Why merging NC and "No" values for a binary (Yes-no) variable?
In some cases, you may have many unknown values for a binary variable. This would lead to excluding a lot of patients when analyzing this variable. However, you have a high certainty that patients without value could be considered as "No". For example, if your Yes-no variable answer the question "Does the patient have a rare disease?". In most cases, unknown values may correspond to "No".
Merging NC and "No" values would increase the number of patients that will be analyzed and therefore the power of your analysis and its representativeness.
How to merge NC and "No" values during a statistical analysis?
- Go to "Test variables"
- Pick a Yes-no variable
- Click on the settings icon next to the name of the variable
- In front of "Unknown values", select "Consider NC as No"
- Validate with the "Confirm" button
Note: in rare cases, you may want to consider NC values as "Yes". This can be done in the same way. Instead of "Consider NC as No", choose "Consider NC as Yes".
How to merge definitively NC and "No" values for all statistical analyses?
You may decide that you want to replace all NC values with "No" in your series. This can be achieved by updating your database as a whole with the Bulk update function.
- Filter the patients that have unknown values for your variable (How to filter patients)
- Go to the menu "Data Manager" and then "Bulk update"
- Pick the variable you want to update
- Choose the value "No"
Beware! Bulk updates are irreversible and should be performed with great caution.