I think that a snap-shot impression in political poll results says very little, particularly as political variables are so much more prone to volatility. Therefore, I think it is always important to look at the trend in polling results over time, and consider what specific intervention would change the trend (since if we treat politics in an objectivistic, deterministic, reified way we are being just pundits and in no position to affect outcomes in any way). That is how a politician looks at it.
According to Gallup, Bush's approval rating is 50 percent "approve", 47 percent "disapprove" at the moment. The 3 percent difference isn't very much, and it takes little to tip the scales. This combines with a massive increase in the state of dissatisfaction about the state of the country and a massive decline of economic confidence (see http://www.gallup.com/poll/stateNation/ ). Thus, given severe public doubts, this should alert us to the kinds of events, that would cause a further shift or change in public opinion in the future. In addition, one ought to check the survey questions used, since, the exact formulation of the questions, especially in attitudinal surveys, can enormously affect the results, and these results can in turn influence public opinion or impressions, such that public opinion is moulded and shaped by feeding back a certain scheme of abstraction, which provides frames of reference implying what the legitimate dimensions and boundaries of valid opinion are. One of the most common errors made in attitudinal surveys consists in asking questions which respondents cannot answer or decide on as stated, with the result that they consent to choose an answer which "they think" is closest to their opinion, but which does not reflect their true opinion, because it abstracts from their true motivational structure, true context or overall judgement, or is influenced by the surveyer. At the simplest level, it is already very different story, for example, if I ask a respondent a question of the type "which of these options most closely reflects your opinion" than if I ask "is your opinion actually x, y, or z, and choose one only from these options" (which a critical portion of respondents may in fact not be able to do). It can sometimes be much more revealing to ask a question of the type "what specifically would change your opinion of this issue ?" since opinions are much more subject to change than other human behaviours or characteristics. "The philosophers have interpreted the world, in various ways - the point is to change it" - Karl Marx J.
