>If you are relying on polls, what are the demographics?

It varies per poll but they try to get a sample representative sample of likely voters.

>number of people polled?

They median seems to be around 800 people in the samples. With 500 at the low end and 1400 at the high end samples. Looking at past statistic courses that’s a large enough sample for this type of poll.

>what is the margin of error ?

+/- 3.5% in most of the polls. I think taking an average of multiple polls helps with achieving greater accuracy in this area.

>Did you figure for the coming out of the wood work population / those who usually do not vote ?

Most polls try to select likely voters. Trying to account for people that don’t normally vote sounds more like precognition.

>Remember, Trump brought out more primary voters than anyone in history.

Primary turnout has a very mixed history for predicting general election turnout and is historically an unreliable indicator. In Trumps case he has the most Republicans vote for and against him that any other Republican before. So it’s an even more mixed indicator with him.

>Polls have nothing to do with the electoral process.

Remember that when going to your polling place on election day.