ALGORITHMIC INDUCTION THROUGH MACHINE LEARNING: USING PREDICTIONS TO THEORIZE

Machine learning (ML) algorithms are rapidly advancing research across many fields of social science, including economics, marketing, and management information systems. Management and organization studies are yet to fully leverage these methods beyond application to coding unstructured data. This may be in part due to the distaste for “predictions without causal explanations” that ML algorithms are known to produce. Yet, we argue, precisely because of this property, ML techniques can be extremely useful in theory construction by separating the two key components of inductive theorizing- pattern detection and pattern explanation. ML can facilitate “algorithmic induction”— formally specifiable operations aiding inductive inference that yields identical (or highly similar) conclusions when applied by different observers to the same data. We propose that algorithmic induction is valuable for researchers interested in using quantitative data to both develop and test theories in a transparent and reproducible manner.