Subject: Re: Book referral
From: "Karsten M. Self" <kmself@ix.netcom.com>
Date: Sun, 31 Oct 1999 05:31:28 +0000

Brian Bartholomew wrote:
> 
> > There is an anthology of decisionmaking behavior (individual and
> > group) that I've been meaning to mention for months now.  I don't
> > have it in front of me now.  Please feel free to box me 'round the
> > ears if I don't mention the title.  There's much coverage of issues
> > such as what Stephen mentions above, as well as others I found
> > highly relevant to how free software operates.
> 
> What's the title?

Well, it's brown, for brown shades of green.... [1]

Plous, Scott, _The Psychology of Judgement and Decision Making_,
McGraw-Hill Series in Social Psychology, McGraw-Hill, Inc, New York,
1993 ISBN: 0070504776 
http://www.amazon.com/exec/obidos/ASIN/0070504776

This is a survey of the judgment and decision making literature and
research landscape.

Contents:
Perception, memory, and context
	Selective Perception
	Cognitive Dissonance
	Memory and Hindsight Biases
	Context Dependence

How questions affect answers
	Plasticity
	The effects of question wording and framing

Models of decision making
	Expected utility theory
	Paradoxes in rationality
	Descriptive models of decision making

Heuristics and biases
	The representativeness heuristic
	The availability heuristic
	Probability and risk
	Anchoring and adjustment
	The perception of randomness
	Correlation, causation, and control
	Attribution theory

The social side of judgment and decision making
	Social influences
	Group judgments and decisions

Common traps
	Overconfidence
	Self-fulfilling prophecies
	Behavioral traps



What I liked:

The cover.  [2]

This was an introduction to a topic I'd otherwise have been unlikely to
pick up.  The popular treatments are, well, too popular.  The sources
themselves are a bit thick.  The book was used in conjunction with a
local university extension course which a friend attended.  Makes me
wish I took the course, though her review was mixed (great topic,
participation was inconsistent).

The section on heuristics and biases is very interesting on a personal
basis -- it provides insight on many unconscious processes.

What really grabbed me though was the extion on group judgments and
decisions, and the traps.  

The traps are fairly well known:  social facilitation (changes in work
performance when others are present), social loafing (slackening effort
when participating as a group rather than individually), bystander
effects (modifications of behavior based on actions of other
bystanders), conformity, groupthink, polarization, etc.

More interesting was the structure of groups making effective
decisions.  Groups typically outperform individuals, but the best
members of groups do better working independently.  One of the most
effective group structures is a "dictatorship".  Rather than voting on
decisions, a "dictator" makes the decision.  This is based on work by
Janet Sniezek (1989) [3]
(http://www.beckman.uiuc.edu/faculty/sniezek.html).  Five group
structures were evaluated, each consisting of five college students,
each rotating through all five methods, each estimating campus bookstore
sales in the upcoming month, tested against actual sales.

  consensus  -- face-to-face discussion resulting in one shared
judgment.
  dialectic  -- include discussion of bias factors.
  dictator   -- "best member", essentially peer-selected.
  delphi     -- rounds of anonymous answers leading to consensus.
  collective -- answers averaged to "group" result w/o interaction.

Collective fared worst.  "Dictator" reduced error by a factor of three
over the three remaining methods.  Dictators who revised responses in
the direction of the group tended to *increase* error.

I found a strong resonance with Linus Torvalds who's often described as
a benevolent dictator. While he makes decisions, it's based on the faith
of those following him.  His decisionmaking power is absolute, but his
position itself is not (Linux could fork, another developer could assume
the lead position).  Mix of dictatorship and democracy, position
maintained by effectiveness of results, not political power or force of
leadership itself.

Other interesting traps -- overconfidence (how many CEOs does it
describe?), calibration (weather forecasters are better than you think
<g>), confidence and accuracy, and the importance of feedback.  Makes me
suspect that an open framework is a better decisionmaking environment
than a corporate one, from a sociological perspective (economic
incentives of a corporation may change the mix).

I'd recommend a light reading of the front material and a close study of
the last three chapters.  Comments from anyone with a better grasp of
this than me much appreciated.


-- 
Karsten M. Self (kmself@ix.netcom.com)
    What part of "Gestalt" don't you understand?

SAS for Linux: http://www.netcom.com/~kmself/SAS/SAS4Linux.html
Mailing list:  "subscribe sas-linux" to
mailto:majordomo@cranfield.ac.uk    
  9:31pm  up 2 days, 11:37,  2 users,  load average: 0.15, 0.12, 0.09

[1] Reference to my off-list comment to Brian who'd asked for more info
before I'd found the book again.  Coulda sworn it was brown.  Hmm....
[2] A:  no, not until it was pointed out.
[3] Sniezek, J.A. &  Henry, R.A (1989( Accuracy and confidence in group
judgment.  _Organizational Behavior and Human Decision Processes, 43_,
1-28.