Jom methods

Research Design for Empirical OM (jom-methods)

When to trigger

• You must pick a design that can actually test your operations hypotheses
• A reviewer questions whether the method fits the operations phenomenon or the level of analysis
• You are weighing a behavioral lab experiment vs. archival panel vs. survey vs. field/intervention
• You anticipate the Empirical Research Methods Department method check on incoming submissions

JOM's empirical-only design space

JOM publishes empirical OM research and does not publish purely analytical models or optimization techniques. Every design must rest on observation. The native JOM design families are:

Operations question / claim	Design
Perceptions, practices, constructs across firms/plants	Survey (validated multi-item scales; multi-respondent)
Cause–effect on operational outcomes from secondary data	Archival/secondary panel (recalls, inventory, supply ties)
Human operational decisions, biases, incentives	Behavioral-OM experiment (lab/online, manipulation checks)
How operations work unfolds in context	Field study / case study (process, embedded, comparative)
Effect of an actively introduced change in a real setting	Intervention Based Research (engaged scholarship)
Pooling effects across studies	Meta-analysis of empirical OM findings

Match design to claim and level

State the unit and level (transaction, shift, line, plant, project, dyad, supply network) and ensure the design observes at that level. A plant-level claim tested with firm-level archival data is a level mismatch reviewers catch quickly. For multi-respondent supply-chain dyads, plan both-side data collection.

Intervention Based Research (a JOM-distinctive genre)

JOM formally houses Intervention Based Research, where researchers actively intervene in a real operational problem (engaged scholarship). If you use it: pre-state the theorized effect, document the intervention protocol and timeline, separate researcher actions from observed effects, and address how engagement threatens (and how you protect) inference. Most flagship management journals do not formally house this genre — use it deliberately.

Pre-empt the Empirical Research Methods check

JOM's Empirical Research Methods Department performs method checks on incoming empirical submissions. Defend, up front: sampling frame and response/coverage, construct operationalization, identification strategy (for causal claims), common-method separations (for survey), and reproducibility of secondary-data construction. Weak identification or unvalidated measures stall here.

Execution bridge (StatsPAI / Stata MCP)

For the empirical / causal lane, estimate and audit rather than only specify. Full map: execution-with-mcp. JOM is empirical operations / supply-chain — survey and archival panel data; foreground endogeneity of operational choices and clustered / multilevel inference.

• 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.

Anti-patterns

• An optimization or pure simulation model presented as the empirical contribution.
• Design that cannot observe at the claimed level.
• Single-respondent survey for a relational/dyadic operations claim.
• Intervention work that conflates the researcher's actions with the measured effect.

Design-defense checklist the methods check applies

Each JOM design family carries its own non-negotiable evidence; the mapping below summarizes what the Empirical Research Methods check typically scrutinizes. Confirm exact expectations against current methods-department guidance.

Design	Defense expected up front	Fatal omission
Survey	Sampling frame, coverage, validated scales, CMB separations	Single-respondent data for a dyadic claim
Archival/secondary	Identification strategy, reproducible construction, clustering	Endogenous practice treated as exogenous
Behavioral-OM experiment	Manipulation/attention checks, random assignment, task realism	Confound between manipulation and a cue
Field / case study	Sampling logic, traceable data structure, audit trail	Anecdote substituted for within-case evidence
Intervention Based Research	Pre-stated effect, protocol/timeline, action–effect separation	Researcher actions conflated with the outcome

Desk-reject and return triggers on design

• An optimization or pure simulation model presented as the empirical contribution (redirected to analytical OM venues).
• A line-level claim tested only with firm-level archival data — a level mismatch.
• Survey design with no procedural remedies for common-method bias and no measurement validation plan.

Worked vignette: archival vs. field for a quality-practice question

Does cross-training frontline operators reduce defects? Archival route: a plant panel with HR cross-training records and quality logs, identified via a phased rollout (illustrative). Field route: an intervention introducing cross-training in matched lines with pre/post measurement. The choice turns on the claim and the threat. If selection is the worry and a staggered rollout exists, the archival quasi-experiment identifies the effect at the line level with plant and time fixed effects. If the mechanism (how cross-training changes floor problem-solving) is the contribution, Intervention Based Research observes the process and pre-states the effect. Either is JOM-fit; an analytical model of optimal allocation is not.

Design objections and the JOM-grounded response

• "The method does not fit the operations phenomenon." Re-state the unit and level, then show the design observes at exactly that level.
• "This is analytical work, not empirical." If the core contribution is a model, redirect to an analytical venue; if observed, foreground the observation.

Output format

【Design family】survey / archival / experiment / field-case / IBR / meta-analysis
【Unit & level】...
【Identification / validity strategy】...
【Methods-check readiness】sampling, measures, identification, reproducibility ...
【Next step】jom-data-analysis