There's a great new article on the Monkey Cage. https://www.washingtonpost.com/news...normal-actually-voted/?utm_term=.5e50b5e53bf6 If black turnout in Wisconsin and Michigan reflected merely the average decline in black turnout in 2016, Hillary would have won both states and Trump would have won the Electoral College by only 12 votes - that would also have resulted in a 51-49 Republican majority in the Senate. Food for thought.
Furthermore, according to the study, if 2016 turnout had mirrored 2012 turnout more, on E-Day + 1, we would be looking at this map when we woke up: That's a scenario that is much more...interesting. (Share on 270towin wasn't working for me, anyone else who can make it work please do)
This is the map I get. Is that what you wanted to share? Question: Is it possible to measure the impact of new voter registration laws in the states that were very close? I think that WI was one of the closest states and I recall reading that several thousands of voters were denied their right on election day.
Yeah, this map would be what we woke up to, where Hillary defended small leads in Florida, Pennsylvania, Wisconsin, and NH, while Trump defended small leads in NC and GA.
https://www.vox.com/the-big-idea/2017/7/14/15959682/evangelical-mainline-voting-patterns-trump @American Brummie I thought you might find this interesting.
First of all, these voters were never Clinton voters, so they followed what they wanted and did not have to vote for who the Democratic party told them. Some asked for more detail on how Sanders primary voters behaved in general. This graphic shows this, including small % who abstained 2/n pic.twitter.com/iOjKr7eoYJ— Brian Schaffner (@b_schaffner) August 23, 2017 900164807961305088 is not a valid tweet id https://www.yahoo.com/news/bernie-sanders-voters-helped-trump-210427684.html
Question: What is like the average Berniebro voter? White, middle aged and high school education only?
@American Brummie https://www.nytimes.com/2017/11/16/opinion/trump-white-voting.html?_r=0 We already "knew" this, but now it's backed up by science! Basically, Trump didn't do as well with white people who actually interact with POCs. But he dominated in super-white areas. "The 2016 results suggest that residents of a diminishing number of decisively white American towns and small cities — even those which supported Barack Obama in 2008 and 2012 — can now be politically mobilized around race, ethnicity, multiculturalism and immigration. None of the nation’s whitest municipalities and counties — especially those in the Trump-voting Midwest — is immune." Here's one example. "Or take Elk County in Northwest Pennsylvania. In 2000, Elk had a population of 35,112, of whom 34,746, or 98.96 percent, were white. 62 residents were black and 142 were Hispanic. By 2016, Elk’s population had fallen to 30,480, 97.5 percent of whom were white. 152 were black and 244 were Hispanic, increases of 145 and 72 percent. In 2008, Elk County residents voted 51-47 for Obama over McCain. In 2012, the county backed Romney over Obama 57-41. In 2016, Trump swept Elk County, 69-27."
copy cat copy cat I know you are a dirty rat In all sincerity, what is your point? Comparing Trump to Romney, Trump lost white voters who lived around POCs, and picked up white voters who didn't. I and many others suspected that, but now someone has done the work and verified it. This enabled Hillary to really rack up PVs in minority heavy states like Cali and Texas and Georgia without changing the EVs. That's how Trump snuck out his narrow-not narrow EV win. (He won 3-4 states by very small margins, but didn't need all of them to win election. That's why I say narrow-not narrow.)
Going back to the initial question: What Polling Got Wrong, I pulled up the RCP state polling and compared it to the actual outcomes. Here are the basic patterns of where polls were wrong, starting from what was most off: 1) Little Information. The most significant incorrect margins were in states which did not have good polls conducted near Election Day. This is no surprise. If the last well conducted poll was five weeks earlier, for example, Trump was at his nadir. States which are not going to be close often don't have very many reliable polls, which further draws these off mark. In order of accurate to inaccurate, these states included CT, NE, MS, RI, SC, MT, AR, MD, DE, AK, KY, TN, SD, OK, and WV. West Virginia, Trump's second biggest win, was 24 points larger than the most recent (but quite old) poll suggested. On the other hand, Connecticut and Nebraska were within a couple points, but this is more of a blind squirrel effect than anything. On average, these states were about ten points off actual results, while no other state was off by more than 8.2. (Some states had no good polls: WY, ND, AL, HI, DC; but none were close or were expected to be.) 2) Reluctant Mormons. There are two states with at least 20% LDS: Utah and Idaho. Both were off by almost 8%. While Mormons were generally unfavorable towards Trump, they are reliably conservative Republicans, so some came around to the party's nominee. Not enough to prevent the worst Republican performance in Utah since FDR, but enough that it was significantly off of late polling data. 3) The Midwest. These states were all off by pretty significant amounts. Iowa and Ohio were supposed to be close Trump wins, but turned out solid for him. Indiana and Missouri were supposed to be solid, but were landslides. The upper Midwest was supposed to be Clinton's, with expected margins of 3.4% in Michigan, 6.5% in Wisconsin, and 9% in Minnesota, but each were in the tossup range. Much of Pennsylvania is similar to the Midwest, and had a smaller shift his way. Much of the problem here was modeling; how 'Likely Voters' were determined by polling agencies did not line up with how various demographic groups in the area actually turned out. If the raw data was interpreted with the actual demographic shares measured by exit polling, suddenly the polls predict something close to what actually occurred. Only Illinois bucked this trend, because it fell into a different category... 4) Big, Urbanized States. There's one category of places which broke hard for Clinton--the largest metro areas in the country. This meant she overperformed in California (Los Angeles, SF Bay), Illinois (Chicago), and New York (NYC) gave her 5.5% or more than expected, while Texas (DFW, Houston) and New Jersey (NYC) delivered an extra 3%. Places with many smaller cities and relatively populated rural areas mitigated this surge, as seen in Virginia, Pennsylvania, Florida, Georgia, and Michigan. The remaining states all fell within 3.3% of polling results. Little trends can still be seen, like Clinton doing better than expected in much of the west or Trump in the northeast, but I'm not reading too much into results which generally fell well within the margin of error.
The big early hint about the Midwest went unreported on Election Night when Indiana was called immediately. Nobody on any broadcast I've seen, including right-leaning ones, remarked on it but it should have been thirty to sixty minutes before the state was called, instead of immediately when polls closed.
Back in the 20th century, Indiana was frequently called early for the Republicans, even in big Bill Clinton victories.
Yes, but the way it's been Republican has shifted over time; the county map from each election displays this shift; the river counties went from light blue to deep red, while Indy and Bloomington went the other way. But what I'm talking about is a more widely applicable principle. The difference in late polls indicates roughly how long a state should take to be called, and an 8-9% win takes around 90 minutes, while a margin twice that is called at closing. Indiana being so significantly off was a harbinger of trouble for similar states.