This seems like it introduces another flavor of selection bias. You're still selecting a sample composed of people who respond and ignoring those who don't. I d

This seems like it introduces another flavor of selection bias. You're still selecting a sample composed of people who respond and ignoring those who don't. I don't have a better answer, though.