Listening to Politics

Learning objectives

What this chapter asks you to be able to do.

• Explain the inferential jobs qualitative inquiry can perform, including interpretation, mechanism discovery, institutional process, lived experience, category formation, and theory development, without treating interviews as automatic estimates of population prevalence.
• Analyze interviews as evidence-generating political encounters shaped by recruitment, rapport, power, recall, strategic presentation, interviewer effects, refusal, and gatekeeping.
• Explain how focus groups use group interaction as data and why apparent consensus can be produced by status, conformity, dominance, or confidentiality constraints.
• Identify what observation, field notes, documents, and archives can reveal about the gap between formal procedure and political practice.
• Distinguish purposive, criterion, maximum-variation, referral, theoretical, deviant-case, and position-based selection, and state what each selection logic permits a researcher to claim.
• Evaluate saturation claims without applying a universal interview number or confusing code saturation, meaning saturation, theoretical saturation, and information power.
• Use positionality and reflexivity to inspect how the researcher-participant relationship shapes evidence without treating reflexive disclosure as immunity from error.
• Trace the transformation from encounter to recording, transcript, translation, coding, interpretation, and claim, identifying possible information loss and interpretive intervention at each stage.
• Compare deductive, inductive, and hybrid coding approaches and explain why coding consistency is not identical to interpretive validity.
• Apply qualitative rigor through transparent evidence chains, negative-case analysis, contextual adequacy, rival interpretations, ethical integrity, and method-appropriate standards rather than one universal checklist.
• Evaluate AI-assisted transcription, translation, coding, search, and synthesis as forms of assistance and interpretive influence that require privacy, provenance, challenge, and human responsibility.
• Produce Decision Dossier VIII: a Qualitative Evidence and Interpretation Record that preserves the chain from person, practice, or document to research encounter, record, interpretation, and bounded claim.

Chapter summary

The argument in inspectable form.

Key terms

Concepts to carry forward.

qualitative inquiry

A family of research approaches that investigate meaning, process, context, experience, institutions, practices, and mechanisms through evidence such as interviews, observation, documents, archives, and other records.

semi-structured interview

An interview organized around planned topics or questions while allowing probing, follow-up, and reordering in response to the participant and emerging evidence.

elite interview

An interview with a person selected because of institutional position, specialized access, authority, or involvement in a process rather than because the person statistically represents a population.

focus group

A facilitated group discussion in which interaction among participants is itself part of the evidence, not merely a way to conduct several individual interviews at once.

participant observation

Observation in which the researcher participates to some degree in the setting or activity being studied while systematically recording relevant practices, meanings, and context.

purposive sampling

Deliberate case or participant selection based on characteristics or positions relevant to the research question and inferential purpose.

maximum-variation sampling

Purposive selection designed to include consequentially different cases or participant characteristics so that variation, commonalities, and contrasts can be examined.

snowball / referral sampling

Recruitment in which existing participants or contacts help identify additional participants, useful for hard-to-reach networks but capable of reproducing network and gatekeeper bias.

saturation

A family of stopping concepts referring to diminishing analytic returns, whose meaning depends on the research tradition and whether the target is codes, meanings, theoretical categories, or another analytic objective.

information power

A sample-size logic proposing that fewer participants may be needed when cases contain more information relevant to a focused aim, specific sample, strong dialogue, established theory, and intensive analysis.

positionality

The research-relevant relationship between the researcher’s social, institutional, linguistic, professional, or political position and the setting or people being studied.

reflexivity

Systematic examination of how research assumptions, relationships, decisions, and changes in understanding may shape the production and interpretation of evidence.

transcription

The creation of a written research artifact from recorded or remembered speech and interaction; it necessarily includes choices about what features to represent.

coding

The process of assigning analytic labels or categories to portions of qualitative material in order to organize, compare, interpret, or develop an argument.

negative / deviant case

A case, passage, or observation that does not fit an emerging interpretation and is deliberately examined for what it reveals about scope, mechanism, categories, or rival explanations.

transferability

A judgment about whether and how an interpretation developed in one context may illuminate another context, based on theoretically and empirically relevant similarities rather than automatic statistical generalization.

audit trail

A reconstructible record of sampling, fieldwork, transformations, coding, analytic decisions, changes, evidence, and challenges that allows later inspection of how a qualitative claim was produced.

Review & discussion

Questions for seminar, revision, or assessment.

1. Why is it misleading to describe qualitative research as merely anecdotal because the number of cases is small? Give an example of a strong small-N claim and an illegitimate prevalence claim from the same interviews.
2. Mara gives a trust score of six and then distinguishes technical competence from procedural responsiveness. What does the interview add to the number, and what does it still fail to establish?
3. How can an interviewer effect be methodologically important without implying that the participant's answer is false?
4. When would an elite interview be stronger because the interviewee is not representative of the population? What scope limit follows from that selection logic?
5. Why is a focus group not simply a multi-person interview? Describe one finding that depends on participant interaction and one way the interaction could suppress evidence.
6. A formal meeting minute omits an argument that appears in three participants' field notes. What can each source establish, and what additional evidence would you seek?
7. Compare maximum-variation sampling with referral sampling for former transit riders. What problem does each help solve, and what selection risk does each create?
8. What would be wrong with a methods section that states only, 'Saturation was reached after twelve interviews'? Rewrite the statement so its stopping logic becomes inspectable.
9. Give an example in which researcher positionality is load-bearing and another in which a long positionality statement would add little to evaluating the claim.
10. Why can a word-perfect transcript still be an incomplete research record? Which contextual features would you preserve for the Asterbridge sarcasm example?
11. When can literal translation be less faithful to political meaning than a nonliteral translation? What provenance should accompany the choice?
12. Under what research design would inter-coder agreement be important? Under what design could disagreement among coders be analytically valuable?
13. What is the difference between triangulation and three derivative sources repeating the same institutional narrative?
14. A participant rejects the researcher's code for an accurate quotation. Does the participant automatically control the scholarly interpretation? Does the researcher automatically control it? Explain the responsibilities on both sides.
15. How can protecting confidentiality reduce evidentiary richness? When should the ethical cost outweigh the analytic benefit of publication?
16. An AI summarizer preserves every interview in a linked archive but omits minority interpretations from the executive summary. Which safeguard failed: preservation, provenance, salience, validity, or authorization? More than one may apply.
17. Would using three different runs of the same model family count as independent interpretive review? What kinds of diversity or challenge would be more meaningful?
18. Which current NousPolis qualitative safeguards are architectural commitments at N0, and which operational details must remain version-gated until later maturity?
19. Design one reopening trigger for a qualitative codebook and one for a recruitment plan. Each trigger must identify evidence that would materially change the research object or scope.
20. At what point in Chapter 8 does a qualitative interpretation become a public decision? Explain why the correct answer should remain compatible with the rule 'AI analyses. Humans authorise.'

Further reading

Continue into the literature.