Causality and Validation of System Dynamics Models
incorporating Soft Variables: Establishing an Interface with
Structural Equation Modelling.
Santanu Roy
National Institute of Science, Technology and
Development Studies
Dr. K.S. Krishnan Marg
New Delhi 110 012
India
Email: santanu @csnistad.ren.nic.in
Pratap K.J. Mohapatra
Department of Industrial Engineering and
Management
Indian Institute of Technology
Kharagpur 721 302
India
Conventional methods and models are based on hard (quantitative, cardinally-measured)
information. The problems are different in the analysis of soft, qualitative or
categorically measured data. Social scientists have been more and more concerned with
measuring qualities in order to graple with complex configurations and the ambiguities
inherent in human perceptions and behaviour. The authors have earlier attempted to
model the work climate of an R&D laboratory using the system dynamics (SD)
framework. Problems occur at two stages in developing such a system dynamics model
incorporating soft variables. First, most of the variables encountered in such systems are
measured using a quasi-quantitative framework. The question of reliability and validity
of such measurement would have to be addressed. Second, the causal relationships
among the variables would have to be ascertained in a way that takes into consideration
this quasi-quantitative measurement approach. Reliability refers to the stability of
replicated measurements. Construct validity refers to whether the measure really
measures what it is supposed to measure, as opposed to measuring some similar yet
conceptually distinct variable. Causality or causal linkages are central to the paradigm of
system dynamics. The causal relationships in the above-mentioned system dynamics
model were largely derived from correlations, regression analysis, cluster analysis and
multiple classification analysis. But in all these methods of analysis, causality cannot be
inferred or verified. Further, there is the critical question of validating such a system
dynamics model. Our approach towards soft system modelling is quite apart from the
methodological thrust of soft systems methodology (SSM) and other problem structure
methodologies. For one, SD itself has moved away from the hard system paradigm, with
the relativist/holistic philosophy of validation. Secondly, in SSM, the problem situation
could be ill-structured and messy whereas the variables in the model need not be so. The
central theme of structural equation modelling is the establishment of causal relationships
among latent variables taking into consideration the reliability and validity of quasi-
quantitative measurement of such variables. It is, therefore, argued that establishing and
interface between system dynamics and structural equation modelling could be
appropriate to address the problem of establishing causality in and validation of a system
dynamics model incorporating soft variables.
CC BY-NC-SA 4.0