Group process losses

advertisement
Introduction to Groups:
Process losses
Agenda



1.
Groups are valuable.
a.
2.
Groups often do better than the individual in them. They allow people to handle
projects that are too large or complex for a single individual.
The success of a group consists of three components:
a.
b.
c.
Getting the work done
Supporting the needs of individual members
Keeping the group as an unit functioning.
3.
Much groups research adopts a functionalist perspective, trying to identify
inputs & process that help groups succeed.
4.
Groups often perform worse than optimal. Afflicted by "process losses", which
prevent them from doing as well as they are capable of doing:
a.
b.
Problems in coordination
a.
Brainstorming
b.
Shared information bias
Problems in motivation
a.
Social loafing
b.
Group think
Criteria for Group Success
What were the criteria for success in the rowing crews?
1.
The success of a group consists of three components:
a. Production: Getting the work done & meeting
needs of stakeholders
b. Member support: Supporting the needs of
individual members
c. Group maintenance: Keeping the group as an
functioning unit and developing it with time and
experience.
2.
These components can be in tension
Functionalist Perspective
•
Normative approach that seeks to identify the inputs
to groups and the group processes cause groups to
be more or less successful.
– Groups are goal oriented
– Both group behavior & performance can be evaluated
– One can control group interactions to make them more
appropriate for achieving group goals
– Other factors (both internal & external) influence group
performance through group interaction
•
“Normative” means that there are better or worse
ways to organize groups to achieve the goals for
which they were formed.
Traditional Input-Process-Outcome Model
of Group Effectiveness
Input
Process
Output
Forsyth, D. (2010). Group dynamics (5th ed.). Belmont, CA:
Wadsworth Pub Co.
The tension in group work
•
•
Groups are valuable: Perform better than the
individuals who comprise them
But they rarely live up to their potential = Process
losses.
Knowledge
Skills
Attitudes
Time
Effort
Actual
Potential
Performance
Process losses
Coordination
•
•
•
•
Production blocking: members can
not think of new ideas while
listening to someone else
Common knowledge effect:
discussions focus on shared
information
Unequal participation:
participation  expertise
Coordination costs of
– Scheduling
– Developing consensus
– Doing the work
Motivational
•
•
•
•
•
Social loafing: members expend less
effort
Conformity pressures & group
think: members feel pressured to
agree with other group members.
Effects strongest with cohesive
groups.
Conflict: interpersonal conflict is
disruptive
In-group vs. Out-group bias: Mere
group membership leads to in-group
favoritism.
Escalation of commitment: groups
persist in following a course of
action despite evidence against it
Coordination process loss:
Brainstorming in interactive groups
Alex Osborne’s
Rules for Brainstorming (1953)
•
•
•
No criticism
Defer criticism.
Encourage the wild
– Wild ideas may trigger more practical suggestions
from others
– It is easier to tone done crazy ideas than to be creative
•
The more the better
– The more ideas, the greater likelihood of one winner
– It is easier to eliminate than to generate
•
Build off of others
– Combinations and improvements are welcome
– How can you improve what others offered?
– Can you get creativity from combinations?
Osborne, A. F. (1953). Applied imagination: Principles and procedures of
creative problem solving. Charles Scribener’s Sons, New York.
