You commission factfinding polls to get a legitimate idea of where the public stands on an issue. For example, if you are representing a brokerage house, you might commission a survey to find out whether (or how much) the Enron case will discourage people from putting their money in the market.
You commission manipulative polls to gather “factual” information that you can use to build your case. For example, if you are representing a generic drug manufacturer and you are lobbying government for more freedom in the market, you might commission a poll that would “reveal” that an overwhelming number of people would prefer cheaper drugs.
In trying to distinguish between the first and second kind of poll, it's helpful to look at the questions (Were they phrased to elicit a specific answer?) and to consider whether the the information might have been useful for anything other than manipulating public opinion. Be particularly wary of any poll that trumpets obvious but not very illuminating conclusions (eg., people like cheaper drugs).
Now, consider this story  from the National Post, which reports that
Fifty-two per cent (of business leaders) said they support the Asia-Pacific Partnership, which has voluntary emissions targets aimed at controlling global warming, while 24 per cent of those surveyed said they support Kyoto.
Notice, first of all, the general spin in the story, the degree to which the writer is building a case for the Asia Pacific Partnership rather than impartially reporting the news. For example, “Unlike Kyoto, the partnership does not penalize countries that do not meet emissions targets by forcing them subsidize those that say they do.” (My emphasis.) Notice the writer's own emphasis, choosing to pick out that “Eleven per cent of those surveyed said they have some doubt about whether global warming is taking place,” rather than focusing on the apparent (and interesting) corollary: that 89 per cent of business leaders believe the climate in changing.
Pay attention to the quality and reliability of the information. For example, half the sample (49 per cent) reports that they “believe global warming is taking place but they are not sure why or whether it is a normal variation,” and of that ill-informed half, 51 per cent endorse the Asia-Pacific Partnership. The National Post would have us believe that, within their business leader sample, people who have not even formed an opinion on the cause of climate change have already selected the best mechanism for dealing with the problem.
The reporter's agenda is clear enough, and it doesn't seem to encompass impartial reporting of the news. But consider finally that the National Post commissioned this poll: NP management paid for this information. The poll's most obvious result is that people who don't know much about climate change don't care much about how it might be solved - but when pressed to answer a pollster on the issue, they think that voluntary agreements sound better than penalties.
Does that advance the public discourse or does it merely give the National Post a manipulative “fact” with which to encourage Prime Minister Stephen Harper's abdication of responsibility on this issue?