if the percentage is not high, may consider using imputation approach if the unknown happens randomly. If the percentage is high, not sure if there is a good approach. Also, The person who responds may select unknown on purpose, in this case, imputation is also not good since something is hiding in the unknown response.
other people may have better suggestions?
扫码加好友，拉您进群
全部版块
我的主页
收藏
