You are on page 1of 13

Group Diversity and Turnover Intention:

The Moderating Role of Social Cohesion

Nikodimos Takele Geberetsadik


Hakim Bwanika

Malmö University
Department of Urban Studies

Abstract
This study examines the role of social cohesion in moderating turnover intentions in diverse group
settings. Time spent socializing among group members was also hypothesized to increase the social
cohesion of diverse groups. The sample consisted of 21 student work groups (60 students) in two post
graduate programs at Malmö University in Sweden. The results support our hypothesis that social
cohesion moderates turnover intention in diverse groups predicting a nearly one-to-one relationship
between the variables. However the results only provided a partial support for our second hypothesis that
social cohesion in diverse groups is increased through time spent socializing among group members. The
results also show that the extent of the effect of the different factors of social cohesion on turnover
intention varies.

Introduction
Figures indicate that work force is increasingly becoming diverse. In 2007 more than15% of the
total labor force was foreign born in the USA (American community survey reports 2009).
Similarly in Sweden 2010 figures indicate that 12% of the total labor force is foreign born, the
figure goes up to 18% in cities like Stockholm and Malmo (SCB 2010). Increased diversity in the
labor force also implies increased diversity in organizations. Cross-functional project teams,
mergers, acquisitions, and joint ventures also add to the diversity in organizations (Knippenberg
et al. 2007). The effects of diversity are especially to be felt more in group works and group
based organizational structures where workers have to work closely together.

Different authors have examined group processes that take place in diverse groups. Research has
demonstrated the effects of diversity on different group level processes and individual level
attitudes and behaviors. Barnett, Caldwell and O'Reilly Ill (1989) identified a relationship
between group demography and turnover and found out that increased demographic diversity

1
increases turnover. Sargent and Sue-chan (2001) also demonstrated that racioethinic diversity
relates with higher group efficacy and for this effect to be enhanced by social cohesion.

In a meta-analysis Friedkin (2004) developed a perspective on social cohesion as a causally


interrelated aspect concerned with individuals’ membership attitudes and behaviors. Conflicting
theories also portray individual membership attitudes as antecedent and consequent factors of
social cohesion, but, we haven’t found any empirical research that analyzes their relationship.
Thus our study, pertaining to view of turnover as a consequent factor of social cohesion,
primarily seeks to investigate the role of social cohesion in moderating members’ turnover
intentions in diverse group settings.

Diversity and Turnover


Group diversity is conceived as the dispersion of group members’ positions on a given attribute
of diversity (Knippenberg et al. 2007). Attributes of diversity include age, gender, race,
organizational tenure, group tenure, education and functional background (Pelled 1996).
Harrison et al. (1998) further conceptualize diversity in to surface level – characteristics that are
manifested in physical features and deep level – obscure characteristics that are identified only
through extensive interaction. Surface level characteristics include elements such as
race/ethnicity, sex and age while deep level characteristics comprise of attitudes and values.
Different studies (Sargent et al. 2001, O'Reilly et al. 1989) have analyzed the effects diversity (a
group level attribute) on both individual and group level outcomes such as performance and
turnover and found conflicting results on the effects of diversity on performance but have found
consistent results on the effects of diversity on turnover.

Turnover is the termination of membership from a group, at the group level it is conceptualized
as turnover rate. At the attitudinal level it is also conceptualized as turnover intention – desire to
leave or remain in a group. Studies have consistently showed a positive relationship between
diversity and turnover. O'Reilly et al. (1989) studied 20 actual work units and found out that
group tenure increases turnover indirectly through decreasing social integration in the group. Tsui
et al. (1992) using a sample of 151 groups tested the effects of demographic diversity (measured
in terms of age, tenure, education, sex and race) on individuals physiological and behavioral
inclination to the organization and revealed increasing work-unit diversity to link with lower
levels of physiological attachment to group members. Similarly Jackson et al. (1991) observed
team heterogeneity (measured in terms of age, tenure, education level, curriculum, alma mater,
military service, and career experiences) to increase turnover. In a similar vein, the self-
categorization theory (Turner 1987) in psychology establishes that people classify themselves and
others using demographic characteristics such as age, race and gender. Effects of such
categorization processes would be that group process will be smoother when groups are
homogenous and individual membership attitudes would be more positive when members
resemble each other.

