Measuring the Ethical Awareness of Corporate Data Scientists

• Main Author: Weaver, Jessica L
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01.01.2022
• Subjects: Decision making; Employees; Ethics; Innovations; Leadership; Management; Morality; Occupational psychology; Organizational behavior
• ISBN: 9798358485723

Data is the new oil (Hand, 2018b) and yet data and the science of leveraging it to learn new insights, AKA data science, remain largely unregulated in the U.S. The U.S. Bureau of Labor Statistics (2022) predicts the occupation of data science to grow by 22% between 2020 and 2030. However, data-enabled organizations have experienced backlash and public outcry with their mishandling of information, e.g., Google Flu, Target scandal, Facebook scraping, algorithm biases in hiring and criminal justice. Academia is trailing behind, creating more data science programs focused on the tech, but lacking in human subject research and ethics training. The purpose of this quantitative study is to build and validate a psychometrically sound ethical awareness scale for corporate data scientists. The key population are data scientists and practitioners that work in organizations that do not employ an Institutional Review Board (IRB) like process to oversee data research projects. A panel of 3 data and ethics SMEs reviewed the initial scale items, ensuring appropriateness of the measures, i.e., face validity. A sample of 569 data scientists were surveyed via online crowd sourcing tools, using Yurtsever’s Moral Imagination Scale and the Corporate Data Science Ethical Awareness Scale (CDSEAS). The CDSEAS demonstrated reliability, with a Cronbach’s alpha > 0.7, and discriminant validity, as a separate construct from Moral Imagination, however the dimensionality of this instrument requires further research. The proposed moderating variable of Contextual Knowledge was tested to see if it can influence the relationship between a data scientist’s moral imagination and ethical awareness in data science. While this study failed to reject the null hypothesis regarding the moderating variable, it did provide a reliable scale to measure ethical awareness and key insights for future research assessing this target population.