Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› jmgmt-methods
jmgmt-methods
GitHub针对《Journal of Management》稿件，解决研究设计与测量瓶颈。匹配理论与设计，处理共同方法偏差、内生性、构念效度及多层结构，提供元分析编码方案，确保研究严谨性以符合期刊标准。
Trigger Scenarios
研究设计与理论主张（水平、时序、因果）不匹配
数据存在单一来源或自报风险，需防范共同方法偏差
理论为因果但设计仅为横截面/相关分析
构念缺乏已验证的测量工具
需要进行元分析的编码协议与误差校正
审稿人质疑设计无法检验假设或未解决内生性
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill jmgmt-methods -g -y
SKILL.md
Frontmatter
{
"name": "jmgmt-methods",
"description": "Use when research design and measurement are the bottleneck for a Journal of Management (JOM) manuscript — matching design to the theoretical claim, construct validity, common-method bias, endogeneity, multilevel structure, and (for meta-analyses) coding\/artifact corrections. Designs the study; it does not run the estimation (jmgmt-data-analysis)."
}
Research Design & Methods (jmgmt-methods)
When to trigger
- The design may not match the theory's level, timing, or causal claim
- Data are single-source, single-wave, self-reported (common-method bias risk)
- The theory is causal but the design is cross-sectional/correlational
- Constructs lack established, validated measures
- A meta-analysis needs a defensible coding protocol and artifact-correction plan
- A reviewer says "the design cannot test this hypothesis" or "endogeneity is unaddressed"
Match the design to the claim
JOM welcomes all empirical methods — survey, experiment, archival panel, multilevel field study, qualitative, and meta-analysis — and judges on fit and rigor, not a preferred method. JOM's research-methods identity (it explicitly covers research methods and runs methods-focused reviews) means design choices are scrutinized closely.
| Theoretical claim | Design that earns it |
|---|---|
| Causal effect of a manipulable cause | Experiment (lab/online/field), or natural experiment |
| Process unfolding over time | Multi-wave panel; longitudinal/lagged design |
| Firm/strategy outcome from archival cause | Panel archival with fixed effects + an endogeneity strategy |
| Cross-level mechanism (team→individual) | Multilevel/nested data analyzed with HLM, not OLS |
| Synthesis across a literature | Meta-analysis with a pre-registered coding protocol |
A two-study design (field study for generalizability + experiment for the causal mechanism) is a recognized JOM strength — it buys internal and external validity at once.
Designing against the threats JOM referees punish
- Common-method bias (CMB): separate the sources of predictor and outcome; separate them temporally across waves; use objective/archival outcomes where possible. Procedural remedies beat statistical fixes (the Podsakoff et al. guidance is the field reference). Plan this before collecting data; a Harman single-factor test alone will not satisfy a JOM reviewer.
- Endogeneity (archival/macro): anticipate omitted variables, reverse causality, and selection. Specify an identification strategy — instrument, natural experiment, panel fixed effects, difference-in-differences, Heckman/2SLS, propensity matching — and state the assumptions each requires.
- Measurement / construct validity: use validated multi-item scales; pilot new measures; plan a confirmatory factor analysis (CFA) with fit indices and a discriminant-validity test (AVE vs. shared variance, or the HTMT ratio). State the level at which each construct is measured.
- Multilevel discipline: if data are nested, justify aggregation with ICC(1), ICC(2), and r_wg; model the nesting (random effects/HLM). Theorizing at the team level but running OLS on disaggregated individuals is a standard rejection trigger.
- Sampling & power: justify the frame, response rate, and statistical power — especially for interactions, which JOM reviewers know are underpowered when authors present null moderation as a "boundary condition."
Meta-analysis design
- Pre-specify inclusion/exclusion criteria and a transparent search; report a PRISMA-style flow.
- Double-code a subset; report inter-coder agreement.
- Choose the artifact-correction model (Hunter–Schmidt psychometric meta-analysis vs. Hedges–Olkin random-effects) and justify it; correct for sampling error and, where defensible, measurement unreliability and range restriction.
- Plan moderator/meta-regression analyses that map to competing theories, plus publication-bias diagnostics.
Referee pushback mapped to the design fix
- "This is single-source, single-wave — common-method bias is unaddressed." → Add temporal/source separation or an objective outcome; a Harman test alone will not close it.
- "Your archival regressor is endogenous." → Specify and defend an identification strategy (IV/NE/FE/DiD/matching) and report first-stage strength.
- "The new scale's discriminant validity is unestablished." → Report a CFA with AVE vs. shared variance or an HTMT ratio, plus an alternative-model comparison.
- "You theorize at the team level but test individuals." → Justify aggregation (ICC, r_wg) and model the nesting, or re-pitch the theory at the individual level.
- "The interaction is underpowered." → Report power for the interaction specifically; if it is a true null, theorize the boundary rather than presenting an underpowered null as a finding.
Designing a multi-study program
JOM rewards study programs that triangulate rather than merely accumulate. A canonical pairing is a field study (external validity, real outcomes) plus an experiment (causal mechanism, manipulation of the antecedent). Decide what each study is for — generalizability, causal identification, or mechanism evidence — and make sure together they license the central claim. A second study that merely re-runs the first in a new sample adds length without adding inferential leverage, and the 50-page limit punishes it.
Execution bridge (StatsPAI / Stata MCP)
For the empirical / causal lane, estimate and audit rather than only specify. Full
map: execution-with-mcp. Journal of Management covers empirical management broadly (including meta-analysis); the chain below serves primary causal / panel work.
detect_design→recommend→ fit withas_handle=true→audit_resultto enumerate the checks the design owes.- Panel / staggered DiD:
callaway_santanna/sun_abraham+bacon_decompositionhonest_did_from_result. IV:effective_f_test+anderson_rubin_ci. RDD:rdrobust+mccrary_test.
- Experiments: randomization-based inference and
romano_wolffor 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
- Design can actually test each hypothesis (causal claims have causal leverage)
- CMB addressed by procedural design (separate sources/time), not just a post-hoc test
- Endogeneity strategy specified for archival/observational causal claims
- Validated measures; new scales piloted; CFA + discriminant validity planned
- Levels aligned across theory/measurement/analysis; aggregation (ICC, r_wg) justified
- Sampling frame, response rate, and power (incl. interactions) justified
- (Meta) coding protocol, inter-coder agreement, artifact-correction model, bias diagnostics
Anti-patterns
- Cross-sectional causal claims: "X causes Y" from one-wave correlational data
- CMB as afterthought: a Harman single-factor test instead of designed separation
- Ignored endogeneity: an archival "effect" with an obviously endogenous regressor and no strategy
- Mismatched levels: theorizing at the team level, testing individuals via OLS
- Home-grown scales with no reliability or discriminant-validity evidence
- Underpowered interactions presented as null "boundary conditions"
- Vote-counting meta-analysis with no artifact corrections or bias checks
Output format
【Design】experiment / panel-archival / multilevel survey / qualitative / meta-analysis
【Hypothesis-design fit】each H testable? notes ...
【CMB plan】procedural remedies ...
【Endogeneity strategy】instrument / NE / FE / DiD / matching ...
【Measures】validated? new (piloted)? CFA + discriminant?
【Levels】theory / measurement / analysis aligned? aggregation (ICC, r_wg) ...
【Power & sampling】frame, N, power for interactions ...
【Meta only】coding / agreement / artifact model / bias checks ...
【Next step】jmgmt-data-analysis
Version History
- 1839142 Current 2026-07-05 13:47