2
Similarly Pelled (1996) established a theoretical model that classifies each demographic diversity
variable according to its visibility1 and job-relatedness2. Pelled proposes that the more visible a
particular type of diversity the stronger its relation with emotional conflict and emotional conflict
to be a strong determinant of turnover. In other words surface level characteristics of diversity
indirectly increase turnover via increasing emotional (non-task) conflicts. This is also supported
by Jackson et al. (1991) finding that team heterogeneity is a strong predictor of team turnover
rate.

Taking advantage of the empirical consistency in the relationship between diversity and turnover
and the theoretical prepositions by Pelled (1996), we contend that diversity within groups
naturally induces turnover and we focus on a moderating variable – social cohesion. Here we
want to mention that usage of the word ‘diversity’ in our research mainly pertains to diversity in
ethnic background which is the major diversity attribute exhibited in the groups we sampled. The
need to measure diversity in our study is also minimized by the meager differences in diversity
that prevails across groups in our samples since the students groups we sampled have been
deliberately established to have equal and maximum levels of diversity; we will elaborate more
on this in subsequent sections.

A Moderating Variable – Social Cohesion

Cohesion is defined as the total field of forces which act on members to remain in the group.
These forces rely on that attractiveness of the members of the group, prestige of the group and/or
activities in which the group engages (Festinger 1950). Various meanings such as we-feeling,
group loyalty, harmony and coordination has also been attached to cohesion, the attractiveness of
the group as an end object to cohesion has proved vital (Libo 1953). Libo also identifies two units
of analysis – the individual member and the group and points that the cohesion construct is
captured differently in each unit of analysis.

Different authors portray cohesion as a multidimensional construct. Mullen and Copper (1995)
based on a meta-analysis of the cohesion literature identified three dimensions of cohesion –
interpersonal attraction, liking or commitment to the group task and group status or pride.
Braaten (1991), proposed a five factor model of cohesion which are attraction and bonding;
support and caring; listening and empathy; self-disclosure and feedback; and process
performance and goal attainment. The author identifies these factors as important in-group
dimensions of cohesion and argues that certain pre-group and early group preconditions, which
are given in table 2.1, also must be fulfilled for groups to establish and maintain cohesion.
Similarly Wilbmeyer et al. (1985) provide a two dimensional construct of cohesion – task
cohesion and social cohesion. Task cohesion is the degree of motivation towards achieving
group’s objectives while social cohesion is conceptualized as motivation to develop and maintain

3
social relationships within the group; at the individual member unit of analysis it is
conceptualized as individual team members’ feelings about personal involvement in the social
interaction of the group. Consequently Liking and commitment to the group task from Mullen
and Cooper and process performance and goal attainment from Braaten can be encapsulated
under task cohesion while interpersonal attraction from Mullen and Cooper and attraction and
bonding and support and caring from Braaten can be encapsulated under social cohesion.

Gross and Martin (1952) contend the definition of social cohesion should be constructed on
members attitude about the attractiveness of the group due to the case that such attitudes are
proximate causes of members’ decision to stay or leave from a group. Similarly Libo (1953)
establishes that individual members’ attraction to group induces strengthening membership. In a
meta-analysis of the social cohesion literature Fredriken (2004) ascertains that a group can be
labeled as cohesive when group conditions generate positive membership attitudes. Sargent et
al. (2001) postulate that if social cohesion is considered as an expression of
the psychological attachment of group members to the group then groups
that are endowed with diversity are likely to experience high turnover and
call for future research to address the moderating role of social cohesion in
diverse groups. Thus we posit the following hypothesis

Hypothesis 1: Social cohesion conceptualized as members’ psychological attachment to


the group, moderates turnover intention in diverse groups.

