Quantitative Methodologies using Multi-Methods

Models for Social Science and Information Technology Research

Sergey Samoilenko, Kweku-Muata Osei-Bryson

• Publisher: Routledge
• Year: 2021
• ISBN:
  9781000431131
• Edition: 1
• Topic: Social Sciences
• Subtopic: Social Science Research & Methodology
• Index: Social Sciences

Quantitative Methodologies using Multi-Methods is a multifaceted book written to help researchers. It is a user-friendly introduction to the popular methods of data mining and data analysis. The book avoids getting involved into details that are more suitable for more advanced users; it is written for readers who have, at most, a surface-level knowledge of the methods presented in the book. The book also serves as an introductory guide to the subject of complementarity of the tools and techniques of data analysis. It shows how methods could be used in synergy to offer insights into the issues that could not be dissected by any single method alone.

This text can also be used as a set of templates, where, given a set of research questions, the investigator could identify a set of methodological modules for answering the research questions of interest. This is not entirely unlike the relationship between theanalysis and design phases of the systems development life cycle—where the Wha t of the analysis phase has to be translated into the How of the design phase. The book can guide the identification of modules (the How ) that are suitable for answering research questions (the What ). It can aid in transitioning a conceptual domain of the research questions into a scaffolding of data analytic and data mining methods.

The book is also a guide to exploring what data under investigation holds. For example, an investigator may use the methodological modules presented in this book to generate a set of preliminary questions which, after a careful consideration and a requisite culling, could be formulated into a set of questions consistent within a selected theory or a framework. Finally, the book can be used as a generator of new research questions. Applying every method in each of the book's modules opens a new dimension ripe with follow-up questions such as, Why is this so? The answers to this question may provide new insight and lead to the development of a new theory.

Section III

Methodological Modules – Examples of Their Application

18

A Hybrid DEA/DM-based DSS for Productivity-Driven Environments

INTRODUCTION

Modern organizations typically operate in dynamic, competitive environments. Within this context, the critical issues of organizational survival and advancement often lead to calls for improvements in the levels of effectiveness and efficiency of performance. However, due to the relativity of the concepts of efficiency and effectiveness, productivity-driven organizations must take into consideration the performance of their competitors. This requirement is due to the dynamic nature of the business environment which will cause the levels of performance of competing organizations to change over time, and if the efficiency of the competitors has improved, then a productivity-driven organization must respond with its own improvements in efficiency.

A desired capability of an organization to successfully respond to efficiency-related challenges suggests the need, first, for an effective mechanism that allows for discovering appropriate productivity models for improving overall organizational performance and, second for a feedback-type mechanism that allows for evaluating multiple productivity models in order to select the most suitable one.

The dynamic nature of the business environment also suggests the presence of a concept that is central to a productivity-driven organization, namely, that of the superior stable configuration. Given the goal of achieving a high level of efficiency of conversion of inputs into outputs, a superior stable configuration in the context of a productivity-driven organization may imply a model of conversion of inputs into output (input–output model) characterized by a high level of efficiency.

Overall, a decision maker tasked with a responsibility of improving performance of productivity-driven organization existing within a dynamic business environment must take into consideration internal (organizational) and external (environmental) factors. Similarly, if a decision making is to be added by an Information System, then the designers of such system must implement two sets of functionalities: externally oriented and internally oriented. The externally oriented functionality is directed toward evaluating the external competitive environment of a productivity-driven organization, as well as identifying the differences between the current state of the organization and the states of its competitors. The internally oriented functionality, on the other hand, is directed toward the optimization of the level of productivity of the organization, as well as toward an identification of the factors impacting the efficiency of the input–output process.

In this chapter, we will describe a decision support system (DSS) that allows assessing and managing the relative performance of organizations. Specifically, we focus on organizations that consider the states of their internal and external organizational environment in the formulation of their strategies, such that the achievement of an organizational goal is dependent on the level of performance that is commonly measured in terms of the levels of the efficiency of utilization of inputs, effectiveness of the production of outputs, and efficiency of conversion of inputs into outputs.

DESCRIPTION OF THE DSS

The focus on the efficiency assessment suggests that an important component technique of the DSS is data envelopment analysis (DEA). However, other techniques are also required for providing answers to several questions that are relevant to the organization's search for the productivity model that is most suitable with respect to survival and advancement. Next, we outline how a DSS could be implemented using a combination of parametric and non-parametric data analytic and data mining techniques including DEA, cluster analysis (CA), decision tree (DT), neural networks (NN), and multivariate regression (MR).

Let us discuss how each of the above mentioned methods, alone, or in combination with other methods, could be used in the DSS. First, we will discuss externally oriented functionality.

EXTERNALLY ORIENTED FUNCTIONALITY

CA allows for segmentation of the data set into naturally occurring heterogeneous groups. An application of this method allows for detecting the presence of multiple disparate groups of competitors in the external business environment. A decision maker can also determine whether the clusters comprising the data set differ in terms of the relative efficiency of utilization of inputs or production of outputs – DEA will help in this regard. By specifying a DEA model and running the analysis, we can obtain scores of the relative efficiency for each cluster, as well as see how the scores differ between the clusters.

If the data for the same group of competitors available for two points in time, let us say, Year 1 and Year 2, then a decision maker can obtain insights regarding possible changes in the number of clusters, as well as changes...