Fickle public opinion
Polls are pretty bad indicators of long term anything.
I mean for the most part you get several different organizations asking different questions of very few but different people.
The realclearpolitics aggregate poll you cited in another thread, most of the polls were questions asked of like 1,200 people, some fewer (if I remember correctly) like 300.
There was only one poll that questioned something like 12,000 people.
Not to mention, people ♦aren't generally being asked "do you openly support hillary over trump?" or "are you going to vote for hillary or trump," but more convoluted questions.
And a lot of poll questions are worded in ways to lead to a desired answer.
There are around 320,000,000 people in the us, speaking dozens of languages from diverse cultures and backgrounds.
320,000,000
12,000
1,200
You really think polls are an accurate indicator of anything all that relevant?
Let alone long term trends?
.003% of the population, at best, accurately represents 100%?
Polls are misused by the media to represent things that aren't really asked.
Polls are misused by politicians for propaganda and psych warfare purposes.
Then people use those headlines for confirmation bias.
So
Fickle public opinion
IMO not really.
Determining accurate public opinion is flawed, and made worse trying to determine anything over the long term.
Polls and public opinion numbers are good for confirmation bias and propaganda, that's about it.