The problem of dealing with non-responses is extremely complex and must be approached with caution.

The "data" you provided is not sufficient for a comprehensive answer.

I can recommend the following text, which deals with the problem with adequate completeness:

Bethlehem, Cobben, Schouten (2011), Handbook of Nonresponse in Household Surveys, Wiley & Sons

Indeed, this is a complex question to give a precise or detailed answer. The main problem with non-response is how to assign plausible estimates to these cases to compute it as a response. That is, how to determine that the non-response is distributed in a similar way to the response. This cannot be known a priori, so the non-response is usually estimated with imputation procedures, assuming that the non-response IS distributed like the response (proxy case, average of the group that does respond, average of the stratum, etc.). If this strategy is accepted (with its risks), the weighting of the non-response can be done in the same way as the response, with little deviation from reality if the non-response values ​​and their errors are normally distributed (which is usually the case). ). Otherwise, the recommendation of Dr. Onipe Adabenge Yahaya. is the most appropriate.

To adjust survey weights for non-response rate in a stratified random sample design, calculate the inverse of the response rate within each stratum and apply these as adjustment factors to the original survey weights.