Although it is not completely foolproof, I think polling can be trusted as a fairly accurate predictor of a political outcome. The polls provide information as to the popular opinion of candidates or issues, which provides a good indicator of how these individuals will vote in any pertaining election and thus the outcome.
Polling ahead of recent elections has proven to be unreliable. This has been the case in U.S. elections, as well as elections in other countries. Therefore, polling should not be trusted as an accurate predictor of political elections. The media should not try to use polls to sway public opinion about political candidates, because these polls are so unreliable.
Polling should not be trusted because the only people who participate in polls are those who choose to do so or polls are usually found on websites that typically lean in one direction or another thus making the participants all a little like-minded and not based on the general population.
Polling has become a popular means of representing outcomes of political races before they actually happen, but one should not rely upon these polls to predict certain outcomes. Polls are merely an informal and unscientific method used to forecast which candidate, for example, has the better chance of winning a particular political race.
The problem is that often media outlets and even the candidates themselves often use poll results to draw conclusions that aren't accurate. People's opinions change drastically from one point in time to another and many polling results are based on the opinions or input of people who don't actually vote in the official election. While polling serves a purpose and provides interesting, timely commentary, it should not be trusted as an accurate measure of a final outcome.