Process Loss in Brainstorming
Real, interacting groups
(versus nominal ones)
produce






Fewer ideas
Fewer good ideas
Lower average quality
Lower feasibility
Lit review: 18/22 studies
show nominal groups
surpass real groups (Diehl &
Stoebe, 1987)
Fixes depend on causes
120
100
Number of ideas

80
60
40
20
0
Real group
Nominal group
Number of ”good” ideas produced by interacting
and nominal 4-person groups discussing
how to improve relationships among
Germans & guest workers (Diehl & Stoebe, 1987)
Possible explanations
Explanation
Solution
Possible explanations
•
•
•
Explanation
Conformity pressures
Social loafing
Production blocking
Solution
Anonymity
Surveillance systems
Simultaneous input
Disentangling causes
•
Diehl & Stoebe (1989): 5 experiments to identify
importance of causes
– Evaluation apprehension: High (Your ideas on controversial topic
recorded & judged) vs Low (no recording & judgment)
– Social loafing: Personal (each person compared) vs Collective (group as
a whole is compared) to a standard
– Production blocking: High (Stoplights prevented subjects from producing
ideas when another subject was producing) or Low (no lights)
•
Production Blocking was the main problem
–  Brainstorm at home & use group meeting to consolidate
•
Other techniques to enhance brainstorming
– Take a break
– Brainstorm within categories
– Division of labor
Coordination process loss:
Common knowledge effects
Hidden profiles & shared information bias
•
•
One reason groups succeed is that together members
have more knowledge than any single member
Yet groups
– Over-discuss information held by all members
– Under-discuss information held by a subset of members
•
Often leads to worse decision-making than if group
shared all their information
Sample Coordination Problem:
Lack of Information Sharing
•
•
•
Team members have some shared & unshared information about a
candidate – all positive
If they use all the information, the choice is clear. Pick candidate
with most positive attributes
But if they share only some of the information, choice may be wrong,
depending on what is shared
Information sharing determines
quality of group decision-making
•
If they combine all the information, A dominates B
•
But partial sharing can lead to wrong decision
Stasser, G., & Titus, W. (1985). Pooling of unshared information in group decision
making. Journal of Personality and Social Psychology, 48optional, 1467-1478.
Shared information
•
•
•
•
More likely to be mentioned (d=2.03, k=33)
Will be discussed more
More likely to be remembered
More influential in decision-making
 By not talking enough about information held only by
a subset of members, group is not taking advantage
of one of a group’s primary asset
Lu, L., Yuan, Y. C., & McLeod, P. L. (2012). Twenty-Five Years of Hidden
Profiles in Group Decision Making A Meta-Analysis. Personality and Social
Psychology Review, 16(1), 54-75.
Failure to discuss all the information
leads to worse decisions
•
Odds of a correct answer were 8x larger when all
group members had all the information than when
only a subset of members had some information
Moderators
•
Hidden profiles led to
less info sharing when:
– Groups were larger
– When there was more
information overall
– When more of the initial
information was unique
•
No effects:
– Communication media
•
Hidden profiles led to
worse decisions when:
– Groups were larger
– When there was more
information overall
How to fix the problem
•
Helps
Group interaction:
– Explicitly ask for unshared info
– Have recognized
specialization (i.e. roles)
– Build group trust
•
Structure the decision
– Consider alternative one at a
time
– Rank, not choose
– Suspend initial judgments
– Approach task as “problem to
be solved” not “judgment”
•
•
•
•
Doesn’t help
Increase discussion
Separate review & decision
stages
Increase team size
Poll before discussion
Coordination reflected in
participation rates
•
•
Uneven
distribution in
groups
Unevenness
increases with
group size
Example of coordination loss in
Wikipedia
Generally articles with more editors have higher
quality
N Editors (log2)
Article quality X Number of Editors
11.0
10.0
9.0
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
1
Quality
Stub
2
Start
3
C
4
5
B
6
Good
Featured
Wikipedia article quality
But coordinating large numbers of editors could be a
Coordination types
•
Explicit coordination
– Direct communication among editors planning and
discussing article
 More process loss
•
Implicit coordination
– Division of labor and workgroup structure
– Concentrating work in core group of editors
– Development of group norms
 Less process loss
Predicting changes in Wikipedia quality
Longitudinal Analysis: What leads to
changes in quality
Predicting Changes in Article Quality
Intercept
Initial Quality
Article Age
# Editors
Editor Concentration
Editors X Concentration
Quality X Concentration
Age X Concentration
# Talk Edits
Editors X Talk
Quality X Talk
Age X Talk
Coef.
