Leadership & Variance in Homophily

advertisement
Birds of a Feather
Sometimes
Flock Together
Team Leadership, Heterophily, and
Team Performance
Andrew Knight
University of Pennsylvania
Overview

Team diversity

Leadership, heterophily, and performance

Method, Analyses, & Results

Implications
Team Diversity
A Double-Edged Sword
4
3
2
1
0
Type 1 Type 2 Type 3 Type 4 Type 5
Diversity as Separation
Costs of diversity
Strongly
Agree

Slightly
Agree

Opposing views
Conflict
Diversity as separation
Neutral

12
10
8
6
4
2
0
Slightly
Disagree

5
Strongly
Disagree

Greater breadth of
resources
Creativity
Diversity as variety
6
Frequency


Diversity as Variety
Benefits of diversity
Frequency

Common Research Approaches:
Group & Individual Level

Link group-level metrics of diversity to group
processes



Group processes: Cohesion, conflict
Standard deviation as diversity metric
Link an individual’s average dissimilarity to
individual-level outcomes


Individual outcomes: Satisfaction, turnover
Euclidean distance as diversity metric
General Assumption:
Birds of a Feather Flock Together

Homophily often assumed…



…but seldom tested



Team members have positive perceptions of similar others and
negative perceptions of different others
Grounded in similarity-attraction theories
Very little team diversity research at the dyad level
Little exploration of actual relationships
Workplace may place boundary conditions around homophily


Roles and role structures
Leadership, power, social influence
Hypothesis 1:
Homophily varies across
teams
Team Leadership

Team leaders set the tone for a team




Shape team climates
Reinforce certain types of behaviors
Model appropriate behaviors
Leaders who take an inclusive approach
may model positive cross-category
relations
Hypothesis 2:
Leader inclusiveness is
positively related to
heterophily
Heterophily & Team Performance
Back to the Sword

Gaining the benefits of diversity



Positive cross-category relations yield access to
diverse information and resources
Diversity in KSAs aids in problem-solving,
creativity
Avoiding the costs of diversity


Positive relations aid in coordination
Positive relations mitigate the effects of conflict
Hypothesis 3:
Heterophily is positively
related to team
performance
Method:
Research Setting

Team-based military
competition

9-person teams navigate a
9km obstacle course

Teams train for nearly 4
months to prepare for the
one-day event
Method:
Sample

33 teams



Composed of cadets from the hosting academy
Training teams ranged from 10 to 16 members
381 individuals



86% male
79% White
Mean age = 20.3 (SD = 1.4)
Method:
Procedure
Month 1
Team
Roster
Confirmed
Month 2
Time 1
Survey
OPORD
Published
Month 3
Time 2
Survey
Time 3
Survey
Start of
Formal
Training
Month 4
Time 4
Survey
Competition
Method:
Predictor Variables (All at T1)

Team-level

Leader inclusiveness: 5-item scale completed by team leader



Controls: Prior competition experience, Athletic GPA, Military GPA
Individual-level




“Effective team leaders carefully weigh members’ opinions.”
Class (i.e., freshman, sophomore, junior, senior)
Gender
Branch choice (e.g., infantry, artillery, medical, intelligence)
Dyad-level

Same or different category membership for class, gender, branch
choice (0 = Same; 1 = Different)
Method:
Criterion Variables

Dyad-level criterion: Friendship (T3)



“How much did you socialize with X in your free time
during the past week?”
Members rated one another on a 5-point scale in a roundrobin fashion
Team-level criterion: Team performance (T5)


Team total score in the military competition
Scored by trained competition officials
Analyses:
The Social Relations Model via RCM

Random coefficient model




Random intercepts for team, actor, partner
Estimate A-P covariance and within-dyad covariance
Random slope for “homophily” effects
Fixed effects for category membership and diversity

