This cumulative dissertation examines the impact of predictive analytics on organizational planning and decision-making. It focuses on the factors that encourage or discourage managers from adopting respective analytics tools, and on management accounting strategies to guide their use. Three studies combine theoretical reasoning, qualitative interviews and behavioral experiments to address these issues. The first study explores the adoption of a human resource analytics tool by potential users through interviews. It identifies beliefs that shape perceived behavioral control, attitudes and social norms, linking them to machine learning features. The second study proposes a framework for 'algorithm aversion', defined as the tendency to discount algorithmic advice more than comparable human advice. Based on the Theory of Planned Behavior, it demonstrates that algorithm aversion stems from a combination of unfavorable attitudes towards algorithmic judgement and lower perceived behavioral control when using algorithms compared to relying on human judgement. The third study examines how management accounting can influence the perceived credibility of algorithmically generated sales forecasts. In a laboratory experiment, capability-based metrics consistently increase perceived credibility, even when unwarranted, whereas predictability-based metrics align credibility more closely with the actual accuracy of the forecast. In summary, managers’ use of predictive analytics is shaped by their attitudes, their perceived behavioral control, social norms and the characteristics of the underlying algorithms, which can either foster or hinder adoption. Management accounting can influence this by carefully designing communication, such as performance metrics, to mitigate algorithm aversion and calibrate the perceived credibility of forecasts.
Christian Ertel (Wed,) studied this question.