科研 Skill

质性访谈设计

qualitative-interview-design

Designs semi-structured interview guides with probing questions, sampling strategy, and saturation tracking for qualitative studies. Use when planning interviews before fieldwork.

发布者 Codar安全审核通过v0.1.0更新于 2026-11-18

SKILL.md

Qualitative Interview Design

Purpose

Supports the design phase of a qualitative or mixed-methods study that relies on semi-structured interviews or focus groups: building an interview guide that elicits rich, relevant data without leading participants, planning a defensible sampling strategy, and setting up a coding workflow (thematic analysis, grounded theory, or framework analysis) before data collection starts. Weak interview guides and undocumented coding decisions are a common reason qualitative work is challenged at review — this skill front-loads the design choices that make the analysis traceable later.

Use when planning a new interview or focus-group study, revising an interview guide after a pilot, or setting up a coding scheme for already-collected qualitative data.

When to Use

  • Designing a new semi-structured interview guide for a qualitative or mixed-methods study.
  • Converting a research question into interview questions and probes.
  • Deciding on a sampling strategy (purposive, theoretical, snowball, maximum variation) and a stopping rule (saturation).
  • Setting up a thematic analysis or grounded-theory coding plan, including a codebook template and intercoder-reliability process.

Workflow

1. Anchor the guide to the research question

  • Restate the research question(s) in one sentence each.
  • For each research question, list the 2-4 topics that must be covered to answer it — these become interview guide sections, not the literal questions themselves.
  • Decide the interview type: structured (fixed wording and order), semi-structured (a guide with flexible probing — the most common choice for exploratory qualitative research), or unstructured (narrative/life-history interviews with minimal pre-set structure).

2. Write questions, then probes

  • Open with a low-stakes, easy-to-answer warm-up question (e.g., "Can you tell me a bit about your role / background with X?").
  • Write main questions as open-ended, non-leading prompts: avoid yes/no framing, avoid embedding the researcher's hypothesis in the question wording (e.g., prefer "How did you decide X?" over "Didn't you feel pressured to do X?").
  • Under each main question, list 2-3 optional probes to use only if the participant does not spontaneously address them: "Can you give an example?", "What happened next?", "How did that make you feel?", "Who else was involved?"
  • Order questions from general to specific and from least to most sensitive; place demographic or sensitive questions near the end unless they are needed to screen eligibility.
  • End with a closing question that gives the participant space to add anything not covered ("Is there anything else you think is important that we haven't discussed?").
  • Keep the guide to roughly 8-15 main questions for a 45-60 minute interview — more than that produces superficial coverage of each topic.

3. Pilot and revise

  • Pilot the guide with 1-3 participants similar to the target sample (colleagues or convenience contacts if target participants are scarce).
  • After each pilot, note: questions that were misunderstood, answered in one word despite being open-ended (reword to be more concrete), or that produced data unrelated to the research question (cut or rework).
  • Time the pilot interview; trim the guide if it runs long.

4. Plan sampling and recruitment

  • State the sampling logic explicitly: purposive (deliberately recruiting information-rich cases), maximum variation (deliberately recruiting across a range of a key dimension), snowball (participant referrals), or theoretical sampling (grounded theory — sampling driven by emerging categories, decided iteratively during analysis).
  • State a stopping rule in advance: a target range (e.g., "15-25 interviews, informed by comparable published studies in the field") combined with a saturation check (stop recruiting once consecutive interviews stop producing new codes or themes relevant to the research question) — pre-registering a fixed number alone is weaker because saturation is sample- and topic-dependent.
  • Draft a short recruitment script and a plain-language consent script covering purpose, voluntary participation, recording, anonymization, and withdrawal rights (see the research-ethics-irb-protocol skill for the full consent/IRB workflow).

5. Set up the coding workflow

  • Choose an analytic approach and commit to it before coding starts:
    • Thematic analysis (Braun & Clarke): familiarization -> generate initial codes -> search for themes -> review themes -> define/name themes -> write-up. Good default for most applied qualitative work.
    • Framework analysis: builds a matrix of cases x a priori and emergent themes; useful for applied/policy research with multiple coders and a need for transparent, auditable comparison across cases.
    • Grounded theory: open coding -> axial coding -> selective coding, with theoretical sampling and constant comparison; used when the goal is to build theory inductively rather than describe themes within an existing framework.
  • Build an initial codebook with code name, short definition, and one example excerpt per code; keep it as a living document rather than fixing it before any data is seen (except a small set of deductive/a priori codes tied directly to the research question).
  • For multi-coder teams, plan an intercoder-reliability check: 2+ coders independently code the same subset (10-20% of transcripts) and compute agreement (Cohen's kappa for discrete codes) before finalizing the codebook; reconcile disagreements by discussion and document the resolution rule.
  • Choose a tool for coding: a spreadsheet for small studies; NVivo, ATLAS.ti, MAXQDA, Dedoose, or the R package qualCoder/Python taguette for larger ones.

6. Report rigor and reflexivity

  • Report the number of interviews, duration range, sampling strategy, saturation justification, coding approach, and any intercoder-reliability statistic.
  • Include a brief positionality/reflexivity note: the researcher's relationship to the topic or participants and how it may have shaped data collection or interpretation — expected by most qualitative reviewers (e.g., following COREQ or SRQR reporting guidelines).

Prerequisites

  • A clearly stated research question the interviews are meant to answer.
  • Access to or a plan for recruiting the target participant population.
  • Institutional ethics approval or a plan to obtain it before recruiting (see the research-ethics-irb-protocol skill) — this skill covers guide and coding design, not the ethics submission itself.

Limitations

  • Provides a general-purpose framework for semi-structured interviews and common qualitative analysis traditions; it does not cover highly specialized methods (conversation analysis, discourse analysis with linguistic transcription conventions, ethnographic participant observation) in depth.
  • Reporting-standard references (COREQ, SRQR) are mentioned by name but not reproduced in full; consult the published checklists directly when preparing a manuscript.
  • Assumes a single-language, in-person or video interview context; cross-language studies need an explicit translation/back-translation protocol not detailed here.
  • Software suggestions (NVivo, ATLAS.ti, MAXQDA, Dedoose) are commercial tools not bundled with this skill and may require institutional licenses.

Quality Checklist

  • Every interview question traces back to a stated research question.
  • Questions are open-ended and non-leading; sensitive questions are placed later in the guide.
  • Guide has been piloted and revised at least once.
  • Sampling strategy and stopping rule (saturation) are stated before recruitment begins.
  • Coding approach is chosen and an initial codebook drafted before full-scale coding.
  • Intercoder-reliability plan exists if more than one coder is involved.

示例提示词

我要做半结构化访谈研究,帮我设计访谈提纲和编码方案。

所属插件

跨学科科研方法跨学科通用科研方法:系统综述与元分析、预注册、实验与问卷设计、文献矛盾与研究空白、科学计量与专利分析、可复现研究、答辩与审稿回复。

安装该插件会一并获得同领域的 54 个 Skill / MCP。

同插件的其他科研 Skill

常见问题

质性访谈设计能做什么?

Designs semi-structured interview guides with probing questions, sampling strategy, and saturation tracking for qualitative studies. Use when planning interviews before fieldwork。

如何安装质性访谈设计?

在科搭 Codar 左侧点击「插件」,在「组件」页签搜索「质性访谈设计」并安装;在 Claude Code 或 Codex 中可以发送本页的安装提示词,或下载 Skill 包后解压到 ~/.claude/skills 或 ~/.codex/skills。

质性访谈设计需要哪些权限?

该科研 Skill没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。

质性访谈设计经过官方审核吗?

是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。