.304
-.146
-.006
-.020
.600
.216
-.222
-.041
.087
-.010
-.001
-.003
SE
.033
.005
.004
.003
.038
.020
.035
.028
.004
.002
.003
.003
P
***
***
***
***
***
***
***
***
More Talk & Concentration
Improve Article Quality
•
Effect of number of editors disappear, when
examining change in quality at average levels of
talk & concentration
Longitudinal Analysis: What leads to
changes in quality
Predicting Changes in Article Quality
Intercept
Initial Quality
Article Age
# Editors
Editor Concentration
Editors X Concentration
Quality X Concentration
Age X Concentration
# Talk Edits
Editors X Talk
Quality X Talk
Age X Talk
Coef.
.304
-.146
-.006
-.020
.600
.216
-.222
-.041
.087
-.010
-.001
-.003
SE
.033
.005
.004
.003
.038
.020
.035
.028
.004
.002
.003
.003
P
***
***
***
***
***
***
***
***
Effects of communication depends
on number of editors
Direct communication is effective with small
number of editors, but harmful with many editors
Predicted change in quality
.25
.3
.35
.4
.45
By number of editors and amount of communication
0
2
4
6
Number of editors (log2)
High editor communication
8
Low editor communication
Longitudinal Analysis: What leads to
changes in quality
Predicting Changes in Article Quality
Intercept
Initial Quality
Article Age
# Editors
Editor Concentration
Editors X Concentration
Quality X Concentration
Age X Concentration
# Talk Edits
Editors X Talk
Quality X Talk
Age X Talk
Coef.
.304
-.146
-.006
-.020
.600
.216
-.222
-.041
.087
-.010
-.001
-.003
SE
.033
.005
.004
.003
.038
.020
.035
.028
.004
.002
.003
.003
P
***
***
***
***
***
***
***
***
Effects of Number of Editors
Depends on Concentration
•
Concentration helps overall
– Helps most with when many editors contribute
– Many editors without concentration harms quality
Predicted change in quality
.2
.25
.3
.35
.4
.45
By number of editors and editor concentration
0
2
4
6
Number of editors (log2)
High editor concentration
8
Low editor concentration
Motivational process loss:
Social Loafing
Social Loafing
RINGLEMANN’s Discovery (1913)
•
•
•
•
A French agricultural engineer who conducted most of
his research in late 1880’s.
Device measured the exact mount of forced exerted
on the rope
1, 2, 3, or 8 people pulling on rope
Force didn’t increase
linearly with the
number of people
Social Loafing: RINGLEMANN
(1913)
•
Mean force pulled by
individuals = 85.3 kg of force
•
Eight people should produce
(8*85.3kg) or 682.4 kg of force,
but really produce less than
half
` Why?
Nominal or co-acting
groups. Subjects think
they are in the
presence of a group,
but in fact acting alone
motivation
Real (or collective)
groups need to shout at
the same time
coordination
Sound pressure (Dunes per cm2)
Distinguishing Coordination
Problems from Motivation
10
Nominal group
(coactive)
9
8
7
6
5
4
Real group
(collective)
3
1
2
Group size
6
Social Loafing:
Working in a group decreases effort
•
•
Social loafing occurs in
both interacting and
nominal groups
Across many
performance outcomes
–
–
–
–
Physical
Intellectual
Quantity
Quality
When is social loafing
reduced?
Brainstorm uses of a knife.
Place ideas into
– Separate boxes (coactive)
– Common box (collective)
Group cohesion
– High Cohesion: Prior pleasant
interaction
– Control: No conversation
– Low Cohesion: Prior
argumentative interaction.
Numberof
ofunique
uniqueideas
ideas
Number
Loafing reduced in cohesive groups
40 Coactive group
Coactive group
35
30
25
20
15
10
Collective group
Low
Social loafing occurs
– In no history control group
– In low cohesion group
– Eliminated in high cohesion
group
Control
High
Cohesion
Karau, S. J., & Hart, J. W. (1998). Group Cohesiveness and Social Loafing: Effects of a Social Interaction Manipulation
on Individual Motivation Within Groups. Group Dynamics: Theory, Research, and Practice, 2(3), 185-191.
