View
214
Download
0
Embed Size (px)
Citation preview
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20071
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Improvising Around Roles and Scripts in Group Model Building
Group Decision and Negotiation INFORMS, Mt. Tremblant, Quebec, May 2007
David F. Andersen and George P. RichardsonRockefeller College of Public Affairs and Policy
University at Albany, SUNY
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20072
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Outline of Remarks
• What is System Dynamics Group Model Building?• Improvisational Jazz as an Organizing Metaphor
• Initial Agreements— Who is paying for the gig?
• Roles and Teamwork— Getting the players together
• Scripts—Learning the Basic Rhythms and Melodies.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20073
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Outline of Remarks (continued)
• Improvisation with Roles and Scripts• Understanding the Improvisational Playfield
• Two Principles that guide improvisation
• A catalogue of Improvisational Behaviors
• Improvisation that Ignores Roles and Breaks Scripts• Role Reversals and other dangerous moves
• Working “on the fly” (to create new scripts)
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20074
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
What is Group Modeling?
• A form of group decision support, involving a group of stakeholders with a complex problem• Group facilitation
• Model building and refinement in public
• Simulation of scenarios and options
• Extensive facilitated discussion and analysis
• Facilitated policy design and decisions
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20075
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
What is Group Modeling?
• Management team (10-20) with a Modeling/Facilitation team (2-4)
• Four full days over 3-to-4 months• Extensive between meeting work• Rapid prototyping of model with finished
simulation product• Facilitation of implementation plans
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20076
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Why System Dynamics Modeling?
SystemConceptualization
Representation ofModel Structure
Perceptions ofSystem Structure
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20077
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Why System Dynamics Modeling?
SystemConceptualization
ModelFormulation
Representation ofModel Structure
Comparison andReconcilation
Perceptions ofSystem Structure
Diagramming andDescription Tools
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20078
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Why System Dynamics Modeling?
Empirical Evidence
SystemConceptualization
ModelFormulation
Representation ofModel Structure
Comparison andReconcilation
Perceptions ofSystem Structure
Alternative Models,Experience, Literature
Diagramming andDescription Tools
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 20079
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Why System Dynamics Modeling?
Empirical Evidence
SystemConceptualization
ModelFormulation
Representation ofModel Structure
Comparison andReconcilation
Perceptions ofSystem Structure
Alternative Models,Experience, Literature Literature
Empirical andInferred Time
Series
Comparison andReconciliation.
Deduction OfModel Behavior
Diagramming andDescription Tools
Computing Aids
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200710
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Why System Dynamics Modeling?
Empirical Evidence
SystemConceptualization
ModelFormulation
Representation ofModel Structure
Comparison andReconcilation
Perceptions ofSystem Structure
Alternative Models,Experience, Literature Literature
Empirical andInferred Time
Series
Comparison andReconciliation.
Deduction OfModel Behavior
Diagramming andDescription Tools
Computing Aids
StructureValidatingProcesses
BehaviorValidatingProcesses
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200711
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
The Albany Teamwork Approach
• Facilitator / Elicitor• Modeler / Reflector• Process coach• Recorder• Gatekeeper
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200712
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Components of the Process
• Problem definition meeting• Group modeling meeting• Formal model formulation• Reviewing model with model building team• Rolling out model with the community• Working with flight simulator• Making change happen
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200713
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
A Typical Room GMB Session
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200714
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Typical First Group Model Building Meeting
• Introductions: Hopes and Fears• Stakeholders• Introduction to simulation: Concept models• Client flow elicitation• Policy resources and clusters• Mapping policy influences• Next steps for client group and modeling team
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200715
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Introduction to Simulation
• Concept models• Introduce the stock, flow, and causal link icons used
throughout the workshop
• Demonstrate there are links between feedback structure and dynamic behavior
• Initiate discussion about the structure and behavior of the real system
• Less than 30 minutes
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200716
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Concept Model Progression:“Models are ours to change and improve.”
