This is up to you. It's more of an M&E question than a data management question. There are pros and cons to adding an "unknown" option for disaggregated data.
Cons for adding "unknown": Less certainty about disaggregated data
- In some instances, the "unknown" option lets people get away with bad record keeping. Allowing "unknown" means that partners do not necessarily have to record disaggregations in order to report data.
- Analysis on disaggregated data with a large number of "unknowns" is questionable. You could not draw reasonable conclusions about gender breakdown for an indicator result if you had 23 males, 40 females, and 165 unknowns.
Pros for adding "unknown": Capture more accurate totals
- If data cannot be reported in the case that the disaggregation is unknown, it could lead to underestimates of total results.
- In some cases, like surveys where you have little control over acquiring complete details, it's valid to use "unknown" to make sure you've at least counted everyone/everything.