humrel-methods

Methods & Research Design (humrel-methods)

When to trigger

• You are choosing among a qualitative/ethnographic, critical, quantitative, or mixed-methods design
• The design is chosen but its rigor and transparency are not yet defensible
• A qualitative study lacks a sampling logic or trustworthiness safeguards
• A quantitative study lacks construct validity or theorizes only the coefficient
• A mixed-methods study cannot say why both strands and what each adds

Principle: method follows the theoretical question — and HR privileges none

HR treats qualitative, critical, quantitative, and mixed-methods work as equally first-class, in the Tavistock human-relations tradition. There is no house preference for an estimator or a paradigm; what is non-negotiable is that the design fits the social-theoretical question (from humrel-theory-development) and is executed with rigor, transparency, and reflexivity. A sophisticated method cannot rescue a thin theory, and a single immersive ethnography can carry an HR paper if the insight is deep and the craft is high. Note also the double-anonymous review: write methods so they establish credibility without revealing the authors' institution, field site identity, or grant.

Branch A — Qualitative / ethnographic design

For how/why process, meaning, identity, and contested dynamics at work.

• Theoretical (not convenience) sampling. Sites/cases/informants chosen to illuminate the mechanism; state the logic (polar types, extreme/critical case, longitudinal, theoretical replication).
• Access and immersion. Specify duration, depth, and your role (participant vs. non-participant); for ethnography, time in the field; for historical work, the archive.
• Triangulated data. Interviews (count, who, when, guide), observation, documents — and how they corroborate.
• Trustworthiness. Credibility, transferability, dependability, confirmability: member checks, prolonged engagement, audit trail, negative-case analysis.
• Reflexivity. State your standpoint and how it shaped access, interpretation, and the relations you studied — HR expects this, not as confession but as method.

Branch B — Critical design

For research interrogating power, ideology, discourse, and taken-for-granted arrangements.

• Make the critical-theoretical lens explicit (labour process, Foucauldian, feminist, postcolonial, CDA) and tie analytic choices to it.
• Treat reflexivity and positionality as central, not optional.
• Ensure the critique is generative: the design must surface a constructive theoretical claim, not only an unmasking.

Branch C — Quantitative design

For whether/how much/under what conditions across many cases.

• Sample and unit of analysis justified relative to the theory (individuals nested in teams/organizations; dyads; events).
• Construct validity. Operationalizations defended; multi-item measures with reliability; address common-method bias if same-source/cross-sectional.
• Multilevel structure. If theory is cross-level, use HLM/mixed models and justify aggregation.
• Identification in service of theory. Where causal claims are made, state the threat and the strategy (panel FE, longitudinal lags, quasi-experiment) — but identification is a means to a theoretical end at HR, not the contribution itself.

Branch D — Mixed methods

• State the integration logic: sequential, concurrent, or nested, and what each strand adds that the other cannot. Avoid bolting a few interviews onto a survey for decoration.

Either branch

• The design must let you see the mechanism, not just endpoints.
• Pre-empt obvious alternative explanations at the design stage.
• Plan the data-to-theory link now (feeds humrel-data-analysis and humrel-tables-figures).

Execution bridge (StatsPAI / Stata MCP)

For the empirical / causal lane, estimate and audit rather than only specify. Full

map: [execution-with-mcp](../../../shared-resources/empirical-methods/execution-with-mcp.md). Human Relations blends critical/qualitative and quantitative work; apply the chain below to its survey / experimental quantitative papers.

• 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](../../../Journal-of-Finance-Skills/resources/worked-examples/02-execution-walkthrough.md).

Checklist

• [ ] Design matches the theoretical form (process → qual/critical; variance → quant)
• [ ] Qual: theoretical sampling stated; access/immersion specified; trustworthiness named
• [ ] Critical: lens explicit; reflexivity central; critique is generative
• [ ] Quant: construct validity and (if needed) multilevel structure addressed
• [ ] Mixed: integration logic and the added value of each strand stated
• [ ] Obvious alternatives are designed against, not just discussed later
• [ ] Methods text is anonymized (no institution/site/grant tells)

Anti-patterns

• Convenience sampling dressed up as theoretical sampling
• Qualitative work with no transparency about coding, sources, or fieldwork depth
• A fancy estimator chosen to signal rigor when the question doesn't need it
• Critical work that is reflexive in tone but methodologically opaque
• "Mixed methods" that is a survey with ornamental quotes
• Methods that reveal author/site identity in a double-anonymous submission

Output format

【Journal】Human Relations
【Skill】humrel-methods
【Design】qualitative / critical / quantitative / mixed (type)
【Why it fits】link to the social-theoretical question
【Sampling/identification】logic + key threat addressed
【Rigor safeguards】trustworthiness / construct validity / integration logic
【Anonymization】no site/institution/grant tells (yes/no)
【Next skill】humrel-data-analysis