brycewang-stanford Theory-Construction Method (amr-methods)
AMR publishes NO empirical data. There are no samples, no measures, no estimation, and no results. "Method" here means the craft of building theory: how you construct constructs, derive relationships, ground mechanisms, and bound the theory. If your project needs data to make its point, it belongs at AMJ / ASQ / SMJ.
When to trigger
- Constructs and propositions exist but the way they were built feels ad hoc
- A reviewer would ask "where did this construct's domain come from?"
- The theory needs a more disciplined construction approach than intuition
- You need to choose among theory-building styles for your puzzle
Choosing a theory-construction approach
| Approach | When it fits | Core moves |
|---|---|---|
| Conceptual elaboration | Extending a known theory to a new domain | Carry over the core logic; re-specify constructs for the new setting; add boundary conditions |
| Analogical / metaphorical transfer | A construct from another field illuminates a management phenomenon | Map source → target carefully; theorize where the analogy holds and breaks |
| Typology construction | The phenomenon has distinct, theoretically meaningful types | Derive dimensions from theory (not data); ensure types are mutually exclusive and jointly exhaustive; theorize transitions (e.g., Oliver's strategic-responses typology, AMR 1991, DOI 10.5465/amr.1991.4279002) |
| Process theorizing | The phenomenon is a sequence/becoming, not a variance relationship | Specify stages, triggers, feedback loops, and timing; theorize the engine that drives movement |
| Construct re-specification | An existing construct conflates distinct phenomena | Split or merge; re-define the domain; re-derive downstream relationships |
| Cross-level theorizing | The action spans individual ↔ collective | Specify emergence (bottom-up) and top-down effects; avoid level confusion |
Building each component well
- Constructs: a good construct is necessary (the theory cannot be stated without it), distinct (not a synonym), and has a clear domain and scope conditions (Suddaby's construct-clarity criteria, AMR 2010, DOI 10.5465/amr.2010.0419). Define what it excludes, not only what it includes. Construct domain is the conceptual analog of avoiding a contaminated sample — but done in argument, not data cleaning.
- Relationships: derive, don't assert. Each link should follow from the constructs and an explicit mechanism. Specify form (causal / recursive / moderating / mediating / constitutive).
- Mechanisms: name the engine — the why that makes theory explanatory rather than descriptive (Whetten, AMR 1989, DOI 10.5465/amr.1989.4308371). Prefer mechanisms with micro-foundations — actors doing things for stated reasons — over black-box "is associated with."
- Boundary conditions: state the contextual, temporal, and level limits. Strong theory specifies its own scope; this is contribution, not hedging.
- Falsifiability: design propositions to be refutable in principle (AMR's scope demands "testable knowledge-based claims"). This is the AMR analog of identification rigor at AMJ — credibility rests on tight, falsifiable logic, not on a clean estimate.
Checklist
- A named construction approach (or explicit hybrid) is used, not intuition alone
- Each construct is necessary, distinct, and domain-bounded
- Each relationship is derived from constructs + mechanism, not asserted
- Mechanisms are specified with micro-foundations where possible
- Typologies (if used) are mutually exclusive and jointly exhaustive, derived from theory
- Process theories specify stages, triggers, feedback, and timing
- Boundary conditions are stated as part of the theory
- No empirical apparatus has crept in (no samples, measures, tests)
Anti-patterns
- Treating "method" as data collection — there is none at AMR
- A typology whose dimensions came from convenience, not theory
- A process "model" that is really a static box-and-arrow diagram with no engine
- Borrowing a construct from another field without theorizing where the analogy fails
- Cross-level claims that confuse the level of theory with the level of the mechanism
- Asserting relationships and back-filling a mechanism afterward
Output format
【Construction approach】elaboration / analogy / typology / process / re-specification / cross-level
【Constructs built】[name + how its domain was derived]
【Mechanism basis】micro-foundations / structural / cognitive / ...
【Boundary conditions】contextual / temporal / level
【Empirical apparatus present?】must be: none
【Next step】amr-data-analysis (stress-test the logic)