返回提示词目录第 1412 页

问卷调查结果报告与建议撰写

将调查数据整理为关键发现、主题分析与建议,注明样本基数、方法局限,并附图表建议和题目表。

完整提示词

<context>
A survey report turns tables into a story people can act on, without claiming more than the data can bear. Experienced survey researchers follow a few rules. Every percentage carries its base (n), and bases differ between questions because of routing and skips. Percentages on small bases (commonly under 30 to 50) are shown as counts or not shown, and subgroup differences are only called differences when a test supports them. A margin of error applies only to probability samples; an open-link or volunteer survey has no margin of error, and its results describe respondents, not the whole population. Response rate and who responded matter as much as the answers. Multi-select questions add up to more than 100%, which readers must be told. Agreement scales are summarised as top-two-box (agree plus strongly agree) with the full distribution available, and the question wording is quoted where it shapes the answer. Chart choice follows the question: horizontal bars sorted by size for categorical answers, stacked diverging bars for agreement scales, and no pie charts for multi-select. Open-ended themes are reported with counts and representative, non-identifying quotes. Survey results show association, not cause.

{{#audience}}Audience: {{audience}}{{/audience}}
</context>

<task>
<survey_method>
{{survey_method}}
</survey_method>

<survey_results>
{{survey_results}}
</survey_results>

1. Write the "About the survey" section: purpose, who was invited, completions, response rate, dates, sampling method, weighting, and what population the results can and cannot describe.
2. Pick five to seven key findings, each a plain sentence with the figure and base, ordered by importance to the audience's decisions.
3. Group questions into themes and write each theme section with the findings, the recommended chart per question and why, and subgroup differences only where tested.
4. Report open-ended themes with counts and short non-identifying quotes.
5. Draft recommendations that follow from the findings, marked as suggestions for the owners to decide.
6. Write the limitations plainly and list the questions with exact wording and base sizes.
</task>

<constraints>
- Show the base for every percentage; replace percentages on bases under 30 with counts and say so.
- Give a margin of error only for probability samples, and say why none is given otherwise.
- Do not describe a subgroup difference as real unless a test result was supplied; otherwise call it "a difference among respondents that may not hold more widely".
- Do not use causal language for survey associations.
- Do not quote open answers that could identify a respondent, and suppress groups small enough to identify people.
- Do not add findings that are not in the results supplied.
{{> output/uncertainty}}
</constraints>

<output_format>
## Summary
Three to five sentences for a reader who reads nothing else.

## About the survey
Short paragraph and a small table: Invited | Completed | Response rate | Dates | Method.

## Key findings
Numbered findings, each with figure and base.

## Findings by theme
One subsection per theme: findings, chart recommendation per question, tested subgroup differences.

## Recommendations
Numbered suggestions linked to findings.

## Limitations
Bullets.

## Question list
Table: Question | Exact wording | Base.
</output_format>

可替换内容

  • {{#audience}}
  • {{audience}}
  • {{/audience}}
  • {{survey_method}}
  • {{survey_results}}
  • {{> output/uncertainty}}

分类与标签

办公效率、问卷分析、调查报告、样本基数、研究局限

前往 AI 图片创作