.1177/0018726701545004

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (pp. 20-26). Hillsdale, NJ: Lawrence Earlbaum Associates.

Cohen, P. R., & Levesque, H. J. (1991). Teamwork. Nous, 25(4), 487-512. doi: 10.2307/2216075

*_{Colquitt, J. A., Noe, R. A., & Jackson, C. L. (2002). Justice in teams: Antecedents and} consequences of procedural justice climate. Personnel psychology, 55(1), 83-109. doi: 10.1111/j.1744-6570.2002.tb00104.x

Daft, R. L. (2015). Organization theory and design: Cengage learning.

De Dreu, C. K., & Weingart, L. R. (2003). Task versus relationship conflict, team performance, and team member satisfaction: a meta-analysis. Journal of Applied

Psychology, 88(4), 741-749. doi: http://dx.doi.org/10.1037/0021-9010.88.4.741

DerSimonian, R., & Laird, N. (1986). Meta-analysis in clinical trials. Controlled clinical

trials, 7(3), 177-188. doi: https://doi.org/10.1016/0197-2456(86)90046-2

*_{Dierdorff, E. C., Bell, S. T., & Belohlav, J. A. (2011). The power of “we”: Effects of} psychological collectivism on team performance over time. Journal of applied

The effect of team collectivism on team performance : a meta-analysis

MSc. In Business Administration – Leadership & Management Track