Table 1 Condition and Dimensions of Cohesion


Pre-group Early group conditions In-group dimensions
Conditions
Selections of suitable Resolving conflict Attraction and Bonding
Participants and rebellion

Balanced composition of Constructive norming and Culture Support and caring


the group building

Effective orientation, training Reducing Avoidance and Listening and empathy


and contracting Defensiveness
Self-disclosure and
feedback

Process performance and


goal attainment
Adopted from Braaten (1991)

4
Mckenna (1994) cited from Hayes (2002) posit that time spent together among team members to
provide the opportunity to explore common ideas and interests and for it to be an important factor
in bringing about cohesion. Thus, the second hypothesis to be examined in this study is as
follows:

Hypothesis 2: Time spent socializing among group members increases the social
cohesion.

Sample

This sample consisted of 60 Students enrolled in two postgraduate Urban Study programs;
Masters in Leadership for Sustainability (LFS) and Sustainable Urban Management (SUM) at
Malmo University. Malmo is a commercial centre of southern Sweden and an international city.
This is expressed not least, by the fact that Malmo has 286,500 residents from approximately 170
different nationalities of which 5% are students and 7% of which are international students
coming from America, Asia, Europe, Middle East, Australia and Africa. The enrollment statistics
in these two programs depict that the university is a representative of diversity in the city of
Malmo. i.e LFS comprised of 18%, 20%, 6%, 6%, 30%, 20% coming from Africa, Europe, North
& Central America, South America, Middle East and Asia respectively. SUM comprised of 12%,
24%, 14%, 12%, 20%, and 16% from Africa, Europe, North & Central America, South America,
Middle East and Asia respectively. The authors to this essay were given a go ahead to conduct
the study because there was a desire from the authors to understand how group process affect
individual member attitudes towards groups in the two Programs: (LFS) and (SUM). The
Participants in LFS and SUM had a Mean age of 27.8 and 28.6 respectively. They worked in
groups with a group size ranging from 4 to 8 members in LFS, where as in SUM all groups
comprised of 8 members.

In LFS the instructor advised that the groups were developed on a rationale that would assert
respect to group member’s diversity and individual members’ desires to belong to a particular
group with interests of accomplishing tasks within similar topic. For example one group
comprised of members that had interests of working with issues to do with social media and
communication. And for SUM, the instructor advised that diversity, education background and
demographic attributes were put into consideration while the groups were being formed. In these
groups education backgrounds and demographic aspects were evenly distributed among groups,
as much as possible, by the instructor.

Task
With broad content guidelines provided in groups, students were required to accomplish various
tasks or assignments from different study program module. The groups would then structure the
assignment and assign roles and subtasks to group members. For instance a) In LFS the students

5
would select a topic of their choice within a specific module. Select distinct articles or scientific
materials that are related to the problem question. Devise plans and structure the assignment,
allocate roles and subtasks to group members. The accomplishment of these tasks was a
collective responsibility, whereby every member of the group was to write a one page synthesis
to an article chosen and contribute to a one page group synthesis that would be submitted to the
instructors.
b) In SUM the nature of the assignment required students to collect field data, analyze, report and
present the finds upon being given a problem question within a module. The members of each
group as well had to accomplish the tasks in an interdependent manner however in this case each
group had to submit 5-6 page report and a power point presentation.

Procedure
Initially email traffic was exchanged between the researchers and instructors on seeking advice
and information of how the groups in both programs were developed. A questionnaire assessing
demography, social cohesion and turnover intention were distributed to the students after they
had completed all their group assignments. For LFS program, an online survey was posted, for
the students to respond. And for Sustainable urban management the questionnaires were
distributed during the last course lecture, the students were assured that participation was
voluntary and their answers would remain confidential to the two researchers.

Measures

Social cohesion: To asses social cohesion we adopted a scale from Sargent et al. (2001)
which was specifically developed to asses critical factors of social cohesion in student groups.
The scales consisted of four items (I am friends with members of my group, I feel a sense of
belongingness to my group, I get along with members of my group and I like my group)
assessing members emotional attitudes towards their groups. Each item had a 5-point
response scale ranging from strongly disagree to strongly agree.

