Transcript The Hidden Power of Social Networks and Knowledge Sharing

The Hidden Power of Social
Networks: Developing Knowledge
Maps Using Social Network Analysis
Dr. Jay Liebowitz
Professor
Graduate Division of Business and Management
Johns Hopkins University
[email protected]
Sharing Knowledge is Power?!
(December 10, 2004)
Where is the “Business Intelligence”?
“How to Battle the Coming Brain Drain”
(Anne Fisher, Fortune, March 21, 2005)
• “By 2010, more than half of all workers in the
U.S. will be over 40. Tens of millions of babyboomers turn 60 this year, and the decade
ahead will see vast numbers of people retiring,
or at least leaving their current full-time careers.”
• General Electric, Dow Chemical, Northrop
Grumman: capture their oldsters’ knowledge &
disseminate it to younger workers before it’s too
late
Lost Knowledge
• “In sectors such as government,
manufacturing, energy, health care,
education, and aerospace, knowledge
retention will become an overriding
concern in the years ahead.” [David
DeLong, Lost Knowledge: Confronting
the Threat of an Aging Workforce,
Oxford University Press, 2004]
Keynote Address: Medical Informatics and Emergency
Medicine (Dr. Feied, 2004 AEM Consensus Conference on
Emergency Medicine Informatics)
• “Work to reduce institutional dependence
on specialized personnel with ‘secret
knowledge’ that allows them to complete
tasks nobody else can perform. If the unit
secretary is the only one who knows how
to place or cancel an order, every coffee
break can put a congested department
further behind.”
The Hidden Power of Social
Networks (Cross, 2004)
• Networks of informal relationships have a
critical influence on work and innovation
• Research shows that appropriate
connectivity in well-managed networks
within organizations can have a
substantial impact on performance,
learning, and innovation
• Actors/nodes (individuals/units) and
links/arcs (relationships/ties)
Conducting a Social Network
Analysis
• Step 1: Identify a Strategically Important Group
– Integrating networks that cross core processes
– Promoting innovation through connectivity in new
product development or process improvement
initiatives
– Facilitating post-merger integration and large-scale
organizational change
– Supporting CoPs
– Forming strategic partnerships and alliances
– Improving learning and decision making in top
leadership networks
Size of the Network
• Typically limit to 250 people for
visualization purposes (but could have
more for analysis purposes)
• Ask each person to rate his/her
interactions with other members of the
group
Step 2: Assess Meaningful and
Actionable Relationships
• Relationships that reveal collaboration in a
network (e.g., Communication, Information,
Problem solving, Innovation)
• Relationships that reveal the information sharing
potential of a network (e.g., access)
• Relationships that reveal rigidity in a network
(e.g., decision making, influence)
• Relationships that reveal well-being and
supportiveness in a network (e.g., liking,
friendship, trust)
Survey
• Confidentiality
– Full disclosure of the results via an allemployee debriefing or other communication
– Allow only one person or a select group to
see the full results so that helpful action can
be taken
– Disclose no names whatsoever
Step 3: Visually Analyze the
Results
•
•
•
•
•
UCINet-Netdraw
Pajek
Netminer
Inflow
International Network of Social Network
Analysts (INSNA)
Staff for General Advice
Questions
Junior-Senior Staff
Relationships
Network Diagram: General
Advice (Individual)
Network Diagram: Subject
Matter Expertise (Department)
Social Network Analysis (www.orgnet.com)--Org.
Mapping
Step 4: Quantitatively Analyze the
Results (Indiv. Network Measures)
• In-degree centrality (the number of incoming ties
a person has for a given relationship (such as
communication or trust)
• Out-degree centrality (the # of outgoing ties a
person has for a given relationship)
• Betweenness centrality (the extent to which a
particular person lies “between” the various
other people in the network—networks that
contain individuals with high betweenness are
vulnerable to having info flows disrupted by
power plays or key individuals leaving)
Indiv. Network Measures (cont.)
