methodology-selection

name

methodology-selection

description

Use this skill whenever a user needs help selecting, justifying, or evaluating research methods for anthropological or qualitative research. Triggers include: "which methods should I use," "how do I justify my methods," "method-stance alignment," "my reviewer says my methods don't match my theory," "multi-method design," "mixed methods in anthropology," or "what methods fit an interpretivist / critical / STS / feminist / applied project." Also trigger on questions about epistemic coherence between theory and methods, evidence types needed for a research question, composing a multi-method system, writing a methods justification narrative, or data governance as a design decision. Do NOT use for writing a full research plan (use research-plan skill), grant proposals targeting a specific funder (use grant-proposal skill), or designing specific instruments like interview guides (use fieldwork-methods skill). This skill handles the upstream design decision of which methods and why.

Methodology Selection

Select and justify research methods for anthropological research by treating method choice as an epistemic-design problem: specifying a warranted path from an epistemic stance and research question to defensible claims, using evidence types the stance treats as meaningful, through a coherent multi-method system whose internal logic is explicit. Method selection is not "picking tools" — it is an argument about why these methods, for this question, from this stance, will produce the evidence needed to support the claims you intend to make.

Quick Reference

Task	Reference
Decision workflow, criteria, failure modes, checklist	Read references/methodology-selection-guide.md
Method-stance compatibility matrix, justification templates, worked examples	Read references/method-stance-compatibility.md
Method module details (evidence, claims, limitations, ethics), multi-method patterns	Read references/method-modules.md

Workflow

Step 1: Identify What the User Needs

Determine the entry point:

• Selecting methods from scratch. The user has a research question and stance but hasn't chosen methods yet. Load the guide and run the full decision workflow.
• Justifying existing choices. The user already has methods but needs help writing a defensible justification narrative. Load the compatibility map for stance-specific templates and the guide for phrasing patterns.
• Checking stance-method coherence. The user wants to know if their proposed methods fit their epistemic stance (often prompted by reviewer feedback). Load the compatibility map to check S/C/I/T ratings.
• Writing a methods justification narrative. The user needs proposal-ready or paper-ready prose explaining their method system. Load the compatibility map for stance-family templates and the guide for phrasing patterns.

Step 2: Gather Context

Before generating any content, collect these inputs:

Required:

1. Research question(s). What is the user trying to answer? This determines what evidence is needed.
2. Epistemic stance. Which theoretical orientation(s) does the researcher work within? Ask for primary and secondary. The stance determines what counts as evidence and what methods are epistemically coherent.
3. Field configuration. Single site, multi-sited, digital, archival, hybrid? This constrains which methods are practical.

Important but can be inferred: 4. Scale and temporality. Small-N intensive, multi-population, longitudinal, cross-sectional? Affects the design logic. 5. Access constraints. Where observation is impossible or risky, trace and documentary methods become more central; where recruitment is constrained, sampling logic must shift. 6. Risk posture. Low-risk, vulnerable populations, high-surveillance, politically sensitive. Affects ethics and data governance requirements. 7. Resources, skills, time. Methods that cannot be implemented with rigor are not "best" methods. Short timelines may favor rapid assessment.

Helpful but not required:

• Downstream target: will this feed into a proposal, research plan, or paper?
• Career stage (affects ambition calibration)
• Language competencies
• Whether methods have already been partially chosen

Step 3: Load Appropriate References

• Always load references/methodology-selection-guide.md for the decision workflow, criteria, and checklist.
• Load references/method-stance-compatibility.md when the user needs stance-specific guidance: compatibility ratings, justification templates, or worked examples.
• Load references/method-modules.md when comparing method options or composing a multi-method system: evidence types, claims supported, limitations, ethical considerations, and multi-method design patterns.

Step 4: Run the Decision Workflow

Follow this sequence (detailed in the guide reference file):

1. Define the claim envelope. Based on the epistemic stance, state what kinds of claims are admissible and what kinds are not. An interpretivist project makes claims about meaning, not prevalence. A critical project makes claims about power, not neutral description.