Turnover intention: To asses turnover intention we developed an item that assesses individual
member’s attitude towards strengthening or weakening membership. “I would like to
continue with the same group for the second semester” was the item developed with a 5-point
response scale ranging from strongly agree to strongly disagree.

Socializing: To assess whether time spent among group members increases social cohesion,
we developed an item (me and my group spend time together in non-academic activities) with
a 5-point response scale ranging from strongly disagree to strongly agree.

6
Consistent with Sargent et al. (2001) we aggregated each of the items under social cohesion
in order to have a single measure of social cohesion for each respondent. We then specified
an ordinary least squares regression model to test our hypotheses.
Y= β0 + β1x+ u (hypothesis 1) Where y is turnover intention and x is social cohesion and,
Y= β0 + β1x + u (hypothesis 2) where y is social cohesion and x is socializing

Following our discussions with the program administrators for the two programs, SUM and
LFS, we decided to regress the data from the two programs separately since our discussions
with the administrators revealed that important pre-group conditions differed for the groups in
the two programs and aggregating the data from the two programs would bias our results.

Additionally a separate regression was also conducted holding turnover intention as the
dependent variable (y) and each of the items in social cohesion as independent variables in
order to identify which social cohesion factor was dominant in determining turnover
intention. The regression model we specified is as follows:

Y= β0 + β1x1 + β2x2 + β3x3 + β4x4 + u; where x1 represents friendship with fellow members, x2
represents feeling a sense of belongingness to the group, x3 represents getting along with
group members and x4 represents individuals affective feeling towards their group.

To conduct the regression, we assigned numerical values to the Likert scales i.e. 1,2,3,4 and 5
representing strongly disagree, disagree, neutral, agree and strongly agree respectively. For
turnover intention since we used an affirmative item which measures desire to stay to serve as
a surrogate for turnover intention which measures desire to leave thus we assigned the
numerical values in a reverse order i.e. 5,4,3,2, and 1 for strongly disagree, disagree, neutral,
agree and strongly agree respectively. Consistent with Sargent et al. (2001) we treat the scales
as continuous variables.

Results
Using Excel we ran the regression on our first hypothesis” Social cohesion decreases
members’ turnover intention” for each of our data sets, SUM and LFS. First we run the
regression of turnover intention (y) on social cohesion (x) (taking the mean of all the items
in the social cohesion) for both data sets. The coefficient of social cohesion (β 1 = -.91, P <
0.01) for SUM data set supports the hypothesis and predicts that a one level increase in
social cohesion (e.g. from 2 to 3) decrease turnover intention by .91. The coefficient of
social cohesion predicts a nearly one-to-one relation between social cohesion and turnover
intention indicating that a one level increase in social cohesion almost decreases turnover
intention by the same amount. The P-value less than 0.01 indicate that the data supports our
hypothesis even at 1% significance level.

7
Table 2: Regression results of y on x for SUM data set for the first hypothesis
Coefficients Standard Error T stat P-Value

x -.91 .14 6.44 <.01

N = 32 R2 = 58%

Likewise for LFS data set, the coefficient of social cohesion (β 1 = -1.25, P < 0.01) supports
our hypothesis and predicts that a one level increase in social cohesion decreases turnover
intention by more than one. The P-value less than 0.01 indicate that the data supports our
hypothesis at 1% significance level.

Table 2: Regression results of y on x for LFS data set for the first hypothesis

Coefficients Standard Error T stat P-Value

x -1.25 .28 4.55 <.01

N = 24 R2 = 48

The regression of turnover intention (y) on each of the factors of social cohesion for SUM
data set reveals interesting results. The coefficient of individual members’ affective feeling
towards their group (x4) estimates a nearly one-to-one negative relationship between
individual members affective feeling towards their group and turnover intention at 1%
significance level. The coefficient of feeling of sense of belongingness to the group (x2)
estimates a minor but positive relationship between turnover intention and sense of
belongingness with a value of .22 at 10% significance level. Friendship with fellow

8
members (x1) and getting along with group members (x3) are predicted to have no
relationship with turnover intention, both with p-values greater than .1.

