About 15 national political opinion polls were wrong for a year. That is about 600 erroneous polls. The reported error based on sample size was about 4%+/-. Most of these 15 were off by 10%. Wrong 600 times. Two polls were right, one was the LATimes tracking poll and the other was Rasmussen.
Because of the stability of the poll results over time, they never waivered, I can make one factual statement: the universe these pollsters were sampling was not voters in the United States (except for two).
So what universe were all these bad polls sampling?
They were sampling the fake Universe of Americans as created in the fetid minds of the Democrat Party. Democrats believe that people vote based on their skin color, ethnicity and genitals.
Therefore the bad polls specifically sampled the ‘correct proportion’ of blacks, hispanics and women based on the past two elections. The two good polls sampled the Trump universe: ordinary people who voted in the past based on random sampling and self definition as party affiliated or not. All polls of the whole country weighted the results based on the Electoral College.
The reason the bad polls were wrong is that most people do not vote their skin color or genitals. More white women voted for Trump, despite their genitalia. The women who were black or hispanic ignored their genitalia and voted only slightly more than 50% for Clinton. But they had mostly voted for Bernie the White male Jew hater in the primaries against their genitals.
Also it turns out that only first generation hispanics vote heavily Democrat. An older hispanic or a second generation hispanic was more likely to vote for Trump. Skin color seems to wash off with age from the Democrat perspective.
Enough. You get the point. Nearly all the polls were in the thrall of Democrat racist ideology so they sampled the wrong universe and got the election results terribly wrong.
The worldview of survey researchers influences the survey results. Nearly all pollsters are Democrats.