Description
A two-step argument that generalises findings from a small qualitative study to concepts and theory, and then from those concepts to other similar situations, instead of to a population.
Yin argues that every study, including a laboratory experiment, is a single study, and that the way findings travel beyond it is usually not statistical. Analytic generalisation is a two-step process. First, the researcher makes a conceptual claim: the findings are shown to inform a particular set of concepts, theoretical constructs or a hypothesised sequence of events, ideally ones already present in the literature. Second, the same concepts are applied to argue that other similar situations are implicated. The claim is an argument, not a proof, so it has to be built carefully and be resistant to logical challenge, and it is much stronger when plausible rival explanations have been actively tested and rejected. Yin distinguishes it from transferability, which makes a more modest claim that depends on how similar the sending and receiving contexts are. For UX teams this addresses the recurring objection that eight interviews cannot generalise: the answer is to state which mechanism or concept the study supports or challenges, show the evidence, and say which other products, segments or flows share that mechanism, rather than implying population percentages.
Attributes
- Discipline
- Research Design & Rigor
- Loss type
- Siloed
- Stage
- Discovery, Definition
- Data
- Qualitative
- Effort
- Days
Steps
- Before fieldwork, name the concepts, theories or hypothesised sequence of events the study could inform
- Identify plausible rival explanations and plan to collect evidence that could support them
- After analysis, state whether and how the findings supported or challenged each concept, citing the evidence
- Test and document the rivals; reject them only where diligent search found no supporting evidence
- Raise the claim to the conceptual level and write it as a mechanism, not a frequency
- Name the other situations (products, segments, flows) that share the mechanism and argue why the concept applies there
- Treat the generalised claim as a working hypothesis to be checked by later studies
Sources
- Yin, R., Qualitative Research from Start to Finish, 2nd ed. (2016)
Also known as
- Analytic generalization
- Theoretical generalisation
- Generalising to theory, not population
Categories
- Research Operations
- Research Design & Rigor