Research Dataset, Sustainability Dictionaries, and Word2Vec Model for "Are We Practicing What We Preach? A Balanced Sustainability Pyramid Assessment of Business and Management Literature"

Abstract

This repository contains the research materials developed for the manuscript “Are We Practicing What We Preach? A Balanced Sustainability Pyramid Assessment of Business and Management Literature,” submitted to the Schmalenbach Journal of Business Research. It includes the dataset of 759 publications from journals included in the 2016 Financial Times Top 50 (FT50) journal list, the complete set of 22 sustainability dictionaries, and the trained Word2Vec model used in the analysis. These resources support the hybrid review design presented in the manuscript and its computer-aided, human-in-the-loop analysis of sustainability discourse in business and management research. They operationalize the study’s conceptual framework—the Balanced Sustainability Spinner (BSS) and Balanced Sustainability Pyramid—across the environmental, social, economic, and governance and strategy pillars, the 17 United Nations Sustainable Development Goals (SDGs), and an overarching sustainability construct. The repository is provided to enhance transparency and reproducibility and to facilitate the reuse and further development of the study’s analytical resources. Methodology and Repository Contents Research Dataset: The dataset contains the 759 publications included in the final sample and retrieved from the Scopus database. Publications were identified according to the systematic review and eligibility criteria described in the manuscript, including the requirement that “sustain*” appear in the article title. Sustainability Dictionaries: The 22 dictionaries cover the four BSS pillars, the 17 SDGs, and an overarching sustainability construct. They were developed iteratively by combining deductive and inductive approaches. Initial terms were derived from established frameworks and prior literature, including GRI, ASSET4, SASB, and the SDGs. The dictionaries were subsequently expanded through frequency analysis, Word2Vec-based cosine-similarity analysis, and generative AI-assisted identification of related terms and expressions. Candidate terms were then qualitatively contextualized, validated, refined, reassigned, or removed through researcher-led keyword-in-context analysis. Word2Vec Model: The repository includes the trained Word2Vec model developed from the study corpus using the Gensim Python package. The final model uses 100-dimensional word embeddings with a context window of four words. It was used to examine conceptual similarities among sustainability-related terms and to support the inductive expansion of the sustainability dictionaries.