Jmgmt Theory Development

---
name: jmgmt-theory-development
description: Use when the theoretical argument is the bottleneck for a Journal of Management (JOM) manuscript — building constructs, mechanisms, boundary conditions, and a priori hypotheses (or a review's organizing framework). Constructs the theory; it does not run the analysis (jmgmt-data-analysis) or write the contribution paragraph (jmgmt-contribution-framing).
---

# Theory Development (jmgmt-theory-development)

## When to trigger

- Hypotheses read as bald predictions ("A is positively related to B") with no mechanism
- You have results and are tempted to write hypotheses around them (HARKing risk)
- The argument leans on one borrowed theory name (e.g., "social exchange") without using its logic
- Mediators/moderators sit in the data but are not theorized
- A meta-analysis or review needs an **organizing framework**, not just a coding scheme
- A reviewer says "the theory is thin," "underdeveloped," or "why would this be true?"

## The JOM theory bar

JOM publishes empirical, theoretical, *and* review work, but in every case the **theoretical contribution is the gate**. For an empirical paper, each hypothesis must be **derived from an articulated mechanism, before the results are known**, and a reader should finish the front end understanding *why* the effect occurs and *under what conditions* it strengthens or reverses. For a **meta-analysis**, the synthesis must be organized by theory so that moderators *adjudicate* competing accounts — JOM does not reward a coefficient census. For a **systematic review** (Review Issue), the deliverable is an integrative framework plus a forward agenda; a chronological catalogue is desk-rejected.

## Building a hypothesis (the mechanism chain)

For each focal hypothesis, write the explicit chain — do not skip steps:

1. **Antecedent** — the predictor and why it matters in this context.
2. **Mechanism** — the theoretical process linking antecedent to outcome (social exchange, signaling, attention allocation, identity threat, resource dependence, conservation of resources). Name the theory *and* the process verb (motivates, signals, depletes, legitimates).
3. **Outcome** — the dependent construct, defined at a stated level of analysis.
4. **Direction & form** — sign, and whether linear, curvilinear, or threshold (justify a turning point with two opposing forces, not a significant squared term).
5. **Boundary** — the moderator(s) and the *substantive reason* the mechanism is stronger/weaker.

State the level of analysis explicitly (individual, team, firm) and keep the theory at the level where the mechanism operates; flag cross-level mediation as such. Draw a theoretical-model figure (boxes and arrows) that maps **one-to-one** to the hypotheses.

## Constructs earn their keep

JOM (and its sibling research-methods reviewers) is unusually alert to **construct proliferation and jangle/jingle problems**. Before adding a construct, show: a clear conceptual definition, its dimensionality, where it sits in the nomological net, and how it is *distinct* from neighboring constructs already in the literature. A new construct with no discriminant story is a liability, not a contribution.

## Mediation, moderation, and rival theories

- **Mediation**: theorize the process construct *before* testing it; a significant indirect effect is evidence *for* a theorized mechanism, never a substitute for the theory.
- **Moderation**: "it depends" is not a theory until you say *why* it depends.
- **Rival explanations**: name at least one competing theoretical account and explain how your design or moderators distinguish it — JOM reviewers push hard on alternatives.

## Theory for a review or meta-analysis

- **Meta-analysis**: state the theoretical question the synthesis answers; pre-specify moderators that map to *competing* theoretical predictions, so the meta-analytic interaction is a theory test.
- **Review**: build an organizing framework (a typology, a process model, an integrative 2×2) that imposes structure on the literature and surfaces the unanswered questions that become the agenda.

## Worked vignette (illustrative)

Consider a claim that CEO humility raises firm innovation. A bald version states the sign and cites a humility paper. A JOM-grade version writes the chain: **antecedent** (CEO humility) → **mechanism** (humility lowers status threat among the top management team, so members voice dissenting ideas — name the process: *legitimates* upward voice) → **outcome** (innovation, at the firm level) → **form** (positive, possibly attenuating at very high levels as decisiveness suffers) → **boundary** (the effect is stronger under environmental dynamism, because dissent is more valuable when the future is uncertain). Now mediation (TMT voice) and moderation (dynamism) are theorized a priori, the level shifts (individual trait → firm outcome) are explicit, and a rival account (humility is just agreeableness) can be named and addressed — the difference between a desk reject and a contribution.

## Checklist

- [ ] Every hypothesis has an explicit mechanism, not just a predicted sign
- [ ] The driving theory is named and its *logic* used (not a one-line citation)
- [ ] Level of analysis stated and consistent across argument and design
- [ ] New constructs have a definition, dimensionality, and a discriminant-validity story
- [ ] Mediators/moderators theorized a priori with substantive reasons
- [ ] Hypotheses fixed before analyzing the data (no HARKing)
- [ ] A theoretical-model figure matches the hypotheses one-to-one
- [ ] At least one rival explanation named and engaged
- [ ] (Review/meta) an organizing framework / theory-mapped moderators exist

## Anti-patterns

- **HARKing**: hypothesizing after results are known — reviewers detect it when every hypothesis is supported and none is surprising
- **Bald predictions**: "A is positively related to B" with no *why*
- **Borrowed theory**: invoking a famous theory by name without using its actual logic
- **Mechanism by mediation**: claiming a process exists only because the indirect effect is significant
- **Construct proliferation**: a "new" construct that is an old one relabeled, with no discriminant evidence
- **Catalogue review**: a literature summary with no framework and no forward agenda

## Output format

```
【Article type】empirical / meta-analysis / review
【Focal theory】... (process verb: ...)
【H1 (focal effect)】antecedent → mechanism → outcome; direction/form; level
【H2 (mechanism/mediation)】...
【H3+ (boundary/moderation)】reason the slope changes
【Construct check】new construct? discriminant story: ...
【Rival explanation addressed】...
【HARKing check】hypotheses fixed before analysis? yes/no
【Model figure / framework】matches hypotheses one-to-one? yes/no
【Next step】jmgmt-literature-positioning
```