Generalization and Induction: Misconceptions, Clarifications, and a Classification of Induction
In “Generalizing Generalizability in Information Systems Research,” Lee and Baskerville (2003) try to clarify generalization and classify it into four types. Unfortunately, their account is problematic. We propose repairs. Central among these is our balance-of-evidence argument that we should adopt the view that Hume’s problem of induction has a solution, even if we do not know what it is. We build upon this by proposing an alternative classification of induction. There are five types of generalization: (1) theoretical, (2) within-population, (3) cross-population, (4) contextual, and (5) temporal, with theoretical generalization being across the empirical and theoretical levels and the rest within the empirical level. Our classification also includes two kinds of inductive reasoning that do not belong to the domain of generalization. We then discuss the implications of our classification for information systems research.
|Author||Eric W. K. Tsang and John N. Williams|
|Keywords||Research methodology, generalization, generalizability, induction, deduction, statistical generalization, statistical syllogism, inductive analogy, Hume’s problem of induction|