Table 4: Regression results for y on x1, x2, x3 and x4 for SUM data set for the first hypothesis
Coefficients Standard T stat P-Value
Error

X2 .22 .21 -1.07 <.1

X4 -.93 .16 5.67 <.01

N = 32 R2 = 77%

For LFS data set the regression of turnover intention (y) on friendship with fellow members
(x1) and sense of belongingness to the group (x2) predict no correlation between y and x1 and
x2 with p-values greater than .1. For the variables getting along with group members (x3) and
feeling to like the group (x4), a negative relationship is estimated with coefficient’s .57 and .
41 respectively both at 10% significance level.

Table 4: Regression results for y on x1, x2, x3 and x4 for LFS data set for the first
hypothesis

Coefficients Standard T stat P-Value


Error

X3 -.57 .29 1.96 <.1

X4 -.41 .22 1.91 <.1

9
N = 24 R2 = %

To test our second hypothesis we ran the regression of social cohesion (y), taking the mean of
all the items in the social cohesion, on time spent together socializing (x) for both our data
sets SUM and LFS. The coefficient of x (β1 = .63, P < 0.01) for SUM data set supports our
hypothesis and predicts that one level increase in agreement to the item “me and my group
spend time together in non-academic activities” in the questionnaire increases social cohesion
by 63%.

Table 5: Regression of y on x for SUM data set for the second hypothesis
Coefficients Standard T stat P-Value
Error

x .64 0.10 6.23 <.01

N = 32 R2 = 56%

For LFS the results do not support our hypothesis with p-values greater than .1 and an R2
close to zero, predicting that time spent together socializing (x) does not add to the social
cohesion (y) in groups in LFS.

Table 6: Regression of y on x for LFS data set for the second hypothesis
Coefficients Standard T stat P-Value
Error

x .001 .096 .01 >.1

N = 24 R2 = -4 %

Discussion
The main purpose of the present study was to investigate the relationship between social
cohesion and turnover intention in group and to investigate whether time spent in social
activities increases social cohesion in groups. Fortunately, research on social cohesion and its
relationship to individual turnover has been desired (Sargent et al. 2001)

10
The findings of our study provide evidence for the hypothesis that social cohesion decreases
turnover intention. The aggregated measure of the effect of social cohesion on turnover yields
an approximately one-to-one negative relationship between social cohesion and turnover
intention for both our samples from SUM and LFS. However a disaggregated measure of the
effect of social cohesion reveals that individuals affective feeling towards the group to be the
most dominant social cohesion factor determining turnover intention for SUM and friendship
with fellow members and getting along with group members are predicted to have no effect
on turnover intention. For LFS a moderate positive relationship was estimated between
getting along with members of the group and affective feeling towards the group and
individual’s decision to remain in a group, while no correlation was found to exist between
friendship with group members and feeling of sense of belongingness and turnover intention.

The study contributes to the understanding of the effects of each of the factors of social cohesion
in moderating turnover intention. Unlike Whiteoak (2007) which estimated no correlation to exist
between cohesion (including task and social cohesion) and turnover; our findings suggest the
reverse while treating social cohesion separately. This provides support for the multidimensional
view of cohesion (Friedkin 2004 and Mullen and Copper 1995) to better understand the effects of
cohesion on group processes. Furthermore the differences in results that prevailed between SUM
and LFS when we disaggregated social cohesion indicate that the effects of social cohesion on
group processes are better understood when different factors of social cohesion are analyzed
separately. We regard this as the most important contribution of our study to the literature of
social cohesion.

