[lbo-talk] Why the GOP is sitting pretty, despite the polls

[Gar Lipow]

On 8/10/05, Gar Lipow <the.typo.boy at gmail.com> wrote:
> On 8/10/05, Jim Devine <jdevine03 at gmail.com> wrote:
> > WS: >Opinion polls and focus groups are based on very different models of human
> > behavior and even science. <
> >
> > right. I would guess that smart marketing experts would see polls and
> > focus groups as complementary tools...
> > --
> > Jim Devine
> 
> Right - which I  acknowledged in  my post. The point is that they are
> an in depth and partially subjective  analysis. (I compared them to
> clinical experience in medicine  - which is far from a dismissive
> analogy.)   So  to  evaluate the validity of a focus groupp you need
> more detail than  you  do on a poll.  The particuluar  pdf referred
> (I'm not going to  repeat it since it has been in two posts now)
> carries about the level of summary information you would get in a
> poll.  That amount of infformation is valid for a poll. It is not for
> a focus group. Nuance is extremely important in  evaluating a focus
> group.  Unless you want to  take the word of the person doing the
> study, not only their honesty (which I frankly  I'm not at all
> confident of  in this case) but their judgement, you  need enough
> detail to see if the  analysis makes sense.  It is not that focus
> groups have less validity than polls. (Actually I suspect they do -
> but that is another and long argument I'm not going to get involved
> in).  It is that those conducting the focus group need to disclose a
> great deal more of the raw data than you do for a poll - statistical
> or summary level information does capture focus group results. You
> need to know what a large percent of the respondents actually said. If
> you are a sociologist on this list, think of comparable  sociological
> studies. Would you endorse the validity  of the analysis of a small
> group sociological study with the level of detail  you were  given for
> this focus group?
> 

Really bad typo: Should have been
> statistical  or summary level information does  NOT capture focus group results