Primary Research

Primary Research

Use When

• Use when the project needs first-hand qualitative evidence such as interviews, observation, focus groups, field visits, or coded open-ended responses.
• Use when desk research alone cannot answer the decision with enough fidelity.

Do Not Use When

• The work is purely secondary-source research.
• The task is about writing up existing findings rather than designing or interpreting primary evidence.

Primary Research Required Context

• Decision or question the fieldwork must inform.
• Proposed respondent / setting universe and access constraints.
• Expected output form and evidence-sensitivity level.

Primary Research Method Detail 2

• Read this SKILL.md first, then load only the relevant reference file for the method being used.
• Design the instrument and sampling logic before collection.
• Preserve traceability from raw material to codes, themes, and final claims.

Quality Standards

• Sampling logic is explicit.
• Notes distinguish quote, paraphrase, observation, and inference.
• Findings are theme-based, not anecdote-based.

Primary Research Additional Failure Modes

• Casual interviews with no guide.
• Treating vivid respondent anecdotes as representative evidence.
• Coding only what confirms the working theory.

Primary Research Core Deliverables

• A defensible primary-research design, collection plan, coding approach, or interpretation with traceable claims.

References

• Use the references/ files for the specific method being run.

Inputs

Capability contract

Read access to the approved protocol, consent basis, and field records is required. Participant contact, recording, field collection, data editing, or publication needs explicit ethical and sponsor authority.

Degraded mode

Fallback if consent, field access, transcription, or coding tools are unavailable: return a qualified protocol or partial analysis, name the evidence gap, and mark credibility checks unassessed.

Decision rules

Outputs

Primary Research Evidence Notes

• Preserve consent status, participant criteria, field deviations, codebook decisions, dissenting evidence, and credibility checks.

Worked example

For interviews, record consent, ask the approved open question, code responses against a documented codebook, and preserve contradictory responses.

This skill governs evidence the engine helps design or interpret when the source is a person, a setting, or a live interaction rather than a published document.

When to use

Use this skill when the project includes any of:

• Expert or stakeholder interviews
• Key-informant interviews
• Focus groups or small discussion groups
• Field visits, shadowing, or direct observation
• Open-ended survey responses that need coding
• Case-study field evidence beyond desk research

Do not use this skill for purely secondary-source work. Pair it with research-orchestration, not instead of it.

Five rules

1. Design before collection. Every interview, visit, or group needs a question-set, sampling logic, and note plan before the first contact.
2. Purposeful sampling beats convenience sampling. The point is coverage of perspectives, not easy access alone.
3. Raw material stays traceable. Claims in synthesis must be recoverable to notes, transcripts, observation logs, or coded excerpts.
4. Coding precedes conclusion. Do not jump from striking anecdotes to findings without theme-building.
5. Credibility is engineered. Use triangulation, disconfirming evidence, member-checking where appropriate, and explicit limits.

Router

Primary Research Core Method Notes

1. Define the decision the fieldwork must inform.
2. Build the sample frame: who must be heard, who may be heard, who is excluded.
3. Draft the instrument: interview guide, observation sheet, or discussion prompts.
4. Collect with disciplined notes: time, setting, speaker role, direct quotes vs paraphrase.
5. Code the material into themes, contrasts, and negative cases.
6. Write findings with provenance and limits.

Ship gate

• Research question and user of the finding are explicit.
• Sample logic is stated and defended.
• Instrument exists before collection.
• Notes distinguish quote, paraphrase, inference, and observation.
• Theme-building is visible; findings are not anecdote dumps.
• At least one disconfirming or minority view is surfaced where relevant.
• Ethics / consent / sensitivity constraints are stated.
• Every load-bearing finding is traceable back to raw material.

Anti-patterns

• Running interviews as casual conversations with no guide
• Quoting the most vivid respondent as if they represent the population
• Treating open-ended responses as decoration instead of data
• Coding only confirming evidence
• Hiding weak samples behind strong prose

Companion skills

• research-orchestration — wave planning and sub-agent brief structure
• research-design — case study, MROC, and formal design layer
• source-evaluation — published sources still need tiering
• analytic-tradecraft — when field evidence feeds a contested judgment
• academic-writing / report-and-proposal-craft — final artifact layer

Sources for this skill

• Patton, Michael Quinn. Qualitative Research & Evaluation Methods: Integrating Theory and Practice. 4th ed., SAGE, 2015. Tier 1.
• Poynter, Ray. The Handbook of Online and Social Media Research. Tier 1 for MROC and online qualitative practice.

Workflow

1. Confirm the decision, participant criteria, consent basis, and approved protocol.
2. Collect the minimum first-hand evidence with field provenance and safeguarding controls.
3. Stop when consent, participant safety, or protocol authority is unresolved.
4. Code and challenge findings; recover by documenting deviations and limiting claims to usable evidence.

Evidence Produced

Reference Index

• Credibility and quality; interview protocols
• Observation and fieldwork; qualitative coding