On the other hand our second hypothesis is partially supported by our findings. In SUM our
findings predict that time spent on social activities off academic work among group members is
positively related with social cohesion. However in LFS our findings predict that social cohesion
in groups is not the result of time spent on social activities among group members. These results
could be attributed to the fact that important pre-group conditions vary within the two programs.
In LFS students were allowed to freely form groups based on their topic of interest which might
have contributed to decreased task conflicts (see Jehn et al. 2001) and thus helped members to
easily get along. However this is only a speculation and further research is warranted to assess
this relationship. The freedom that was given for students in LFS to freely form groups has also
given some room for students to join and associate with groups that were characterized with
members of similar ethnic backgrounds as their own. Analogous to the similarity – attraction
perspective by turner (1987), this might have contributed to attractiveness of the group and thus
increased social cohesion. A Model that incorporates diversity would help better understand this
relationship which was a limitation of the current study.

Practically the findings could mean that group processes will be smoother when members are
homogenous in terms of interests and ethnic background but its implications should be treated
with caution. Previous research has shown a positive correlation between task conflict and

11
performance (Chiocchio et al. 2009) while both positive and negative relationships have been
posited between performance and diversity. The implications of these for the manager in the
work place and the instructor in schools is to identify and maintain a balanced composition (in
terms of interest and diversity) in a group that could increase performance and minimize turnover
intention.

Acknowledgement
We thank Fredrik Björk and Karin Grundström for their invaluable support. We also want to
thank all the students who participated in our survey.

References

1. Braaten (1991) Group Cohesion: A new dimension model: Group, 15, 1 Brunner/Mazel
Inc /39.

2. Chiochio, F. y Essiembre, H. (2009). Cohesion and performance: A meta-analytic review


of disparities between project teams, production teams, and service teams. Small Group
Research, 40, 382-420

3. Harrison DA, Price KH, Bell MP. 1998. Beyond relational demography: time and the
effects of surface- and deep-level diversity on work group cohesion. Acad. Manage.
J. 41:96–107.

4. Festinger, L. (1950). Informal social communication. Psychological Review, 57, 271-282.

5. Friedkin N.E( 2004) Social Cohesion, Annual Reviews Sociology , 30: 409-25.

6. Gross, N. & Martin W.E (1952). On group cohesiveness. American journal of Sociology,
57, 546-554.

7. Jackson, S. E., Brett, J. F., Sessa, V. I., Cooper, D. M., Julin, J. A., & Peyronnin, K.
(1991). Some differences make a difference: Individual dissimilarity and group
heterogeneity as correlates of recruitment, promotions, and turnover. Journal of Applied
Psychology, 76, 675-689.

8. Jehn, K. A. & Mannix E. A (2001) The Dynamic Nature of Conflict: A Longitudinal


Study of Intragroup Conflict and Group Performance The Academy of Management
JournalVol. 44, No. 2 pp. 238-25

9. Knippenberg D. V. & Schippers, M. C. Work Group Diversity, Annual Review of


Psychology, Vol. 58: 515-541(Volume publication date January 2007) First published
online as a Review in Advance on August 11, 2006

12
10. Libo, L.M (1953) Measuring group cohesiveness. Ann Arbor University of Michigan
Research center for group Dynamics.

11. McKenna H.P. (1994) The Delphi technique: a worthwhile approach for nursing? Journal
of Advanced Nursing 19, 1221± 1225

12. O'Reilly CA, Caldwell DF, Barnett WP. 1989. Work group demography, social
integration, and turnover. Admin. Sci. Q.34:21–37

13. Pelled, Lisa Hope. 1996. "Demographic Diversity, Conflict, and Work Group Outcomes:
An Intervening Process Theory."

14. Sargent L.D and Sue-Chan C.( 2001), Does Diversity Affect Group Efficacy? : The
Intervening Role of Cohesion and Task Interdependence. Small Group Research 32: 426.

15. Tsui AS, Egan TD, O'Reilly CA. 1992. Being different: relational demography and
organizational attachment. Admin. Sci. Q.37:549–79

16. Turner JC, Hogg MA, Oakes PJ, Reicher SD, Wetherell MS. 1987.Rediscovering the
Social Group: A Self-Categorization Theory.Oxford, UK: Blackwell .

17. Widmeyer,W. N., Brawley, L. R.,&Carron, A. V. (1985). The measurements of cohesion


in sport teams: The group environment questionnaire. London, Ontario, Canada: Sports
Dynamics.

13

You might also like