Doug Henwood wrote:
>
> Carrol Cox wrote:
>
> >The two overlap, but there are fundamental differences, beginning with
> >the fact that the kind of "poll" I am speaking of _cannot be done over
> >the phone_ but must be done personally, but the organizer, in direct &
> >protracted conversation.
>
> Ok, I'll agree with this. But that's a completely different ball of
> wax from a "poll," which is, or should be, a snapshot of opinion as
> it exists, apart from efforts to influence it.
>
> Doug
>
Actually, I would think that any kind of poll (provided that one has reasonable information on what sample it represents) gives equally "real" information -- one distinguishes not among "good" and "bad" polls or "accurate" or "inaccurate" polls but between useful and unuseful ones.
I forget the deliberately bizarre question you gave, so I'll invent one of my own.
"If it is true that Bush raped his daughter would you vote for him in November?"
The number of "No" answers, admittedly, would not be interesting, but the number of Yes answers would give real information. If 9% answere yes, than we know that nine percent of the sampled population is so committed to Bush's program that information about the man's personal life cannot sway them. If there is no way of telling what population was sampled and how accurate the sample was, _then_ the poll is useless. But merely stupid or unfair or bizarre ways of asking the question are either useful or not useful depending on one's own purposes. The user merely needs to take the bias into account.
Moreover, if _another_ ('unbiased') poll indicated that at the same time 45% supported Bush, then No answers to the bizarre question would tell us something about how personal information on Bush could affect his support. So just as how the question(s) is(are) asked affects answers to the poll, so the questions a potential user of the poll asks of the poll determine the poll's usefulness. If the question you want to ask the poll is "What percentage of Bush's support is untouchable?", then a poll with the questions strongly biased against Bush is "truer" (i.e., more useful) than a more 'balanced' poll question would be.
I haven't looked at the specific poll you are discussing, but it sounds like it gives accurate information on the following: X% of the population, when and if offered certain information, would support impeachment, Y% even with that information, would oppose impeachment. Useful? Perhaps, perhaps not. But its accuracy depends not on the poll but on how you _use_ the poll's results.
Or put another way, the subject heading of this thread is illuminating. Pick the poll that answers to one's needs. This is _not_ the same as picking the poll which has the most satisfactory answers!!!! Ordinarily I, for example, don't need to know how many people will vote for Bush or for Kerry; even how many (_at this time_)support or don't support the war is only of secondary importance. I _would_ like to know how many still oppose the war _even_ if the torture stories are false or exaggerated. That number would represent the core that the movement might reach _at this time_. And only a poll biased in favor of the war could begin to give us that figure.
The number who would support an impeachment of Bush (given certain facts) is not necessarily (if high) a number to make opponents of the war deleriously happy, for that number indicates the number of people potentially on our side who have naive ideas about how u.s. government works.
Carrol