Drivers of Corporate Digital Responsibility: A Machine Learning Approach

Published: 10 March 2026| Version 1 | DOI: 10.17632/77wmdwxsmn.1
Contributors:
, Heba Ali,
,

Description

The study uses a sample that comprises leading companies listed in prominent stock indices from English-speaking countries, specifically: the FTSE 350 for the United Kingdom, the S&P 500 for the United States of America, the S&P ASX 200 for Australia, and the TSX 300 for Canada, over the period from January 12, 2010, to May 2, 2023. Column Definition Code Code Company ID Company ID Industry Industry Group Year Year Index Market Index Country Market Twitter Account (Dummy) Dummy variable; 1 if company has a twitter Account, and 0 otherwise. Tot Posts Total Number of posts Tot Posts Dummy Dummy variable; 1 if company has posted on twitter, and 0 otherwise. Ln Tot Posts Natural logarithm of number of total posts. CDR Posts Total Number of CDR posts CDR Dummy Dummy variable; 1 if company has CDR posts, and 0 otherwise. Ln CDR Ln CDR: Natural logarithm of CDR-related posts. CDR Index CDR Index: ratio of CDR-related posts to total posts ROA Net income to total assets TA Total assets TL Total liabilities CR Current assets to current liabilities MV market capitalization Ln MV The Natural logarithm of market capitalization Lev (TL/TA) Total liabilities to total assets ESG Score Overall company score based on verifiable reported information in the environmental, social, and governance pillars. Ranges from 0 to 100. Env Score Environmental pillar score, Includes emissions, innovation, resource use categories. Ranges from 0 to 100. Social Score Social pillar score, Includes community, human rights, product responsibility, workforce categories. Ranges from 0 to 100. Gov Score Governance pillar score, Includes CSR strategy, management, shareholders categories. Ranges from 0 to 100. Global Innovation Index Score (0-100), Global Innovation Index Report Global Innovation Index (Rank) Rank, Global Innovation Index Report

Files

Steps to reproduce

To construct a reliable measure for CDR disclosure (dependent variable), this study focuses on self-reported information on CDR provided by firms on their official X accounts. Constructing a measure for CDR disclosure requires a systematic multi-step approach to ensure accuracy and comprehensiveness. The following steps outline the process employed throughout this research. The initial phase of the research involves identifying and verifying the companies' official X accounts. First, companies’ official websites are examined for a direct X link on their home pages. If a link is not visible on the homepage, the website’s internal search bar is utilized to search for keywords that might lead to an X account, including “X”, “connect”, “media”, “follow us”, “stay connected” and “social”. If a single X account is identified, it is included in the sample. In cases where multiple X accounts are linked to a company’s website, the primary or general account is selected for inclusion in the sample. For companies whose websites do not provide any indication of an X account, a direct search is conducted on X using the company’s name. Any identified accounts are verified by the presence of X’s “blue checkmark” or by checking for a link that redirects to the company’s official website. Again, when multiple X accounts are found for a single company, the main account is selected for the sample. By following this structured methodology, the study ensures that the X accounts identified are accurate representations of the companies’ official social media presence.

Categories

Corporate Communication

Licence