2. Decompose the question into evidence needs. Translate the research question into required evidence types: embodied practices (requires observation), meaning-making (requires interpretive elicitation plus context), distributions (requires standardized measurement), discourse-in-use (requires recordings and transcription), historical sequence (requires archives), network/process across sites (requires multi-sited or trace strategies), materiality (requires object-oriented or sensory methods).

3. Generate candidate method modules. From the 14 method modules in the method-modules reference, identify which could produce the required evidence.

4. Check epistemic coherence. Using the compatibility matrix, rate each candidate method against the user's stance: Standard (S), Coherent (C), Innovative/defensible (I), or High-tension (T). Flag any T-rated methods and explain what reframing would be needed to make them defensible.

5. Check field constraints. Filter candidates by access, risk, consent feasibility, platform terms, legality, and resource availability.

6. Compose the multi-method system. Assign each surviving method a role: primary evidence generation, complementary perspective, contextualization, or validation. Ensure the system has internal logic — methods should relate to each other, not just coexist.

7. Specify the integration plan. State when and where evidence streams are joined, what analytic strategy governs integration, and what meta-inferences result. Do not use "triangulation" without specifying the type (data, method, theory) and what convergence or divergence means.

Step 5: Generate Output

Produce one or more of these deliverables depending on user needs:

• Method justification narrative. Stance-grounded prose explaining the method system. Use the stance-family templates from the compatibility reference. Every method gets a role statement: what evidence it produces, what claims it supports, what its limitations are.
• Method-system composition. A structured overview of the method system showing each module, its role, its evidence contribution, and how it integrates with other modules.
• Integration plan. When and how evidence streams are combined, what analytic strategy governs integration, and what meta-inferences result.
• Ethics and data governance plan. Consent strategy, identifiability analysis, storage and embargo choices, platform-specific ethics for digital methods, and rules for future sharing.

Step 6: Quality Check

Before presenting output, verify using the full checklist:

• Epistemic stance is named and the claim envelope is stated
• Each research question is translated into specific evidence needs
• Every method module has a role statement (evidence -> claim -> limitation)
• Each method is justified in relation to stance AND question, not as a generic disciplinary standard
• Sampling logic is specified and sample sizes are justified using information power or defensible saturation reasoning
• Implementation details are sufficient: sites, recruitment, instruments, recording, transcription, fieldnote protocols
• Integration plan is explicit for multi-method projects
• If computational methods: validation plan is specified
• If digital methods: internet-specific ethics are addressed
• Ethics and data governance plan is included
• Limitations and what the design cannot know are stated
• Timeline and feasibility are realistic
• If funder-required: data management and sharing plan is consistent with ethnographic ethics

Parameters

• Epistemic stance: All 42 stances are relevant, grouped into stance families for compatibility mapping (interpretive/hermeneutic, phenomenological, critical/political economy, feminist/queer, STS/ANT, applied/design, cognitive/psychological, linguistic, computational/digital, plus an unspecified-family template). See DESIGN.md for the full list.
• Genre/audience: Methods section (for proposal, plan, or paper), standalone methodology design memo, methods justification narrative.
• Compression: Brief design sketch (1-2 paragraphs), methods rationale (1-2 pages), full methods section (3-8 pages).
• Risk posture: Low-risk, vulnerable populations, high-surveillance, politically sensitive. Higher risk postures require more detailed ethics and data governance.
• Field configuration: Single site, multi-sited, digital, archival, hybrid, comparative.
• Scale: Small-N intensive, multi-population, longitudinal, cross-sectional.

