I think she does a poor job arguing her position.

https://projects.fivethirtyeight.com/pollster-ratings/

sort by pollster rating, and you see that the polls from abc news/washington post are among the highest rated, they are one of only 6 pollsters with an A+ rating. they are found to have a democratic bias of 0.6 percentage points (on average, predicting the margin of the democratic candidate over the republican one too high by 0.6 percentage points), and have pretty good performance relative to other pollsters (advanted +-, predictive +-, negative values are better for these metrics).

now, if you look closely, their numbers are probably the worst of the 6 A+ pollsters, but thats still a really good performance.

This post was edited by Black XistenZ on Feb 6 2019 10:25am

i mean, a lot of the 2016 data was basically correct inside of the margin of error for the individual districts, and HRC won the popular vote.

the Huffington Post polls with 98% chance for HRC to win, sure. but a lot of them showed Trump standing a chance district by district, if his "hidden Trump vote" equated to a few percent.

the main surprise in 2016 wasn't the % of the vote that Trump got (what a poll shows) it was his ability to maintain a winning percent in key purple areas nationwide.

We still saw red states stay red and blue states were blue. accurate polling really should only address deeply purple areas, that's all that matters in a presidential race.

before you waste time digging im aware that you can find snapshot in time polls where HRC was favored far too heavily.

This post was edited by thesnipa on Feb 6 2019 10:26am

the main problem, the main thing the pundits got wrong in 2016, was not the polling itself, it was with translating the polling numbers into a probability to win the presidency.

in essence, there were two contradicting trends, which btw were also in effect in 2018:

- rural america trending towards republicans and also showing higher turnout numbers than in the years before trump
- urban and urban-suburban america trending blue


these trends cancelled each other out to a certain degree in the national popular vote, so that someone who was only looking on the surface thought everything would still move along the known patterns, and hillary was simply winning by a solid but not overwhelming margin. with this understanding of the dynamics of the race, they also thought that the EC would be helping her and not him. people just didnt realize that trump's coalition was much more efficiently distributed in the electoral college than that of Romney or McCain, turning the EC into an advantage for him. they also failed to account for the fact that trump was activating his base better than hillary was turning out hers.

This post was edited by Black XistenZ on Feb 6 2019 10:45am

Sources?

indeed, its an issue with expediency. taking nationwide polls and extrapolating data will always be problematic.

and like anything, the "correct" approach is so time intensive few if any would sit through the explanation or read through it on print.

Stage 1: announce all deep red/blue areas, cover any issues that may arise with the state generally, then award Electoral points to help eliminate variables.

Stage 2: identify any outside candidates for flips. Virginia, etc. And cover the polling in that area, cover rural vs urban differences, cover the electoral votes at play and give a value in % chance they flip.

Stage 3: go state by state in the purple areas covering the % chance they go one way or the other, rural vs urban, etc.

Stage 4: cover potential scenarios of what happens if x, y, and z state go the way they're projected vs flipping. cover basically every electoral outcome that doesn't involve the known outcomes in red/blue states.

it would be exhausting and so much wild speculation.

you also have to account for correlation across state or district borders. the gist of the story in 2016 was that hillary had to underperform with just one single group of voters, white non-college educated folks, in one region of the coutnry, the midwest, to lose the race via the EC.

these voters had backed the democrats for decades, and they had just voted for the first black president during the previous two elections, so no one thought trump's racially charged rhetoric would appeal to them. and imho it didnt. but what almost everyone didnt see, or did not want to see, is that his economic message would resonate deeply with these rust belt voters. imho, that's what won him the decisive states and thus the presidency.

i agree generally. if had to play the role of super computer and compute the effects of billions of factors to find the most influential ones, at a guess, i'd rank them something like:

1. Comey Letter
2. Lack of time/money spent in OH, PA, MI, WI
3. decades of public service record for HRC

I dont think Hillary spending more time in these states would have made a difference. She was just the wrong person with the wrong platform to appeal to them.