The thesis is organised as follows:
Chapter two provides a review of the OR and simulation literature that has con- sidered learning from modelling. This is considered from three perspectives: learning
from model building, learning from model use and relationship between credibility
and learning. This review concludes that learning from models can be thought of
as a process of improving decision makers understanding of a.) the computer model
to changes in attitude about the course of action to take and a deeper understand-
ing about the underlying behaviour of a system that can be transferred elsewhere.
The review also illustrates that although some empirical studies explore learning
from quantitative modelling they are focussed on demonstrating that learning oc-
curs rather than how or where decision makers learn in the process. Moreover, little
learning research has been undertaken in the DES domain.
Chapter three frames the existing OR and simulation literature on learning in terms of Argyris and Sch¨on (1996) concepts of single-loop and double-loop learning.
The chapter links single-loop learning to attitude change and double-loop learning
to both attitude change and transfer of learning. Several simulation examples of
learning under this framework are provided. This is followed by a review of the
attitude change/persuasion and transfer of learning literature from psychology. The
chapter concludes with a discussion of a measurement approach for single-loop and
double-loop learning.
Chapter four details a psychology style laboratory experiment to measure single- loop and double-loop learning. The chapter firstly describes how participants in
the experiment, undergraduate business students, are either involved one of three
experimental conditions: the reuse or building (with high or low experimentation)
of a DES model of a case study A&E department in England. This is followed
by a description of the research hypotheses, the single (attitude) and double-loop
(attitude and transfer) dependent variables and the materials used for measurement.
The chapter ends with a detailed procedure for conducting the experiment.
A descriptive account of single-loop learning in the three experimental condi-
tions, described in chapter four, is presented inchapter five. The aim of this chapter is to familiarise readers with the results by condition without any formal testing of
hypotheses. Conclusions of the analysis serve as a basis for a detailed comparison
Chapter six provides a formal comparison of the single-loop learning results in respect to the research hypotheses. For clarity these are presented as both graphical
and inference tests. The chapter also details the issues in undertaking the analysis
and procedures used to help e.g. bootstrapping. The results are followed by a
discussion focussing on the possible learning mechanisms found in the experiment
inchapter seven.
A descriptive account of double-loop learning in the three experimental condi-
tions is presented in chapter eight. The aim of this chapter is to familiarise readers with the transfer of learning results and correlations with single-loop variables by
condition without any formal testing of hypotheses. Conclusions of the analysis
serve as a basis for a detailed comparison between conditions.
Chapter nine provides a formal comparison of the double-loop learning results in respect to the research hypotheses. For clarity these are presented as both graphical
and inference tests. The results from this chapter are discussed with respect to the
single-loop results (chapters five and six) inchapter ten.
The thesis is concluded with chapter eleven. In addition to a summary of the main findings this chapter provides a discussion of the contribution of the research
to the simulation literature on the value of modelling and model reuse as well as
the limitations. Further work is discussed under three headings: work to address
limitations, work to test the hypotheses generated by the research and work to refine
the experiment. The thesis ends with a brief discussion on how the simulation
community view model building and use and the potential for generic models to
Chapter 2
Simulation and Learning
2.1
Introduction
This chapter provides an overview of the simulation literature, DES or otherwise,
that is relevant for understanding learning that can occur in a modelling inter-
vention. As the aims of the research are to explore decision maker learning given
involvement in model building and model reuse, the chapter tackles these areas
separately.
The chapter starts with a review of the conceptual and empirical literature rel-
evant to model building. This argues that involvement of decision makers in model
building aids transparency of both the computer modeland their own mental mod- els of the problem. The learning outcomes frequently discussed are attitude change
towards action and development of a deep transferable knowledge, although there
is little empirical data available to support these claims.
This is followed by a review of DES and SD literature that explores learning
from a use perspective. In fact, much of the simulation learning research falls into
this category. Furthermore, although many authors are positive about the potential
In order to learn from a simulation model it seems necessary that decision makers
and users feel the model is credible. Thus the third section of this chapter focuses
on simulation study credibility and studies exploring it. Although much has been
written on methods to build credibility in a simulation study little has been done to
measure and test it. Thus a number of other fields that are interested in credibility
are briefly reviewed. The chapter concludes with a summary of the main points
covered in the review.