• Closeness centrality (the extent to which a
person lies at short distances to many other
people in the network. Persons highly central
with respect to closeness tend to hear info
sooner than others)
• Brokerage measures
– Coordinators (people who broker connections within
the same group)
– Gatekeepers (people who broker connections
between their own group and another)
– Liaisons (those who broker connections between 2
different groups)
Group Measures
• Density (the # of individuals who have a given
type of tie with each other, expressed as a % of
the maximum possible—if each person were
connected to every other person in the network,
the density would be 100%)
• Cohesion (the average of the shortest paths
between every pair of people in the network—
average cohesion score should be around 2 in
groups where managers are interested in
employees leveraging each other’s expertise)
Step 5: Create Meaningful
Feedback Sessions
• Conduct interviews with a select number
of people (8-10 network members) to
better understand the dynamics behind the
network
• Look at the network analysis to determine
who to interview (central and peripheral
roles)
• Documentation or workshop with all
network participants
Step 6: Assess Progress and
Effectiveness
• Follow-up (post-audit): 6-9 months after
network analysis is first conducted
• SNA research areas:
– Improving the calculation of interval/ratio
scales for measuring the strength of ties
between actors (AHP)
– Effective representation of complex network
information (IBM’s Raison, data mining tool)
Identifying Types of Individuals in
the Network
• Central Connectors: most arrows pointing
to them
– Unsung Hero: “Invisible” work but critical to
the network and can consume many hours
each day (may be completely unrecognized
by senior management)
– The Bottleneck: some people become so
central to the organization that they end up
holding the group back
Boundary Spanners
• Provide critical links between 2 groups of
people that are defined by functional
affiliation, physical location, or hierarchical
level
• Can play important roles in large-scale
change efforts
Information Brokers
• Indirect connections
• For example, Ian and Frank have no direct
connection, but each has a relationship
with Gayle (so they are connected through
her); people such as Gayle play a
brokering role that can hold together entire
groups
Peripheral Specialists
• Have one connection each and are not
linked to each other
• Some people are “stuck” (e.g.,
newcomers) on the periphery and others
may “choose” (e.g., experts) to be on the
periphery
Case Study: Examination of
Raw Data
• All data downloaded to Excel spreadsheet
• 225 respondents answered questions 1, 2, and some
portion of 13, identifying other employees from whom
they sought a form of knowledge advice
• All others were removed, including duplicates
• All cells containing names were examined identifying
abbreviated names, the use of initials, variations in
spelling, etc.
• Numbering of all names yielded 698 employees
Examination of Raw Data
• The data identified 1,621 knowledge advice connections
or communications between the 698 employees
• Each connection between employees was assigned a
weight based on the frequency of the knowledge being
sought and the importance of the knowledge
• Complete employee data for respondents, nonrespondent employee numbers, employee connections
and connection weights were loaded into NetMiner 2.5
for social network analysis.
• Analysis was conducted across 6 defined advice
communication areas
Social Network Analysis Schematic
Adjacency – Advice
Communication Matrix
Employee
Advice Communications
Employee
Employee Attributes
Communication Measures
Social Network Analysis Schematic
Employee
Adjacency – Advice
Communication Matrix
Advice Communications
•Context K – Advice
•Expert Process K – Advice
Employee
•General K – Advice
•Process K – Advice
•Strategic K – Advice
•Relationship K - Advice
Knowledge Types
Context
K of “what” applications
Expert Process
K of “how” networks and
systems work
K of non-work related
questions
General
Process
Relationship
Strategic
K of “how” the business
works
K of “who” has information
K of “why” business
opportunities reduce cost
and other strategic decisions
Social Network Analysis Schematic
Employee
Adjacency – Advice
Communication Matrix
Employee Attributes
•Department
Employee
•Level (Position Hierarchy)
•Tenure
Social Network Analysis Schematic
Employee
Adjacency – Advice
Communication Matrix
Communication Measures
•Centrality (in and out degree)
•Closeness
Employee
•Betweenness
•Cliques
•Density
•Brokerage
Is there much intra-departmental
communication?
• This can be measured by examining the amount
of communication within each department in two
areas: density and cohesiveness
• Density is the proportion of possible lines that
are actually present in the network. It is the ratio
of the number of lines present to the maximum
possible.
• Cohesion Index is the extent to which ties are
concentrated within a subgroup, rather than
between subgroups.
Is there much inter-departmental
communication?
• Departments “blocked” and then examined for degree
and density measures
• As with individuals, departments yielded low density of
communication with other departments
• Some departments have no communication with others
within certain knowledge areas (“structural holes”)
• XYZ department: unconnected from all others in 4 of the
6 knowledge types, including the knowledge of “who”
has the organizational information
• JKL department: not connected within 3 of the
knowledge types including the “what” of applications and
the “how” of network and systems work
Is the organization well
connected among employees?
• Most communication is occurring between
non-managerial employees
• Of the 1,621 knowledge connections,
1,199 were connections to nonmanagement employees
• 74% of reported advice communications
were employee to employee
Is the organization well connected
among the employees and the
managers, directors and executives?