Onassistance
At riskemployedJob finding
rate
Job loss rate
Avg length of stayon assistance
Fraction losing jobper year
Onassistance
At riskemployedJob finding
rate
Job loss rate
Avg length of stayon assistance
Fraction losing jobper year
Onassistance
At riskemployedJob finding
rate
Job loss rate
Avg length of stayon assistance
Fraction losing jobper year
Jobs
Employmentratio
Onassistance
At riskemployedJob finding
rate
Job loss rate
Avg length of stayon assistance
Fraction losing jobper year
Jobs
Employmentratio
Onassistance
At riskemployedJob finding
rate
Job loss rate
Avg length of stayon assistance
Fraction losing jobper year
Jobs
Employmentratio
Unemployedand
unassisted
Loss ofassistance rate
Avg total time onassistance
Fraction losingassistance per year
Onassistance
At riskemployedJob finding
rate
Job loss rate
Avg length of stayon assistance
Fraction losing jobper year
Jobs
Employmentratio
Unemployedand
unassisted
Loss ofassistance rate
Avg total time onassistance
Fraction losingassistance per year
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200717
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Concept Model Progression:“Behavior is a Consequence of Structure”
At Risk Populations4,000
3,000
2,000
1,000
0
0 2 4 6 8Time (Year)
On assistance : welf1At risk employed : welf1
At Risk Populations4,000
3,000
2,000
1,000
0
0 2 4 6 8Time (Year)
On assistance : welf1At risk employed : welf1
At Risk Populations4,000
3,000
2,000
1,000
0
0 2 4 6 8Time (Year)
On assistance : welf2At risk employed : welf2
At Risk Populations4,000
3,000
2,000
1,000
0
0 2 4 6 8Time (Year)
On assistance : welf2At risk employed : welf2
At Risk Populations4,000
3,000
2,000
1,000
0
0 2 4 6 8Time (Year)
On assistance : welf3At risk employed : welf3Unemployed and unassisted : welf3
At Risk Populations4,000
3,000
2,000
1,000
0
0 2 4 6 8Time (Year)
On assistance : welf3At risk employed : welf3Unemployed and unassisted : welf3
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200718
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Typical First Group Model Building Meeting
• Introductions: Hopes and Fears• Stakeholders• Introduction to simulation: Concept models• Client flow elicitation• Policy resources and clusters• Mapping policy influences• Next steps for client group
and modeling team
How do these conversations take place?
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200719
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Facilitating Group Modeling Conversations
• Scripted routines• Richmond, The Strategic Forum
• Vennix, Group Model Building
• Andersen & Richardson, Scripts for Group Model Building
• Improvisation• Varying scripts “on the fly”
• Creating new GMB processes “on the fly”
• Reflecting on improvisations to create new scripts
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200720
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Client Agreements
about Strategies
and Policies
Initial Agreements
Scripts
Roles
Group Modelling Improvisational
Playing Field
Formal ModellingActivities
Improvisation as a Key Activity in Group Modelling
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200721
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Understanding the Improvisational Playfield
• Four Dimensions of Tension in GMB Communications Processes
• Using Boundary Objects to Manage Communication Tensions—three uses• Remembering
• Facilitating
• Modeling
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200722
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Understanding the Improvisational Playfield
• Four Dimensions of Tension in GMB Communications Processes
• Using Boundary Objects to Manage Communication Tensions—three uses• Remembering
• Facilitating
• Modeling
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200723
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Tensions in Group Model Building
• Client Mental Models
• Natural Language
• High Semantic Requirements
• Coherence
• SD Modeling Principles
• Model Equations
• High Syntax Requirements
• Cohesion
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200724
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.
