Also he doesn't like it when people make generalizations about black voters.
Super Tuesday 4. I'm really feeling (The After Bern) March 22, 26 contests
- Thread starter NeoXChaos
- Start date
- Status
Tesseract
Banned
Yass!
OpinionatedCyborg
Member
A Princess Bride reference, eh? Ted Cruz approves: https://www.youtube.com/watch?v=S_mcie4Nasw
thanks for the clarification
Mmm I think you've missed the point of his model. What he's been trying to accomplish is whether or not we're at the point in the internet age where we can accurately or near accurately predict election results without pollsters and old-fashioned analysis. He's deliberately relying on social media trends and internet search interests and adjusting his model with each new primary's results. It's expected he won't be spot on...
But IIRC he is a statistician. He has been employed with other campaign trails before I think? I'll see if I can find the source on that. Forgot where I read that.
Anyway, the Michigan thing isn't a joke or something to brush off. His numbers were off the next time around. He even said he expects to come across that frequently though. He's adjusting as he goes.
That's just commentary on his actual predictions. But my previous post was based on the stats he shared on Arizona's demographics nothing about social media or search interest. I'm not claiming that's the reliable source of information. Because that's what he's trying to prove or disprove with this whole experiment.
superdeluxe
Member
I'm going to go out on a limb here and say sanders is not going to win
googleplex
Member
I'm 42 and black. I voted for Hillary.
FloydtheFathead
Member
Iunderstoodthatreference.gif
User 463088
Banned
No, I definitely get the point of his model, but I reject it on its face. The idea that social media or search trends are able to predict outcomes across hugely different demographics and regions is something that I can't get behind. It's going to automatically over-estimate support among younger voters, simply because older people aren't getting information from Reddit, Facebook or Google. He also can't use his model on the GOP race by his own admission. So, a model in a two person race, in which one of the candidates has an advantage based on methodology, is not something that I personally see a lot of value in.
Bing's been doing something similar, and they've only managed a success rate of 76%
Of course, I'm a total idiot and a jaded bitch, so my opinion is completely useless.
The Michigan thing was a polling error. There hadn't been a contested primary in Michigan since 1992. The likely voter screens were way off. The pollsters tried to "fix" that in the next five states and overrcorrected. There was a lot of weird situations that went down in Michigan causing that particular result.
perfectchaos007
Member
Phoenix is the US's 6th largest city so Arizona is a decent prize. Utah is meh
Where do claims like this come from? This information is just a quick Google search away. Arizona's median age is actually below the national average.
gutter_trash
Banned
it smells like Florida 2
So, after last week when his numbers were "off" by just a bit, he "adjusted" his model by adding 6 more variables to his previous 3 (hi shinra!), yielding this week's predictions:
In case graphics got you down, here's the text version of Bernie's win%:
AZ-56%
ID-79%
UT-75%
Later tonight we can marvel at its accuracy.
Yeah. Old people.
eBay Huckster
Member
probably the demographic splits of the actual early voting numbers, which are also publicly available
So, after last week when his numbers were "off" by just a bit, he "adjusted" his model by adding 6 more variables to his previous 3 (hi shinra!), yielding this week's predictions:
In case graphics got you down, here's the text version of Bernie's win%:
AZ-56%
ID-79%
UT-75%
Later tonight we can marvel at its accuracy.
All three of his predictions will be way off the actual result even though Bernie could win ID and UT tonight.
So, after last week when his numbers were "off" by just a bit, he "adjusted" his model by adding 6 more variables to his previous 3 (hi shinra!), yielding this week's predictions:
In case graphics got you down, here's the text version of Bernie's win%:
AZ-56%
ID-79%
UT-75%
Later tonight we can marvel at its accuracy.
I think those are his predictions after weighting his numbers a bit.
So this is what his model WOULD HAVE called for all past states knowing the information and data that it knows now. Basically, these are retrospective results, as the model was adjusted according to each state's results.
To me this is fascinating to watch him experiment with. You can disagree with the value behind it all you want, but I still find it fascinating nonetheless.
User 463088
Banned
Arizona is 10th when it comes to states with populations of people over 65, Early voting numbers show that people 65 and older made up 43% of the electorate. Those under 35 made up 12%.
Arizona is also older, on average, than Texas, Georgia, Louisiana and Mississippi, all of which Hillary won by huge margins.
There's not a direct correlation here, because younger people, no matter how many of them they are, simply don't vote in comparable numbers to older voters. We can see that in every single state.
eBay Huckster
Member
so is that what retroactive momentum really means
Early voting is always skewed toward older folks. Nursing homes, senior communities, the responsibility of the older generation itself...etc.
Generally, young and mid aged voters don't vote early.
eBay Huckster
Member
and generally, young voters don't vote
Did Hillary win those states decisively because of age or race demographics? The latter is what's agreed upon, right? Arizona not being a southern state like the rest kind of makes it an outlier in those comparisons.
User 463088
Banned
Based on current estimates, 70% of the people who are going to vote in Arizona voted early. An early vote that's been cast is worth twice as much as someone who absolutely promises to get to the polls. You win by winning the early voting. That's why Obama and Hillary's respective campaigns push it so hard. You can target your voters easier, you can run up margins. You know where you need to focus people on election day. Early voting is why Hillary won places like Ohio and Florida by such huge margins. Places without early voting (Michigan, for example) were closer.
