The Field Guide › Paper
Introduces 'methodological fit', internal consistency among four elements of a field research project (research question, prior work, research design, contribution to the literature), as a criterion for research quality that experienced field researchers had long practised implicitly but that the field had never made explicit. The central organising move is a continuum of theory maturity running from nascent (little or no prior theorising, genuinely novel phenomena) through intermediate (some established constructs, provisional relationships) to mature (well-specified constructs studied precisely across many settings), with each stage calling for a different research design: nascent theory calls for open-ended qualitative work aimed at a 'suggestive theory' that invites further research; intermediate theory calls for a hybrid of qualitative and quantitative data aimed at a 'provisional theory,' often integrating two previously separate literatures; mature theory calls for precise, quantitative hypothesis testing aimed at adding new specificity, mechanisms, or boundary conditions to an existing model. The paper's most striking exemplar for the intermediate case is Edmondson's own 1999 paper introducing team psychological safety, presented here as a worked example of the framework in action rather than merely cited: it married organisational-learning theory with team-effectiveness theory across three explicit phases, qualitative interviews with eight teams to shape new survey measures; a quantitative survey of 496 members across 53 teams, with team-design variables independently rated via blind manager interviews specifically to reduce common-method bias; and a final qualitative extreme-case comparison of high- and low-learning teams to explain what the numbers meant 'behind the numbers.' Barker's (1993) 'Tightening the Iron Cage,' an ethnographic study of self-managed teams that took two years of fieldwork to discover that peer-based concertive control could become more psychologically restrictive than the bureaucratic hierarchy it replaced, anchors the nascent-theory case and is reused later to illustrate a genuinely useful second idea, 'off-diagonal' fit. Self-managed teams were already a mature research area when Barker began, but he didn't ask the well-trodden question of what makes them effective; he asked how members cope with the social pressure of policing each other, a genuinely unexplored question inside an otherwise mature topic, which is what justified starting fresh with open-ended qualitative data despite the broader literature's maturity. The paper catalogues six specific ways fit goes wrong, one for each combination of theory stage and mismatched method, with real, sometimes self-implicating examples: in mature areas, qualitative-only work risks 'reinventing the wheel' by rediscovering already-known factors, while mixing in qualitative anecdotes as if they were evidence produces the 'uneven status of evidence' problem, illustrated by the authors' own account of submitting a mixed-methods paper on tacit knowledge and learning curves, initially reacting defensively to reviewers who pushed back on the qualitative material, then realising on reflection that removing it strengthened the paper considerably. In nascent areas, quantitative or hybrid designs risk 'fishing expeditions' and measures with an 'uncertain relationship to phenomena' that isn't yet understood well enough to measure validly, illustrated by another candid admission: an early Edmondson paper attempting to integrate qualitative and quantitative data on how teams learn a new technology had technically deficient, unvalidated quantitative measures that were eventually cut entirely, becoming the purely qualitative Edmondson, Bohmer and Pisano (2001) 'Disrupted Routines' paper already in this corpus. Models the research process itself as a funnel, borrowed deliberately from the product-development literature: choices are wide open before any data is collected and narrow progressively, so most of the productive iteration should happen early, in refining the question and design, since course-correction gets steadily more expensive once data collection is underway. Closes with implications for training new field researchers: methodological versatility over attachment to a single preferred toolkit, learning fit by deconstructing published exemplars, drafting research proposals for structured peer critique (explicitly noting this requires a climate of psychological safety, since sharing an early, unfinished research design with a room of peers is itself a genuine interpersonal risk), and apprenticeship alongside experienced researchers.