Examples of Boundary Objects in SD GMB
• Client-Authored such as…• Variable Graphs Over Time
• Stakeholder Power X Interest Grid
• Facilitator-Authored such as…• Sketches of Model Feedback
• Structure on the White Board
• Modeler-Authored such as…• Refined Sketches of Model
Structure in Modeler Feedback Script
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200725
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Boundary Objects Manage Tensions BetweenMental Models and Good Modeling Principles
Client Group’s Mental Models
SD Modeling Principles
Current GMB Boundary Object
Already Completed Boundary Objects
Facilitation Zone
Remembering andDisplaying
Modeling Zone
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200726
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
First Improvisational Principle:
LERT
Listen, Edit, and Report with Transformations
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200727
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
LERT has two parts…
• Listen and Report Back• Use exact words, concepts, and phrases that client group uses
• Faithfully record and display their thoughts and words
• Edit with Transformations• Find ways to “filter” clients’ speech — “less is more”
• Add value by structuring speech around modeling principles
• These two components of LERT are at odds with themselves
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200728
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Second Improvisational Principle:Unobtrusive Teaching
• About SD Modeling Syntax and Principles, e.g.• Concept Models
• Modeler Feedback
• About Insights into the System Under Study, e.g.• Feedback while working in small groups
• Modeler Feedback
• Plenary Events that “ring a bell” for a potentially big idea within the project
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200729
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Toward a Taxonomy of Improvisational BehaviorsLive “On the Fly” Behaviors
“Off Line” Behaviors
Mostly Modeler Behavior
• Transform cause-and-effect to feedback
• Propose “cleaned up” model geometry
Mostly Facilitator Behavior
• Select next concept to discuss
• Move Ideas to the Parking lot
Joint Modeler & Facilitator Behavior
• “Banter” with the group in structured ways
• Propose Seed Structures to discuss
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200730
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Key Modeler Improvisational Behaviors
• On the Fly• Challenge “Group Think”
causal assertions
• Provide “Expert Opinion” on aggregation or other modeling issues
• Off Line• Transform cause-and-effect to
feedback
• Clean Up Diagram Geometry
• Add Operational Logic
• Propose formulation-based restatements of key effects under discussion by group
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200731
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Key Facilitator Improvisational Behaviors
• On the Fly• Select next key concept of
variable to discuss
• Do not record on the board aspects of discussion
• Park aspects of discussion for future use
• Select words that create variables from verb phrases
• Be alert to and draw out feedback loops
• Off Line• Add insights, comments to
small working groups
• Rehearse key variables and dynamic mechanisms from earlier boundary objects
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200732
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Key Joint Improvisational Behaviors
• On the Fly• Call out for help
• Live “banter” to exchange insights in front of the client group
• Off Line• Design “seed” structures or
dynamics mechanisms to structure next phase of discussion
• Share insights about key variables or dynamic mechanisms
• Plan or refine sequence of scripts
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200733
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Improvisation that Ignores Roles and Breaks Scripts
• Role Reversals and Other Unplanned Moves• Can lead to sudden break-throughs
• Can have dangerous consequences
• We have rules to constrain these behaviors!
• Creating New Scripts in “Real Time”• To resolve a current problem facing the group
• ***Often to create a new script
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200734
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Role Reversals and Other Unplanned Moves
• Improvised Conversations Between Modeling Team Members
• Four Basic Types:• Facilitator Seeks Assistance
• Facilitator Steps Down
• Modeler Steps Up to Offer Insight
• Modeler Steps Up to Take Over the Meeting
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200735
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Even Improvised Conversations Have Rules
• Rule #1: The person “holding the chalk” calls the shots
• Rule #2: Always know who is “holding the chalk”
• Rule #3: Always seek permission for improvised conversations
• 3A: Facilitator initiates conversations with public requests
• 3B: Modeler initiates conversations with a private signal
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200736
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Facilitator Seeks Assistance
• What and When: At an unplanned time, facilitator requests specific support from modeling team in handling group process or content.
• Why: Facilitator senses that current task is not working well or there is an opportunity that needs to be structured
• How: Facilitator sends public signal with an advance request for help. There may be a public discussion of what is needed. When the modeler is ready to help, he or she sends a private signal back that it is OK now to call for help.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200737
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Facilitator Steps Down
• What and When: This is usually an extreme case of facilitator requesting assistance.
• Why: Facilitator senses that current task is not working well or facilitator is tired or confused for some reason.
