Back to Details

research-data-analysis

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Research Data Analysis

研究数据分析

Analysis should be repeatable, inspectable, and connected to claims.

分析应具备可重复性、可检查性，并与研究结论相关联。

Read First

必读内容

```
references/repository-contract.md
```
```
references/output-contracts.md
```

```
references/repository-contract.md
```
```
references/output-contracts.md
```

Workflow

工作流程

Identify raw, interim, and processed data paths.
Read data dictionary and provenance notes.
Write reusable analysis logic in
```
src/
```
, not only notebooks.
Use scripts or CLI entrypoints for repeatable runs.
Put generated tables in
```
outputs/tables/
```
and figures in
```
outputs/figures/
```
.
Record important findings in
```
wiki/claims/
```
or
```
wiki/experiments/
```
.
Link paper claims to exact table, figure, or experiment artifacts.

确定原始数据、中间数据和处理后数据的存储路径。
阅读数据字典和来源说明。
将可复用的分析逻辑编写在
```
src/
```
目录下，而不只是在笔记本中。
使用脚本或CLI入口点实现可重复运行。
将生成的表格存入
```
outputs/tables/
```
目录，图表存入
```
outputs/figures/
```
目录。
将重要研究结果记录在
```
wiki/claims/
```
或
```
wiki/experiments/
```
目录中。
将论文结论与对应的表格、图表或实验制品关联起来。

Statistical Discipline

统计规范

Match tests to design, distribution, sample size, and hypothesis.
Report effect sizes and confidence intervals when relevant.
Distinguish statistical significance from practical significance.
Flag multiple comparisons, missing data, selection bias, and confounds.
Do not treat exploratory analysis as confirmatory.

根据研究设计、数据分布、样本量和假设选择合适的统计测试方法。
相关情况下报告效应量和置信区间。
区分统计显著性与实际显著性。
标记多重比较、缺失数据、选择偏差和混杂因素。
勿将探索性分析当作验证性分析。

Figure Discipline

图表规范

Use readable labels, units, captions, and colorblind-safe palettes.
Avoid misleading axes and decorative chart types.
Save source data for final figures when practical.
For publication figures, record command and input dataset.

使用易读的标签、单位、标题，以及适合色弱人群的调色板。
避免使用误导性的坐标轴和装饰性图表类型。
可行时保存最终图表的源数据。
对于发表级图表，记录生成命令和输入数据集。

Notebook Policy

笔记本使用规范

Notebooks can explore and explain. If the logic becomes part of the result, move it into

src/

and leave the notebook as a consumer.

笔记本可用于探索和解释分析过程。若分析逻辑成为结果的一部分，需将其迁移至

src/

目录，仅保留笔记本作为结果的展示载体。