Elucidating the conceptual structure of a business domain via exploratory network analysis of business survey data

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal

4 Scopus Citations
View graph of relations



Original languageEnglish
Pages (from-to)6359-6369
Journal / PublicationExpert Systems with Applications
Issue number7
Publication statusPublished - 1 Jun 2012


Traditional hypothesis-driven research domains such as molecular biology are undergoing paradigm shift in becoming progressively data-driven, enabling rapid acquisition of new knowledge. The purpose of this article is to promote an analogous development in business research. Specifically, we focus on network analysis: given the key constructs in a business research domain, we introduce a data-driven protocol applicable to business survey data to (a) discover the web of influence directionalities among the key constructs and therein identify the critical constructs, and to (b) determine the relative contributions of the constructs in predicting the levels of the critical constructs. In (a), we build a directed connectivity graph by (i) using a state of the art statistical technique to perform variable selection, (ii) integrating the variable selection results to form the directed connectivity graph, and (iii) employing graph-theoretical concepts and a graph clustering technique to interpret the resulting network topology in a multi-resolution manner. In (b), based on the directed connectivity graph, multiple linear regression is performed to quantify relations between the critical and other constructs. As a case study, the protocol is applied to analyze opinion leading and seeking behaviors in online market communications environments. The directed connectivity relations revealed provide new ways of visualizing the web of influence directionalities among the constructs of interest, suggest new research directions to pursue, and aid decision making in marketing management. The proposed method provides a data-driven alternative to traditional confirmatory methods in analyzing relations among given constructs. Its flexibility enables the business researcher to broaden the scope of research he/she can fruitfully engage in. © 2011 Elsevier Ltd. All rights reserved.

Research Area(s)

  • Constructs, Exploratory data analysis, Partial correlation, Structural equation modeling, Survey data