1. Double-check your data and your analysis:
- Integrity-check your data.
- Read the documentation.
- Know how many records you should have.
- Check counts and totals against reports.
- Are all possibilities included? All states, all counties, correct ranges?
- Check for missing data, duplicates, internal problems.
- Beware the spurious correlation.
- Use terms such as "significant", "likely" and "correlation" correctly.
- Do a gut check on your analysis.
- What else could explain my findings?
- Did I fill in all possible holes?
- Did I collect all the data I needed to?
- The analysis is just the beginning. Once you start reporting, ask: Is it consistent with my findings?
- Run your results by experts.
- With publication, provide a detailed methodology about the data and your process.
- Invite feedback and corrections.
Jennifer LaFleur, Senior Editor for Data Journalism, Center for Investigative Reporting.
2. Double-check your reporting:
- Test your data.
- Test the questions you intend to ask of your data.
- Challenge your assumptions and your findings.
- Have a second pair of educated eyes look at your data and your conclusions.
- Think through, thoroughly, how you will present your data.
Ricardo Sandoval Palos, Senior editor, NPR's Morning Edition.
3. Present your data clearly:
- Take courses to train yourself how to think about data visualization.
- Find examples in which data is presented in an informative way.
- If you're going to try something beyond a pie or bar chart, include annotation to explain what you are doing and why.
- Make the underlying data available, but always with context to assist analysis.
Chrys Wu, Developer Advocate, The New York Times.
Presented at IRE14, San Francisco, 2014.