M M M:

pucefakeΤεχνίτη Νοημοσύνη και Ρομποτική

30 Νοε 2013 (πριν από 3 χρόνια και 6 μήνες)

65 εμφανίσεις

M
ODELING

THE

M
IND

AND

THE

M
ILIEU
:

Computational Modeling for Organizational
Psychologists

"M
ATHEMATICS

IS

THE

LANGUAGE

WITH

WHICH

G
OD

HAS

WRITTEN

THE

UNIVERSE
."








--


G
ALILEO

G
ALILEI



Mathematics has
not

tended to be the language of theories
in psychology and organizational science


Math models need to be solved, but often seen as intractable when
describing human or organization behavior


Computational models do not need to be solved


Computational models are algorithmic descriptions of
process details, typically operationalized as computer
programs that are dynamic and can be simulated (Taber &
Timpone
, 1996)


JAP

has only published one computational model, ever
(Vancouver,
Weinhardt
, & Schmidt, 2010)


0.3
% of
AMJ
articles are computational models


in last 11
years, only 1.3% of
AMR

articles included
computational models (nearly all macro or
meso
)


Y
ET
, C
OMPUTATIONAL

M
ODELS
:


Increase precision and transparency


Less ambiguity of concepts and explanations


Specific predictions
(compared to natural language
theories)


Assure internal (logical) consistency


Model works


Accounts for phenomena claimed


Identify unanticipated consequences


Simulations can lead to new findings

D
YNAMICALLY

C
HALLENGED


Human ability to predict values in dynamic
variables is low (even among those in STEM
fields)


Dynamic variables are variables with memory


Stocks, levels


Predicting the behavior of dynamic, nonlinear
processes, interacting subsystems


Forget about it

H
INTZMAN

(1990)


To have one's hunches about how a simple
combination of processes will behave repeatedly
dashed by one's own computer program is a
humbling experience that no experimental
psychologist should miss” (p. 111).

O
BJECTIVES
:

B
Y

THE

END

OF

THIS

WORKSHOP

YOU

WILL

KNOW

HOW

TO




Identify a problem worthy of modeling


Define the system to be modeled


Build a model


Evaluate a model

S
TEP

1: I
DENTIFY

P
ROBLEM


Dynamic phenomena


All phenomena?


Existing theory


Lot’s of talk, no models


Existing computational architectures


Neural networks


Systems dynamics


Cybernetics

J
OB

A
TTITUDES

AND

S
TRESS


Cybernetic, natural language theories on both
topics


Hulin & Judge’s (2003) review of job attitude
models

T
HE

U
BIQUITOUS

C
OMPARATOR

S
TEP

2: S
YSTEM

D
EFINITION


Unit(s) of analysis


Individual in context


Problem boundary: just enough


Core dynamic processes


Restrictions


Time frame (100 days)


Variables: add as needed











Well
-
Being



Importance



Coping

Physical and Social

Environment

Perception

Desires

Discrepancy

P
ART

OF

E
DWARD

S

T
HEORY

OF

S
TRESS

AND

W
ELL
-
B
EING

S
TEP

3: B
UILDING

THE

M
ODEL


Vensim


System Dynamics platform (Forrester)


Units often organizations or other larger systems


Coopting for psychological modeling


Individual cognitive processes


Individual in context


Open software


Model setting: Unit of time; “Days”

Menu

Sketch Tools

Main Toolbar

Output File

Window

Simulation

Analysis Tools

Build Window (Where you
build your model)

Lock

Move

Variable

Level
Variable

Arrow

Rate

Shadow
Variable

Input

Output

Comment

Delete

Equation

Reference

S
KETCH

TOOLS

K
EY

V
ARIABLES

IN

THE

M
ODEL

Variable Name

Type of
Variable

Time

Person/
Environment

perceptions

Endogenous

Time
-
varying

Person

Environmental States

Endogenous

Time
-
varying*

Environment

actions/coping behavior

Endogenous

Time
-
varying

Person

discrepancies

Endogenous

Time
-
varying

Person

desires

Exogenous

Constant (0)

Person

Well
-
Being/Job Satisfaction

Endogenous

Time
-
varying*

Person

importance

Exogenous

Constant (1)

Person

T
YPES

OF

D
ESIRES


Optima: Not too much; not to little


Minima: Only values exceeding desire a problem
(e.g., budget)


Maxima


Hard maxima: values exceeding desire ignored


Soft maxima: more is better, but with diminishing
returns

S
TEP

4: E
VALUATING

THE

MODEL


Simulations that works


Postdiction


Assess assessment strategies


Will past designs and analysis have been
diagnostic?


Hypothesis generation and testing


Strong inference via model comparison


Differing predictions


Model fitting


Complexity (# parameters) vs. fit

D
ON

T

M
ARRY

YOUR

M
ODEL
!

Questions?



Further information:


ORM tutorial
:
Vancouver, J.B., & Weinhardt, J.M.,
(online).
Modeling
the mind and the milieu: Computational modeling for micro
-
level
organizational researchers.
Organizational Research Methods
.


Modeling in Org Psych
:
Weinhardt, J. M. & Vancouver, J. B. (in
press
)
.
Is there a computational model in your future? Only the math
will tell.
Organizational Psychology Review
.


Symposium
: Understanding Dynamics Conceptually, Analytically,
Computationally, and Empirically. Tuesday, Aug 7 2012

11:30AM
-

1:00PM.
Boston Park Plaza, Beacon Hill
Room.


Web site:
https://sites.google.com/site/motivationmodeling/home



Help from:
Justin Weinhardt; Mike Warren; Amanda Covey; Justin Purl; Xiaofei Li