Explain a dataset or chart without overstating it
Translate metrics and patterns into a clear explanation while separating observations, possible causes and data limitations.
What to provide
- The dataset or chart with labels, units, time periods and any known missing values.
- The audience and the question they need answered.
Example Illustrative example
Input
Illustrative editorial example — not a recorded model test. Fictional monthly survey: January, 30 of 50 respondents found the guide useful; February, 40 of 50. The respondents may differ between months. Audience: the guide’s editor.
Output
The share reporting that the guide was useful rose from 60% to 80%, an increase of 20 percentage points. Both samples contain 50 respondents. This is an observed difference; the data alone does not show that a particular edit caused it. Different respondents or survey conditions could also affect the comparison.
Improve the result
- Check the calculations and list what additional data would be needed to assess the suggested explanations.
How to use this prompt
Provide denominators as well as percentages, and explain how the data was collected. Label estimates and missing values before requesting an interpretation.
What to check in the result
Charts may hide scale choices or missing data, and small samples may be unrepresentative. Verify calculations and do not treat a pattern as proof of causation.