OldAsUrSock
Banned
Though with the popularity of Bernie Sanders we see which direction America is heading politically once these people age and start voting.
Hellsing321
Member
WTF is facebanking?
That certainly makes sense. Has Hillary been winning most/all of the states so far that had early voting? I've been paying attention to the race but honestly don't remember which states had early voting and which ones didn't. Many were caucuses so far.
So primaries that encourage huge voter turnout favors Sanders, but Hillary's dedication to getting the early vote adds security for her numbers?
eBay Huckster
Member
maybe. we could also see pragmatism 2: pragmatize harder as a sufficiently large proportion of these people settle down. nothing's that set in stone at this point, except for climate change and the continued acceleration of automation
User 463088
Banned
She won them due to a combination of factors. In Texas, the only age group Bernie won was 18-29 year olds (who made up 20% of the electorate) Those 65 and older made up 18% of the electorate, and she won them by nearly 75 points. She won Latino voters by 45 points.
You can go state by state. Hillary's demographic strengths are with people of color, yes, but they're also with older voters. Building a coalition mostly around young, white voters is just not a winning strategy.
kitschykitty
Member
Spamming people on Facebook to vote for a candidate.
Facebook+Phonebanking=Facebanking
You Are Viewtiful
Member
Thanks for the replies. I think a lot of his campaign relies on getting super delegates on his side.
Also this Tyler dude is using way too many variables that rely on Internet traffic, that is going to make most the results skew younger. Also, instead of using median age (which includes people ages 0-18 in the calculation), it would smarter to use population pyramids to see which percent of the population is over 50 or under 50.
2 busy 2 vote. Gotta check the latest memes on imgur
TheAbsolution
Member
Spamming your friends and family with Facebook messages to go and vote.
And not just your friends and family. Friends of friends so strangers get random messages too.
I dunno but Nintendo took it out of their Western localizations and I am FURIOUS
Stumpokapow
listen to the mad man
adding variables at random to a regression leads to chance capitalization (overfit on training data) which lowers predictive accuracy. variables should be theoretically justified before adding, or he should use a regularized process (LASSO, ridge regression) to compensate for the overfit.
superdeluxe
Member
welp
Hellsing321
Member
And people think that works?!Spamming your friends and family with Facebook messages to go and vote.
And not just your friends and family. Friends of friends so strangers get random messages too.
User 463088
Banned
Actually, most of the states Bernie has won have had fairly low turnout. He won Kansas, for example, but got fewer votes in the entire state than Hillary did in one single county in Florida.
Some of the bigger states with early voting were Ohio, Florida, Texas and North Carolina. Hillary's early vote lead in Florida was enough to make it nearly impossible for Bernie to have gotten within 10 points even if pretty much no one had voted for her on election day.
Early voters tend to be older, women and minorities. someone who votes early has a chance to fix any errors that ay come up. For example, some people in Arizona have to vote provisionally because they never changed their party registration. Had they been targeted early by Bernie's campaign for early voting, they could have potentially had this fixed early in the cycle.
Early voting also shows which campaign has the better ground game. Getting people to vote early, tracking them and their ballots is a far better use of time than hounding people and saying "You can vote on this day." It removes excuses, it makes the process more convenient, and it lets the campaign know where they need to allocate resources.
For example, say you know that you expect 10,000 votes from Polk County and 10,000 votes from Orange county. Based on your early voting targets, you know that 8,000 people in Polk have already voted, but only 2,000 in Orange have done so. You know to shift your candidate and resources to Orange. If you're just sitting there on election day saying "We need 10,000 let's get them there in the next 24 hours" it's a lot harder.
If he's relying on SD's when he's 300+ pledged delegates behind, then he has already lost. There's not a melting snowball's chance in hell the SD's would vote for him over Hillary if she wins pledged delegates and popular vote, which she essentially has both wrapped up.
User 463088
Banned
His Superdelegate strategy is absolutely ridiculous. It's the same thing Hillary tried in 2008. A few months ago, Bernie was complaining about how terrible and undemocratic they are. Now he wants them to support him over Hillary.
His argument on CBS on Sunday was "Super Delegates should support me in the states that I won. In states that I didn't win, they should support me anyway because I'm more electable."
If a campaign is saying that stuff publicly, it's not a good sign for their internals.
His argument on CBS on Sunday was "Super Delegates should support me in the states that I won. In states that I didn't win, they should support me anyway because I'm more electable."
If a campaign is saying that stuff publicly, it's not a good sign for their internals.
Setting the stage for a Warren landslide in 2024
Wafflecakes
Member
At what time do we expect to start getting results today?
...or are the networks even going to bother covering it considering the events in Brussels?
...or are the networks even going to bother covering it considering the events in Brussels?
Well, it's possible it's another method of contact that will become more common in future cycles and is being pioneered now. The problem as I understand it is that you are not (and cannot) contact random people, as you would when volunteering for a campaign, so you're not hitting the people you need to. Mainly it seems to generate visibility; whether good or bad.
I don't think anyone has any data as to its efficacy, as it is unorganized and done outside the auspices of a campaign. As for the spam aspect, who knows. People are bombarded with campaign ads and calls which are basically spam, but they continue to be used. Maybe a future campaign will place a more scientific grounding under it.
- Status