Open Collections

Entrepreneurial posture in communications : entrepreneurial orientation (EO) analysis in financial communications with computer-assisted text analysis combined with EO custom dictionaries, natural language processing, and machine learning

University of British Columbia

Given the emergence of new automated text-based measurement and machine learning analytical techniques such as Natural Language Processing (NLP) and Topic Modelling, which serve as research analytical methodologies, there is growing interest in their use and application to business scenarios as well as theory development. Issues arise, however, as existing theory may not be properly reflected in such novel methods, such as computer-assisted text analysis (CATA), natural language processing, and topic modeling. This research joins one such conceptual debate regarding entrepreneurial orientation (EO), a psychometric evaluation of a firm’s behavioral nature related to risk-taking, innovativeness, proactiveness, autonomy, and competitive aggressiveness. It has been recently argued that entrepreneurial words, which form the text-based EO measurement, alone are not necessarily aligned with EO concepts as a behavioral construct (Covin & Wales, 2018). This research proposes a novel extension to EO as a messaging construct called entrepreneurial posture (EP) within an Entrepreneurial Posture in Communications (EPIC) framework, where the moderating influence between EO and Firm Performance (FP) is conceptualized as candor, the quality of being honest and straightforward in attitude and speech. Methodological segmentation by natural language processing (NLP) and machine learning topic modelling techniques is also explored in this research. Topic model segmentation identified communities with differing EO to FP profiles, which supports the theoretical view that firm communities differ in candor profiles. Results from this research demonstrates unique EP expression in firms’ communications with respect to numerous accounting and market measures and across time. The results also identified communities with differing EO to FP conceptual relationships, where specific candor differences were suggested. Conclusions drawn suggest that the combined assessment of EP and candor, using automated text-based analysis exclusively, provides predictive insights in keeping with strategic posture. Methodologically, the novel application of NLP and machine learning topic modelling presented in this research improved the proportion of explain variance in the EO to Firm Performance model from 0.04 to 0.21, an affect size substantially higher than typically presented in EO literature.

Text

2021-04-28

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/78022

Interdisciplinary Studies

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/