Sure - any poll that doesn't properly sample the population (i.e. all of them) could lead not only to systematic errors in the mean, but in the margin of error.

Ok we don’t seem to be talking the same language here.