People even work harder in a cohesive group
when they think teammate has low ability
• Subject performs a brainstorming task
• Vary individual vs. collective work
•
Individual: Put ideas in separate boxes
Collective: Put ideas in common box
• Vary group cohesion
•
Friends vs strangers
• Vary perceived ability of others in groups
•
•
Low: “I’m lousy at this type of task”
High: Irrelevant comments or “I’m generally
good at this type of task”
High cohesion
40
Number of Ideas
•
Low cohesion
30
20
10
0
Individual
Collective
Low ability coworker
Individual
High ability coworker
• Social loafing results:
•
•
Collective
With low-ability partners, social loafing occurs in
non-cohesive groups, but reduced in cohesive
groups
With high-ability partners, social compensation
occurs in cohesive groups, but not in noncohesive groups
Karau, S. J., & Williams, K. D. (1997). The effects of group cohesiveness
on social loafing and social compensation. Group Dynamics, 1(2), 156-168.
Karau & Williams Meta-Analysis
Meta-Analysis
– Way to systematically combine evidence from many studies by
averaging effect sizes
– Effect size = power of variable of interest in standard deviations
units
Karau & Williams result
– 163 effect sizes
– 123 of 163 studies show evidence of social loafing
• People working harder in coactive conditions than collective
conditions
– Mean effect size = .44 standard deviations (moderate)
Analyze average effect size & test for
heterogeneity
Illustrating Average Effect Size
•
Difference of .44 standard deviation units btw effort
when individuals are working independently (coacting) versus pooling output (collective)
– Small to moderate effect size
– 66% of people in collective group would exert less effort than averager person in
the co-acting group
– Comparable to difference in height between 14 year old & 17 year old girl or the
difference in reading or math tests of 4th graders vs 5th graders or reading
differences between 12th grade girls vs boys
.44d
Collective
Co-acting
Test for moderator variables
Factors that mitigate social loafing
Social loafing reduced if
• Individual's output is visible
• Task is attractive
• Group is attractive
• Expect others to perform poorly
• Own contribution is unique
• Task is simple
• Task has specific, challenging goals
• Among women
• Among people from collectivist cultures
Karau, S. J., & Williams, K. D. (1993). Social loafing: A meta-analytic review and theoretical
integration. Journal of Personality & Social Psychology, 65(4), 681-706.
Stylized facts vs causal theory
Expectancy-Value Theory (Vroom)
Individuals will work hard in groups to
the extent they believe:
– effort will lead to better performance
– better performance will be recognized
and rewarded
– the rewards are valuable
Utility model of individual motivation
Quasi-economic model
Individuals work hard to the extent that doing so increases
personal payoffs
individual
performance
individual
effort
individual
outcome
x
Valance of
outcome
individual
utility
You study hard (effort) 
–
–
–
–
If you enjoy the topic (intrinsic motivation)
You have a test (individual performance) 
You ace the test (individual outcome) 
You are proud & get praise from parents (evaluation of outcome)
motivation
Collective Effort Model (Karau & Williams)
•
Being in a group
– Changes probability of group performance
– Changes probability of individual outcome
– Changes valence of the outcome
individual
performance
individual
effort
individual
outcome
x
group
performance
Number of others
Own competence
Own unique skills
Group’s incompetence
Valance of
outcome
individual
utility
motivation
group
outcome
Identifiably
Divisibility of outcome
Fairness of reward
distribution
Liking for group members
Identification with group
History of interaction with group
Personal importance of goal
Scenario
•
•
You are member of a 6-person team to select a Wikipedia
article & improve it to good article status
How do you guard against social loafing?
Ways to reduce social loafing
•
•
•
•
•
•
Assign fewer people to work on tasks
(“understaffing”)
Assign individual responsibilities
Make individual performance visible
Define clear, stretch goals
Make the tasks intrinsically interesting
Make the group enjoyable to work in
Download