You believe in polls. You’re in the minority among those in the scientific community.
The polls are not reflective of the nonrespondents which typically make up 90% of a poll based on a scientifically acceptable sample plan.
Independent post-election analyses of the 2012 race pinpointed the 6 million+ demographic ‘Perot Bloc’ of conservative blue collar voters who chose not to vote for Romney or anyone else. Trump is accurate when he says Romney failed to inspire. With that bloc, Romney would have won handily.
The media polls were not accurate as you say. They could never be accurate. Calling them accurate is like claiming global warming models had NYC underwater in 2008 (which was called) and then observing a hotspell and draught to back up the warming predictions.
The Perot Bloc were not sampled by media because they were factored as conservative GOP that would vote for Romney and importantly they were ignored to bolster the media favoured candidate. The media polls did not bother to poll the Perot Bloc so they would never have known regardless of the election outcome. In other words, the media polls got lucky.
Romney’s people factored the Perot Bloc in as a sure vote and Romney’s polls had him winning the day before the election, so much so that he remarked how he could “smell victory”
The 2012 election outcome was not reflected in the polls, it was reflected in the nonrespondents that sat home.
Michael Barone who had hitherto been one of the most in-depth accurate assessors of voting sentiment was off by a very large amount and he attributed his failure to his reliance on landline polling.
Polls can at best be said to reflect those willing to engage of those who are sampled. The 90% nonrespondents may or may not engage in terms of actual voting. If the nonrespondents decide to vote, the polls will look to be ridiculous.
Polls are not legitimate, but they serve as useful shapers of public opinion thereby becoming self-fulfilling. If they were for information gathering only, and never published or broadcast, their social Heisenbergish effect would never be seen and the end results would differ greatly from what actually occurs. The risk to all polls are the undersampled blocs that may or may not be inspired to vote. This was evident in 1980 and 1984
The correlation you are touting is similar to standing outside a sports stadium and monitoring the roar of the crowds. One simply doesn’t know what the score is and doesn’t know if the guest team has their own cheering section. But should the favored team win, the outside ‘roar’ monitors will surely claim they predicted the outcome. Ridiculous.
The only saving grace of polls is the quasi-independence of some from others. That is the only attribute that can possibly restrain cheating.
The bottom line is that polls can't predict voter turnout. It doesn't matter if 60% say that they would prefer a Republican over a Democrat if half of those aren't inspired enough by the Republican to bother to vote while Democrats get 80-90% turnout. Running candidates like McCain, Romney, or Jeb guarantees that a big block of Republican-leaning (but independent) voters just stay home on election day. These people aren't going to turn up at the polls for a Jeb Bush or a Marco Rubio. They would show up for Trump or some other populist candidate who has something to offer besides GOP establishment boilerplate, which is becoming increasingly like Democrat boilerplate.
Remember the FNC poll the sunday before election day that had Kerry five points ahead of Bush.
Please show me two pieces of evidence that say that “four million” stayed home in 2012. I have searched and searched for this. It’s a bogus claim based on a single IN article that actually had no real statistics. I could find no such statistics in OH, where Romney strongly outperformed McCain.