返回提示词目录第 1474 页

设计内容分析研究与编码方案

围绕研究问题制定抽样、编码和信度检验方案,明确可支持的结论及资料使用规则待核事项。

完整提示词

<context>
Content analysis systematically describes communication content. Quantitative content analysis (in the tradition of Krippendorff and Neuendorf) defines variables in advance, samples content, has trained coders apply a codebook, tests intercoder reliability, and reports frequencies and associations. Qualitative content analysis (for example Schreier or Hsieh and Shannon's conventional, directed and summative types) builds or applies categories more flexibly to describe meaning. Typical weaknesses: a convenience sample of whatever a keyword search returned, unclear units (article, paragraph, sentence, image, post), latent constructs like "framing" coded with no decision rules, reliability tested on easy variables only or reported as percent agreement, and claims about audiences or effects that content data cannot support. For digital content, platform search and API limits shape what can be sampled, and deleted or private posts raise ethics and representativeness issues. Automated coding with dictionaries or language models needs validation against human coding.
</context>

<task>
Design a content analysis for:
<question>
{{question}}
</question>
<material>
{{material}}
</material>

1. Choose quantitative, qualitative or mixed content analysis and say why; state which claims the design can support (describing content) and which it cannot (effects on audiences, authors' intentions).
2. Plan sampling: the population of content, the sampling frame and its gaps, search terms with a relevance check, the sampling method (for example constructed weeks for news, stratified by outlet or platform), sample size reasoning, and how duplicates and syndicated copies are handled.
3. Define units: sampling unit, recording unit and context unit.
4. Draft the coding scheme: manifest variables and latent variables, each with a definition, categories, decision rules and an example from the type of material (labelled illustrative). Keep categories mutually exclusive where coded as single choice.
5. Plan coding: coder training, a pilot on material outside the sample, codebook revision, independent double coding of a subsample, and disagreement resolution. If automated coding is proposed, plan validation against human coding.
6. Plan reliability: Krippendorff's alpha (or a justified alternative) per variable, the subsample size, acceptable thresholds and what happens to variables below them.
7. Plan analysis matched to the question (frequencies, trends over time, comparisons across outlets, associations) and list limits and ethics issues for the material.
</task>

<constraints>
- Do not invent example content attributed to real outlets or people; examples are labelled illustrative.
- Do not report percent agreement alone as reliability.
- Do not claim the design shows effects on audiences or authors' motives.
- Note platform terms, copyright and privacy for collecting and quoting online content; tell the user to check current rules.
</constraints>

<output_format>
## Approach
Type of content analysis, reasons and supportable claims.
## Sampling
Frame, method, size and handling rules.
## Units and variables
Units, then a codebook table: variable | definition | categories | decision rules | example (illustrative).
## Coding procedure
Numbered steps.
## Reliability
Coefficient, subsample, thresholds and actions.
## Analysis and limits
Analysis plan, then limits and ethics.
</output_format>

可替换内容

  • {{question}}
  • {{material}}

分类与标签

学习教育、内容分析研究、抽样设计、编码手册、信度检验

前往 AI 图片创作