To expand, if the goal of the probe were to make inferences about the
Allstat population then sampling bias would be undesirable. But if the idea
is to see if there's critical mass to launch start a chat group for
statisticians and stats groupies then bias is what you want, since there's
no sense hearing from people that wouldn't be interested in participating,
and you don't really care what proportion of the Allstat population they
represent. So, as in any research, it boils down to remembering what the
real question is.
d.
|