Research Design, Big-Tent (jm-methods)

When to trigger

• The question is set and you must choose a design that can actually answer it
• A reviewer will ask whether the method supports a causal or managerial claim
• You are deciding between a clean lab experiment and a messier but realer field study
• You have secondary (scanner/CRM/financial) data and need an identification strategy

JM's "big tent" — let the question pick the method

JM is methodologically pluralistic: it welcomes primary data (experiments, field studies, surveys, interviews, observational data) and secondary data, and champions empirics-first research grounded in real-world phenomena. No single method is privileged. The design rule at JM is therefore: choose the method that most credibly answers a substantive question and supports a managerially relevant claim — not the most sophisticated technique. Work centered on mathematical/statistical methods for their own sake is out of scope (route to Marketing Science / JMR); methods here are servants of the substantive insight.

Match design to claim

Combine methods (multi-study or mixed) when one design cannot establish both internal validity (the effect is real) and external/managerial validity (it matters in the market).

Field realism and managerial validity

JM prizes evidence that travels to real decisions. Strengthen the design by: securing a field setting or firm partner where feasible; choosing outcomes managers act on (sales, CLV, conversion, welfare) over proxy attitudes alone; sampling a population the claim should generalize to; and documenting the real-world stimulus, market, and time frame so a practitioner recognizes the setting.

Design for transparency up front

JM requires a replication packet at conditional acceptance and encourages preregistration. Build this in now: preregister experiments (you will later supply anonymized links and an attestation), version-control analysis scripts, and log sample-construction and exclusion rules as you go — not retroactively.

Execution bridge (StatsPAI / Stata MCP)

For the empirical / causal lane, estimate and audit rather than only specify. Full map: execution-with-mcp. JM is empirical marketing — field experiments, panel/CRM data, and quasi-experiments; randomization inference for experiments, DiD / IV for observational claims.

• detect_design → recommend → fit with as_handle=true → audit_result to enumerate the checks the design owes.
• Panel / staggered DiD: callaway_santanna / sun_abraham + bacon_decomposition
  • honest_did_from_result. IV: effective_f_test + anderson_rubin_ci. RDD: rdrobust + mccrary_test.
• Experiments: randomization-based inference and romano_wolf for the many-outcome family-wise correction reviewers expect.

Match the toolchain to the reviewer pool, and report the effect size the venue wants. A run end-to-end (synthetic data, real returns) is in the JF execution walkthrough.

Checklist

• Method chosen to fit the substantive claim, not for sophistication
• Internal validity (randomization/identification) addressed
• External/managerial validity (field realism, actionable outcomes) addressed
• Multi-study / mixed design where one method cannot do both
• Outcomes managers/policy makers care about are measured
• Preregistration planned (experiments); scripts and exclusion rules logged
• Power/sample-size justified a priori for experiments

Anti-patterns

• Method-driven paper: a clever estimator in search of a question (out of scope at JM).
• Lab-only causal claim asserted to hold in the market with no field/external evidence.
• Endogenous treatment in secondary data with no identification strategy.
• Attitude proxies standing in for outcomes managers actually move.
• Retrofitted transparency: no preregistration, exclusions documented after the fact.

Output format

【Substantive claim】[...]
【Design】experiment / field experiment / survey-conjoint / panel-DiD / qualitative / mixed
【Internal validity】randomization / identification: [...]
【External & managerial validity】field realism, actionable outcomes: [...]
【Multi-study plan】[...]
【Transparency】preregistration + script/exclusion logging: [...]
【Next step】jm-data-analysis