From mgsci-skills
Guides selection and defense of analytical or empirical methods for Management Science manuscripts, matching the question to the appropriate Department standard.
How this skill is triggered — by the user, by Claude, or both
Slash command
/mgsci-skills:mgsci-methodsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
- The method may not match the question (wrong model class, weak identification)
Management Science has no single dominant method by design. Each Department sets its own field-appropriate expectations. Pick the lane the question demands, then meet that lane's rigor bar.
| Question / claim | Approach |
|---|---|
| Optimal policy under constraints | Mathematical programming / optimization (LP/MIP/convex) |
| Dynamics, queues, inventory, uncertainty | Stochastic processes, MDPs, dynamic programming |
| Strategic interaction among decision-makers | Game theory / mechanism design; state the equilibrium concept |
| Pricing/incentives under information frictions | Economic-theory model; contracts, signaling, screening |
| Intractable models / policy evaluation | Simulation with variance reduction and honest CIs |
For analytical work the design is the model: define the decision problem, assumptions, solution concept, and what you will prove. Plan the comparative statics that will carry the managerial insight, and the extensions that will demonstrate robustness of the qualitative result.
| Question / claim | Design |
|---|---|
| Causal effect from observational data | Quasi-experiment: DiD, IV, RDD, event study; clustered SE |
| Causal effect under control | Lab or field experiment with randomization & manipulation checks |
| Mechanism / primitive recovery | Structural estimation tied to a model |
| Prediction / pattern at scale | Data-science pipeline with honest out-of-sample validation |
For empirical work, the identification strategy is the heart of the design — name the source of variation and the threats it rules out before estimation. Behavioral and Marketing experiments need pre-registration, manipulation/attention checks, and adequate power.
Whatever the lane, the design must (a) be able to support the specific claim and (b) yield a decision-relevant insight that travels across departments. A method that is rigorous but produces no managerial reading, or a managerial story the design cannot actually test, will not clear the bar.
Management Science verifies data and code for accepted papers before publication (Data and Code Disclosure Policy, articles on/after July 1, 2019). Design your pipeline — analytical numerics or empirical estimation — so a single master script regenerates every result from raw inputs.
Because Management Science is the multidisciplinary INFORMS flagship of many departments, "the rigor bar" is the bar of the department whose editor desk-screens you.
| Candidate department | The standard it enforces |
|---|---|
| Optimization and Decision Analytics | Provable structure or bounds, not just a solver run |
| Stochastic Models and Simulation | Honest confidence intervals; analytical structure where possible |
| Behavioral and Decision Analysis | Pre-registration, manipulation/attention checks, power |
| Finance / Accounting (empirical) | Credible exogenous variation; clustered SE |
A team wants to claim algorithmic shift-recommendations raise gig-driver retention, aiming at the Operations Management / Behavioral lane. Design A compares drivers who opted in versus those who did not — opt-in is self-selected, so a referee reads the gap as selection, not effect. Design B randomizes recommendations on for 50% of new drivers, powered for a 4-point lift (illustrative), and pre-commits market-clustered errors. Design B clears the bar by ruling out selection by construction; Design A is desk-vulnerable on identification despite the data existing.
【Lane】analytical / empirical / combined
【Approach】[model class or identification strategy]
【Why it fits the claim】...
【Department standard met】...
【Reproducibility plan】master script regenerates all results: yes/no
【Next step】mgsci-data-analysis
npx claudepluginhub brycewang-stanford/awesome-journal-skills --plugin mgsci-skillsExecutes and reports analysis for Management Science manuscripts: proves analytical results or estimates/validates empirical models, with replication package preparation.
Guides selection and defense of research designs for Organization Science manuscripts, matching qualitative, quantitative, experimental, or simulation methods to the research question and level of analysis. Addresses reviewer demands for causal inference.
Selects the appropriate analytical model class or empirical identification strategy for M&SOM manuscripts based on the operational decision. Useful when matching method to operations problem.