What needs analysis is the _meaning_ of polls in general, within a given perspective.
For example. It seems to me that what they measure is a passive response to a question _under given circumstances_. Hence they are useful (for a given purpose) _only_ if that purpose is served by that passive response, under the given circumstances. So that moves the discussion from the poll results to a discussion of _what_ are the purposes of those participating in the discussion. And that discussion, of course, proliferates in endless ways.
For example, for a given perspective (that of all of us involved in (a) building the anti-war movement and (b) directing that movement towards unqualified opposition to the Occupation) the _relevant_ poll figures of Iraqi last year was the 15+/-% desiring immediate U.S. withdrawal. That guaranteed the civil (and uncivil) unrest of the last year (and of the year to come). Hence we could confidently urge others (locally and nationally) to direct their efforts towards _preserving_ the organizational structure of the movement and making End the Occupation Now our organizing slogan for reaching new people. That is, we could know that, given that figure, conditions both for the occupying troops and the Iraqi people would steadily worsen, and that organized (and experienced) opposition to the Occupation would find conditions favorable to growth (at least after the quadrennial u.s. vacation from politics ended in November).
And we could also know from October 2001 on that the potential constituency for an anti-war movement in the u.s. was much larger than it had been during the Vietnam war -- this from polls showing 20% or even more of the population was (at least passively) opposed to the war.
It also makes a great difference whether one's interest in the issue is present-oriented or focused on the future. One does not need polls, for example, to tell one that this is no time to call for tearing up paving stones to build barricades. Hence the opinion of the bulk of the population _at this time_ on socialism can be of little interest to socialists. Our interest has to be on those issues that can involve (over the next year -- not next week) substantial minorities of the population in active struggle.
And here even immense numbers _passively_ supporting us are not necessarily hopeful. Health care is, as 'everyone' more or less acknowledges, a horror for 10s of millions in the u.s. Polls I believe usually show much of this. But here we run into a difficulty (in which polls can neither help nor hurt us) over what may be an instance of quantity/quality change. That immense number _potentially_ activated in the demand for decent health care won't move until they _see_ possibilities _in action_ -- which involves, of course, having substantial numbers of people in action. Different or more effective agit/prop or better proposals won't help here, because there is no one to listen to (or even know the existence of) those proposals until they see activity around them. I don't know the solution for this, though I suspect that initial impetus will take place _within_ some movement (such as an anti-war movement) which has more imediate attraction. (I suspect that in 1970 no one would have paid much attention to the drum-beating for Earth Day, which sort of kicked off the Environmental Movement, if they had not been accusomed to such public activity by the civil-rights and anti-war movement. It may be the same with a movement to demand national health care.)
I hope I've said enough at least to define two points.
1. Wankery over the accuracy or inaccuracy of polls does us no good as leftists but merely gets us involved in endless empirical arguments which aren't going to change anyone's views of the world or anyone's participation in politics.
2. We _can_ make use of polls (regardless of how precisely accurate they are) if we view them in terms not of the mostly irrelevant question, who is the majority and who is the minority on this or that matter, but in terms of what they say about the (potential) size over time of those we can (again potentially) involve in mass left activity.
Carrol