Campaign coverage in the United States is full of polls, showing that this candidate is up and the other is down, the margin here is widening and the margin over there is narrowing. There are lots of different polls from different organizations that have adopted different procedures.
Is polling scientific? Yes and no. There are certain insights provided by the sciences that can inform proper polling. But the result of a poll is not a discovery, it is a data point. The real (political) scientists are those who look at all the data points available.
In other words. the polls, precisely because they are taken by distinct organizations, with different methods, tend to provide each other with a context. Approached thoughtfully, they provide a political scientist with evidence in much the same way that temperature readings from lot of patients in different circumstances provides medical science with evidence.
Strange New Worlds:
RealClearPolitics.com publishes the averages of different polls. This is a great resource for people who can’t make following and interpreting poll data their full time employment. It is the dirty-but-quick way to step back from the independent data points and to look for what scientists call the “best fitting curve” amongst them.