Society of Actuaries (SOA) PA Practice Exam

For which missing data percentage should rows be removed according to best practices?

1. Less than 10%

2. Less than 5%

3. More than 10%

4. More than 5%

The correct answer is: Less than 5%

In data analysis and data preparation, the guideline regarding missing data is crucial for ensuring the integrity and reliability of the results. Generally, a missing data percentage of less than 5% is considered acceptable for most analyses. This threshold indicates that the amount of missing information is small enough that it is unlikely to bias the results significantly or compromise the overall quality of the dataset. When the missing data is within this range, analysts often have multiple strategies for handling it, such as imputation, where estimated values are filled in based on other available data. Maintaining rows with such a small percentage of missing values ensures that valuable information is not discarded unnecessarily, while also minimizing the impact on the overall dataset's distribution and variance. Going beyond this threshold tends to raise concerns regarding the robustness of the data. For instance, a missing data percentage greater than 5%, let alone 10%, often necessitates a more cautious approach. Higher levels of missing data could lead to biased outcomes or diminished statistical power, suggesting that removing rows might be a better approach in these scenarios to maintain optimal data quality.