Extracted homophily slope coefficients to test
team performance hypothesis

SAS PROC MIXED
Analyses:
Sample SAS PROC MIXED Code
proc mixed covtest data=t3srm;
class dyad GROUP actbranch partbranch;
model frd = actbranch partbranch difbranch leader
leader*difbranch
/ solution ddfm=SATTERTH;
random
a1-a16 p1-p16 intercept
difbranch
/ solution sub=group type=lin(5) ldata=g;
Fixed Effects
Cross-Level Int.
Random Intercepts
Random Slope
Within-Dyad Cov
repeated / type=cs sub=dyad(group);
ODS Output SolutionR = r_difbranch;
run;
Output Coeffs.
Results:
Null Variance Decomposition
% σ2
Team
1.2%
Actor
43.5%
Partner
6.5%
Dyad &
Residual
48.8%
Reciprocity

Generalized:
.32

Dyadic:
.61
Results: Class
Leadership & Variance in Homophily
Model 1
Model 2
Model 3
Actor Class
.02
.01
.01
Partner Class
.05*
.04
.04
-1.03**
-.96**
-1.64**
.03
-.04
Class Difference
Leader Inclusiveness
Class Difference X Leader Inclusiveness
.20*
-2 Log Likelihood
6288.4
5800.8
5797.8
AIC
6302.4
5814.8
5811.8
Results: Class
Leadership & Variance in Homophily
4
Dif
Class
Same
Class
Friendship
3.75
3.5
3.25
3
2.75
2.5
Minus 1 SD
Leader Inclusiveness
Plus 1 SD
Results: Gender
Leadership & Variance in Homophily
Model 1
B
Model 2
B
Model 3
B
Actor Gender
-.11
-.08
-.08
Partner Gender
-.20
-.20
-.21
-.38**
-.37**
-.60+
.11
.09
Gender Difference
Leader Inclusiveness
Gender Difference X Leader Inclusiveness
.07
-2 Log Likelihood
6590.1
6029.6
6032.0
AIC
6604.1
6043.6
6046.0
Results: Branch
Leadership & Variance in Homophily
Model 1
B
Model 2
B
Model 3
B
Actor Branch Choice
ns
ns
ns
Partner Branch Choice
**
**
**
-.09
.09
-2.41*
.09
.17
Branch Choice Difference
Leader Inclusiveness
Branch Difference X Leader Inclusiveness
2.27*
-2 Log Likelihood
6466.2
5966.0
5964.7
AIC
6480.2
5980.0
5978.7
Results: Branch Choice
Leadership & Variance in Homophily
3.5
Dif
Branch
Friendship
3.25
Same
Branch
3
2.75
2.5
Minus 1 SD
Plus 1 SD
Leader Inclusiveness
Results:
Heterophily and Team Performance
Model 1
β
Model 2
β
Team Military Ability
.01
.02
Team Athletic Ability
.19
.20
Team Experience
.41*
.38*
Branch Heterophily
-.20
Gender Heterophily
.05
Class Heterophily
.46**
F
3.41*
3.17*
R2
.26
.42
Discussion:
Summary of Results

Variance in homophily across teams



Leadership predicts heterophily



Supported for class, gender, branch choice
Models including random slopes were a better fit for the data
Supported for class and branch choice
Members of teams with inclusive leaders are more likely to form friends
with members of different classes and military branches
Heterophily predicts team performance


Supported only for class
Teams with heterophilous relationships with respect to class perform
better in the military competition
Discussion:
Implications of Results

Birds of a feather sometimes flock together



Leaders may shape relational patterns in teams



Key assumption may not hold in all work teams
Some teams are more heterophilous than others
Inclusive leaders model positive cross-category relations
Leadership as a lever for maximizing the benefits and
minimizing the costs of diversity
Diversity can help team performance if homophily is not
the rule


Key assumption of team diversity literature limits benefits
To benefit from diversity, teams may need heterophily
Discussion:
Broader Implications for Team Diversity

Examining diversity effects at the dyad level



A multilevel approach to studying diversity



A fine-grained look at diversity
Relationships are building blocks of team processes
Group composition research is inherently multilevel
Dyadic approach helps “unpack” variance
Develop and test comprehensive theories of
team composition with precision
Download