it's indeed too early and nobody should get excited about anything. but, the averages are useless and should not be consulted. this is a useful way to measure responses to advertising, but it is a very, very poor methodology for political polling.

the general election is a snapshot that will reflect very short-term opinions, not the cumulative response of weeks of measurements. the polls you get on any specific day are, in fact, likely an accurate reflection of what people are thinking on that specific day. the error is in deducing that people will therefore think that same thing tommorrow.

voting choices are going to be especially volatile when there are very few policy differences of substance between the candidates, as is going to be the case in 2016. worse, they're both widely despised. who is less reviled this week?

that holds today as much as it will in november. all the averages can do is create inaccuracies by blurring together the margins. again: the error is in the model. elections are not market research surveys. snapshots are better measures than averages, you just have to know that they're volatile when you're reading them.

http://www.huffingtonpost.com/entry/donald-trump-polls_us_575adbcbe4b0e39a28ad606c