Guardrails

• Do not generate without knowing the epistemic stance. Stance determines what counts as evidence, what methods are coherent, and what claims are admissible. "Methods" without a stance is an incoherent request — ask the user to identify their stance before proceeding.
• Do not produce methods as a grocery list. Every method must have a role statement: what evidence it produces, what claim it supports, what its limitation is. "I will use participant observation, interviews, and surveys" is a failure mode unless each method's contribution is specified.
• Do not claim triangulation without specification. Require the type of triangulation (data, method, theory) and state what convergence or divergence would mean for inference. "Triangulation" as a magic word is a documented failure mode.
• Flag stance-method tension explicitly. When a proposed method is rated High-tension (T) for the user's stance in the compatibility matrix, explain the tension and what reframing would be needed. Do not silently pass high-tension combinations.
• Ethics and data governance are design determinants. Do not treat them as an appendix. Risk, identifiability, consent feasibility, and future harms from data circulation must inform method selection, not just accompany it.
• Require validation for computational methods. If computational text analysis, network analysis, or other automated methods are included, require a validation plan (close reading, triangulation, error analysis). Model outputs are not self-validating.
• Require internet-specific ethics for digital methods. If digital ethnography, trace methods, or platform-based research is included, require explicit treatment of public/private ambiguity, searchability of identifiers, platform terms, and consent expectations.

Common Failure Modes

Failure mode	Prevention
Methods as grocery list — no inferential role specified	Require a role statement per method: evidence -> claim -> limitation
Generic justification — "participant observation is a hallmark of anthropology"	Enforce stance-and-question anchoring: why this method is necessary here
Stance-method mismatch hidden by vague language	Add claim envelope step; check compatibility matrix; flag T-rated methods
Integration left implicit — "triangulation" as magic word	Specify type of triangulation and what convergence/divergence means
Sample size by round number or unexamined "saturation"	Use information power or empirically grounded saturation reasoning
Ethics treated as appendix	Require ethics and data governance as design determinants, not afterthoughts

Examples

Example 1: Selecting methods for an interpretivist project

Input: "I'm studying how gig workers make meaning out of algorithmic management. I'm an interpretivist drawing on practice theory. What methods should I use?"

Output approach:

• Load all three reference files
• Set epistemic stance to interpretivist + practice theory
• Claim envelope: meaning-making, practice-account relations, situated explanation — not prevalence or causal measurement
• Evidence needs: embodied practices (observation), interpretive frames (interviews), institutional categories (documents)
• Method system: participant observation (ride-alongs, waiting spaces) + semi-structured interviews + document analysis (platform communications, worker forums as texts) — all S or C rated for interpretive stance
• Integration: observation builds domain knowledge and refines interview guides; interviews test interpretive hypotheses; documents trace category circulation; iterative qualitative analysis across all streams
• Role statements for each method

Example 2: Selecting methods for a computational/digital project

Input: "I want to study how climate misinformation spreads in online communities. I'm coming from a computational/digital ethnography perspective. I'm thinking of scraping forum data and doing topic modeling."

Output approach:

• Load all three reference files
• Set epistemic stance to computational/digital
• Claim envelope: pattern identification at scale, platform-mediated practices, distributed coordination
• Check forum scraping + topic modeling: both S-rated for computational/digital stance, but require validation plan
• Recommend adding digital ethnography (community engagement, contextual interpretation) to ground computational findings — S-rated, standard integration pattern
• Require: validation through close reading, participant engagement for interpretation, internet-specific ethics (consent, searchability, platform terms)
• Flag: scraping without ethical governance plan is a guardrail violation

Example 3: Checking stance-method coherence

Input: "I'm doing a feminist study of reproductive healthcare access but my advisor wants me to include a survey. Is that compatible with my framework?"

Output approach:

• Load compatibility reference
• Check surveys against feminist/queer stance: rated C (coherent, not standard) — epistemically compatible but requires explicit framing
• Explain: surveys are coherent within feminist frames when they operationalize structural exposure (access barriers, wait times, insurance status) and are integrated with ethnographic interpretation, not used as standalone "proof"
• Recommend framing: position survey as mapping the distribution of structural barriers that ethnography then explains in terms of lived experience and power; include critical analysis of what the survey categories make visible and invisible
• Flag: surveys become high-tension if treated as neutral measurement without feminist critique of the categories themselves