• How: Facilitator sends public signal with an advance request for help. There may be a public discussion of what is needed. When the modeler is ready to help, he or she sends a private signal back that it is OK now to call for help.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200738
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Modeler Steps Up to Comment
• What and When: At an unplanned time, modeler gains permission to interrupt to add value around some specific task.
• Why: Modeler sees an opportunity to clarify a specific task or insight being worked by the group (that the facilitator seems to be missing)
• How: Modeler uses a private signal indicating he or she has a new direction to pursue. If a break is not near, a public discussion may follow. When the facilitator is ready, the modeler comments to add value. The modeler should be prepared to “take the chalk” if necessary.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200739
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Modeler Steps Up to Take Over Session
• What and When: At an unplanned time, modeler gains permission to interrupt to move the group in a new direction —We consider this to be a very risky move.
• Why: Modeler sees an opportunity to move the group in a new (often breakthrough) direction, timing is important, and it seems prudent (to the modeler) to not wait until a break.
• How: Modeler uses a private signal indicating he or she has a new direction to pursue. If a break is not near, a public discussion may follow. When the facilitator is ready, the modeler steps up to manage the conversation, agreeing to stay in control until the next break or “pass back” time can be arranged.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200740
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Creating New Scripts in “Real Time”
• Driven by a specific need or opportunity presented by current work
• Never done “on the fly”• Usually completed over a long break, before the start
of a new day, or late at night (those early morning surprises!)
• Initially “crap detected” using small group working principles
• Extensive critique and re-design at end of session.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200741
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Small Group Principles in a Nutshell
• Divergent, brainstorming tasks• Individuals to pairs to small groups
• Nominal group collection
• Ranking tasks• Sticky dots or software support
• Convergent, design tasks• Hardest – need most thought
• Most risk.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200742
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Borrow or Steal Scripts from other Problem Structuring Approaches…
• Rohrbaugh• Decision Techtronics approaches
• Vennix • Workbook ideas and approaches
• Bryson • Stakeholder mapping
• Leadership principles
• Eden and Ackermann• Decision and Group Explorer techniques
• New direct software linkages
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200743
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
Several Examples of Scripts-to-Order
• Group elicitation of model parameters• Vermont Medicaid project
• Needed numbers for over-night modeling
• Proved to be powerful alignment and discussion tool
• Ratio elicitation of feedback loops• Office of Mental Health Project
• Original intent to decompose too complex stock and flow picture
• Use of ratio of two key stocks seems a “natural” way to generate feedback thinking by client groups
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200744
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
More Examples of Scripts-to-Order• Using Group Explorer Pack of Key Variables to “Seed” Feedback
elicitation• Airport Security Modeling Project• Seeking linkages between Group Explorer and Vensim modeling• Techniques still under development
• Extended Concept Model Approach• Project on Security in Norwegian Oil Fields• Initial intent was to transfer model ownership
to new group• Discovered fast way to engage clients early
in discussion of dynamics.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
D. F. Andersen & G. P. RichardsonGDN, INFORMS, May 200745
Rockefeller College of Public Affairs and PolicyRockefeller College of Public Affairs and PolicyUniversity at AlbanyUniversity at Albany
References• Reagan-Cirincione et al., Decision modeling: Tools for Strategic
Thinking, Interfaces 21,6 (Nov-Dec 1991). • Andersen & Richardson, Teamwork in Group Model Building, System Dynamics
Review 1995.• Richardson & Andersen, Scripts in Group Model Building, System Dynamics
Review 1997.• Richmond, The Strategic Forum: Aligning Objectives Strategy and Process.
System Dynamics Review, 13(2), 1997.
• Vennix, Andersen, & Richardson, Special issue of the System Dynamics Review on Group Model Building, 1997.
• Eden and Ackermann, Strategy Making: the Journey of Strategic Management. Sage: London, 1998.
• Bryson, Strategic Planning for Public and Nonprofit Organizations, San Francisco: Jossey-Bass, 2004