• 336 of the 1,621 knowledge connections (21%)
were to persons in management positions
• 36 of 62 Managers, Directors, and Executives
were named as sources of knowledge advice 5
or more times
• Communication rate from employee to employee
is the highest
Are the junior employees
interacting with the senior
employees?
• Junior employees have limited contact
with senior employees (executives)
• Employee contact is greater with other
employees
• Greatest direct connection between
executives and non-managerial
employees was in the Process K area
• Lowest level: Context K
Are the directors and executives in
the “Power” positions centrally
located?
• Executives and directors are not present or are
relatively weak in most power or central
positions in knowledge advice communications
• A single executive was prominent in Strategic,
General, and Relationship knowledge areas
• Three others in managerial positions appeared
in central positions, with a single one of these
appearing in 4 of the 6 K types
• Most power or central positions are held by nonmanagerial employees or a single expert
Are there correlations as to those
employees sought based on the different
types of knowledge?
• No evident correlation between knowledge type and
employee attributes in central or power positions within
each knowledge area
• While individual actors hold multiple positions across K
area, no single department appears strongly central with
a given K type
• Common factor for all employees sought for advice (with
complete attribute data) was tenure greater than 1 year
(with the exception of employee 188, significantly in the
power position for Strategic Knowledge)
Are there more isolates, transmitters,
receivers or carriers in the organization?
• In all knowledge areas, the greatest
number of employees are isolates
• Receivers are generally greater in number
than transmitters in all knowledge areas
• Carriers are the fewest in number in all
knowledge areas
Node Types by Area
K Area
Context
Exp. Pr
General
Isolate
329
361
378
Process 442
Relation 445
Strategic 396
Transmitter
133
126
118
Receiver
190
171
162
Carrier
46
40
42
110
94
113
109
121
151
37
38
38
How can communication and
collaboration be improved?
• Further departmental examination of results
• Follow-up interviews with persons in central or power
positions
• Task analysis of these persons’ performance in the
positions
• Positive reinforcement program should be developed to
create behavioral change in other positions, based upon
benchmarks set by central or power positions
• Planned insertion of employees or departments to fill all
structural holes should be considered for succession
planning and workforce development
Organizational & Individual
Constraints (Q.25,26,27)
•
•
•
•
•
•
•
37%: Lack of Time/Work Overloads
14%: Too Many Silos/Consolidate into Central Repository
10%: Not knowing who to ask/who has the information
6%: Job Insecurity/Downsizing
6%: Getting People to Share Their Knowledge/Culture Doesn’t Encourage it/Tie to IPEP
6%: Search Engine Needs to be Improved/No Search Engine for Internal Info
Other (21%)(in order):
– Business processes needed/Standardization/People to follow processes
– Politics
– Rigid hierarchical structures
– Supervisor is seated away from Dept/Geographic Dispersion
– Limited access, due to security, to access intranet from home
– More communication needed between people
– Need knowledge capture tool
– Constantly changing organization with unclear roles and responsibilities
– Need to create a continuous learning culture
– Need to capture rationale why things don’t work
KM-Related Recommendations
Based on SNA, Q25,26,and 27
•
•
•
•
•
•
•
•
•
•
•
Portal
Central repository
Google-type search engine
Expertise locator
Recognize knowledge sharing in performance reviews
(extrinsic and intrinsic motivators)
Lessons learned/best practices
Online communities of practice
More cross-training/cross-functional teams
Mentoring program
Knowledge sharing forums (“storytelling”)
Knowledge fairs around certain key topics
Summary
• Low volume of communications: due to an
unbounded network and low levels of advice
seeking
• Departments have relative cohesions of
communication but often fail to communicate
with other divisions across all knowledge areas,
creating structural holes
• Within the Process K area, executive level
actors are most strongly connected to nonmanagerial employees
• Experts, while present in the analysis, do not
appear in powerful or central positions (with the
exception of employee #158)
Summary (cont.)
• Other individuals appear in powerful or central
positions and may be considered experts
relative to the knowledge area (the most
significant of these is employee #157)
• All knowledge areas demonstrate high numbers
of isolated actors and fewer numbers of
transmitters and carriers (employees not
communicated with others within certain
knowledge types)
• Apply knowledge management/knowledge
sharing approaches
Final Comments
• Social network analysis is a wonderful technique
to identify knowledge flows and knowledge gaps
in organizations to help in knowledge
mapping/knowledge audits
• The grapevine effect is stronger than the formal
org-chart effect (i.e., informal networks are
stronger than formal ones)
• SNA helps build a basis for developing a
knowledge management and human capital
strategy
Questions
and Answers