As listed on pages 210 to 214 of the draft Investigatory Powers Act 2016.
Also, interesting titbit from page 37:
EUROPEAN CONVENTION ON HUMAN RIGHTS
[Name to be replaced] has made the following statement under section 19(1)(a) of the Human Rights Act 1998:
In my view the provisions of the Investigatory Powers Bill are compatible with the Convention rights.
Really? Who? And is that even relevant (not that it makes a difference) if you are not part of the EU?
Margin of error != undecided. Not even close. 'Undecided' is a valid response, along with 'Republican', 'Democrat', 'other', and 'no vote'. Margin of error is an estimate of the range of possible values the true population mean could have relative to the sample mean. So when someone says the 95% confidence interval of a value is between x+y and x-y, that means they are 95% certain that the true population mean is within x +/- y of their estimate. Or if they ran the experiment 100 times, then they would expect to be reasonably close around 95 times, and fuck it up about 5 time. Scientists also use the confidence intervals to decide if the results are 'significantly' different or not. For example Clinton 47%, Trump 46%, margin of error 2%, means that the true values could be Clinton anywhere from 45% to 49%, and Trump anywhere from 44% to 48%, thus the confidence intervals overlap and the difference is not significant. True values could be 47% vs 46%, or 45% vs 47%, or 46% vs 46%...
I am sure most polls use statistically rigorous and valid methods. It's just really hard to get a truly representative sample. It's not feasible to poll the whole population, so you try to take a random sub-sample. Where do you even get a truly representative random sample of people? Many people don't answer the polls, but do vote. Some people answer the polls, but don't vote. Some people change their minds between polling and voting. Some people are simply not contactable (I don't have a phone for example). Some people lie to mess with the poll.
And when you add the electoral college system to a tight race, then all bets are off, because who most people vote for != who receives most votes!
The decision doesn't have to be logical; it was unanimous.