prediction of polls Just as an example Scotland has poll to decide whether they need to be independent from UK or not. Here is BBC's summary of different polls:
http://www.bbc.com/news/events/scotland-decides/poll-tracker


It looks like as this progresses the undecided fraction is reducing and yes component is increasing. Provided  time and information we have is there a simple way we can predict the result with certainty ? 
 A: You need to have some indication of the uncertainty of a poll, for starters. And you want to have some idea of how polls from different organization differ. Unfortunately the bbc chose to allow only to show one poll at a time, so we have to click to change pollster results, and somehow "remember" what the last ones were. That being said, a couple of the pollsters have similar trends in results (Survation, TNS-BNRB, You-Gov), with results from all three suggesting the tightening of Yes/No. They disagree on the "Don't know" category. If there were more polls taken, and one could be sure that the pollsters were independent with no vested interest in one opinion or another, then we could get a pretty good idea of the final result. 
If you want to see our plots of US election polls, point your browser to 2012 US election polls 2008 election polls. There are many, many more polls taken, and you can see there is quite a bit of variability at any time point, and there are definite swings over time.
