What are some best practices for documenting and reporting missing data in data analysis?

1 Identify the types of missing data

Missing data can have different causes and implications, depending on the type of data and the context of the analysis. For example, missing data can be due to random errors, non-responses, censoring, or intentional omission. Some types of missing data are more problematic than others, and require different strategies to handle them. Therefore, you should identify and document the types of missing data in your data set, and explain how they affect your analysis.

2 Choose an appropriate method to handle missing data

There are various methods to handle missing data, such as deleting, imputing, or modeling. Each method has its advantages and disadvantages, and may not be suitable for every situation. For example, deleting missing data can reduce the sample size and introduce bias, while imputing missing data can introduce uncertainty and error. Modeling missing data can account for the underlying mechanisms and relationships, but can be complex and computationally intensive. Therefore, you should choose an appropriate method to handle missing data, based on the type, amount, and pattern of missingness, and the goals and assumptions of your analysis.

3 Evaluate the impact of missing data on your analysis

Handling missing data can affect the quality and validity of your analysis, so you should evaluate the impact of missing data on your results and conclusions. For example, you can compare the descriptive statistics and distributions of the complete and incomplete data, and check for any significant differences or outliers. You can also assess the sensitivity and robustness of your analysis to different methods or scenarios of handling missing data, and report the range or confidence intervals of your estimates.

4 Report the missing data information and assumptions

Finally, you should report the missing data information and assumptions in a clear and transparent way, so that your audience can understand and evaluate your analysis. For example, you can report the proportion and pattern of missing data in your data set, and the type and mechanism of missingness. You can also report the method and rationale for handling missing data, and the limitations and uncertainties involved. You can also provide the code or scripts that you used to handle missing data, or share the data set with the missing data indicators.

By following these best practices, you can document and report missing data in data analysis in a systematic and rigorous way, and avoid potential pitfalls and errors.

5 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?