jerseyhoya wrote:

JUburton wrote:No idea what happens tomorrow but I'm having a good lol if may loses seats.

Is very odd. Basically every projection I've read - and I've been avidly following for a while and immersing myself in this all week - is calling for an expanded Tory majority. Except for YouGov who are very smart people and are predicting a hung parliament.

Think I'm going to be too busy at work tomorrow to write a 3000 word essay on how fucking weird the UK polls are, but a few quick bullet points.

-In the US we have voter lists. We can order sample to call registered voters rather than randomly dialing numbers. All campaigns do this, while most media polls rely on random dialing. This goes further in ensuring you're talking to actual voters when conducting interviews. And on that sample we know what elections you have voted in in the past. So if we want to create a likely voter model we can combine a voter's self reported likelihood of voting with their actual demonstrated likelihood of voting based on past behavior. This is obviously not perfect, but it is two massive massive advantages over UK pollsters.

-The UK election is of 650 constituencies, but basically no polling is done at the constituency level. And very little polling is even done at the regional/national (in the case of Wales/Scotland) level. It's like if 90% of polls in the US were nationwide, with even the subnational polls conducted at the regional rather than state level. It seems from things I've read the Tory vote allocation is going to be more efficient this go around than 2015. Boundaries as drawn currently actually favoured Labour at least through 2010, so not talking gerrymandering. But if marginal seats in the Midlands and the North of England experience larger Tory swings than the national average due to UKIP switchers or people hating Corbyn or whatever while Labour make gains in London seats they already hold and Southern seats which were less Leave friendly they can't flip, can see how losing by 7% this time might be worse than losing by 7% last time. And this is just utterly not being looked at by national polling. Of course, this is relying on party canvassing, which anecdotal and flawed in its own way.

-Almost all UK polls are conducted via panels online. This creates a number of problems insofar as you need to be online and often need to care enough to join the panel to be contacted. Weighting is used by pollsters to correct for this, but if the average 22 year old voter who is a member of an online panel is politically different than the average 22 year old voter who is not a member of an online panel, there's really limited options to fix that. At work I'm kind of an online polling evangelist, but if the sole method in tracking a horse race it still comes with serious liabilities.

-The generational gap between those 34 and younger and those 55+ is fucking insane. And determining what share of the electorate these groups will make up is mostly hocus pocus guesswork, and is the determining factor differentiating most of these polls. Younger people are SAYING they will vote in proportion similar to older people. This is getting Labour within a few points of the Tories in the polls where they're going by self reported vote likelihood. History shows this is very unlikely to happen. The polls with modeled turnout based on past turnout show a historic landslide.

-The polls allocate undecided voters by a variety of methods. Past voting history, demographics, etc. But they're making guesses. We, generally speaking, don't do that. I mean, we might model out the whole electorate internally, but public polling isn't so presumptuous. Report undecided figures. As 538 demonstrated so well this year higher levels of undecideds and 3rd party vote share = higher uncertainty. The way UK polls are reported, I have no fucking idea what that number is and how I should feel about the election.

Basically UK polls make a shitton of assumptions US pollsters don't have to, and US pollsters miss badly enough as it is. And so it's probably about as likely that May gets a 100 seat majority tomorrow as she loses seats, but the tails are very, very long so you may get to laugh. But I hope you don't because he's a dirty commie.

pacino wrote:"Of course Trump should expect loyalty from Comey and also what about LORETTA LYNCH?!?!?!?!"

"This is fake news and also who cares anyway!"

"Liberals are just trying to hide their big electoral loss!"

"Trump is a political novice, unlike the swamp people like Comey and the Democrats. We should expect a bumpy start. GIVE HIM A CHANCE."

just cut and paste these taeks if you want to converse with your friends

LANSING, Mich. (AP) — Legislation up for a vote in Michigan would let hunters kill frogs all year round and lift a ban against spearing frogs by using an artificial light.

The Senate is expected to pass the bill Thursday and send it to the House for consideration. Michigan now prohibits killing any species of frog between mid-November and late spring.

It also bars frog-spearing — or "gigging" — which is popular among young people in southern states, according to a legislative analysis.

The bill's opponents say frog-spearing is inhumane and frogs need protection to breed.

Supporters, however, say the legislation would promote outdoor recreation and it makes no sense to prohibit frog-spearing with a flashlight because hunters already can use artificial light to net or hook frogs.

jerseyhoya wrote:

JUburton wrote:No idea what happens tomorrow but I'm having a good lol if may loses seats.

Is very odd. Basically every projection I've read - and I've been avidly following for a while and immersing myself in this all week - is calling for an expanded Tory majority. Except for YouGov who are very smart people and are predicting a hung parliament.

Think I'm going to be too busy at work tomorrow to write a 3000 word essay on how fucking weird the UK polls are, but a few quick bullet points.

-In the US we have voter lists. We can order sample to call registered voters rather than randomly dialing numbers. All campaigns do this, while most media polls rely on random dialing. This goes further in ensuring you're talking to actual voters when conducting interviews. And on that sample we know what elections you have voted in in the past. So if we want to create a likely voter model we can combine a voter's self reported likelihood of voting with their actual demonstrated likelihood of voting based on past behavior. This is obviously not perfect, but it is two massive massive advantages over UK pollsters.

-The UK election is of 650 constituencies, but basically no polling is done at the constituency level. And very little polling is even done at the regional/national (in the case of Wales/Scotland) level. It's like if 90% of polls in the US were nationwide, with even the subnational polls conducted at the regional rather than state level. It seems from things I've read the Tory vote allocation is going to be more efficient this go around than 2015. Boundaries as drawn currently actually favoured Labour at least through 2010, so not talking gerrymandering. But if marginal seats in the Midlands and the North of England experience larger Tory swings than the national average due to UKIP switchers or people hating Corbyn or whatever while Labour make gains in London seats they already hold and Southern seats which were less Leave friendly they can't flip, can see how losing by 7% this time might be worse than losing by 7% last time. And this is just utterly not being looked at by national polling. Of course, this is relying on party canvassing, which anecdotal and flawed in its own way.

-Almost all UK polls are conducted via panels online. This creates a number of problems insofar as you need to be online and often need to care enough to join the panel to be contacted. Weighting is used by pollsters to correct for this, but if the average 22 year old voter who is a member of an online panel is politically different than the average 22 year old voter who is not a member of an online panel, there's really limited options to fix that. At work I'm kind of an online polling evangelist, but if the sole method in tracking a horse race it still comes with serious liabilities.

-The generational gap between those 34 and younger and those 55+ is fucking insane. And determining what share of the electorate these groups will make up is mostly hocus pocus guesswork, and is the determining factor differentiating most of these polls. Younger people are SAYING they will vote in proportion similar to older people. This is getting Labour within a few points of the Tories in the polls where they're going by self reported vote likelihood. History shows this is very unlikely to happen. The polls with modeled turnout based on past turnout show a historic landslide.

-The polls allocate undecided voters by a variety of methods. Past voting history, demographics, etc. But they're making guesses. We, generally speaking, don't do that. I mean, we might model out the whole electorate internally, but public polling isn't so presumptuous. Report undecided figures. As 538 demonstrated so well this year higher levels of undecideds and 3rd party vote share = higher uncertainty. The way UK polls are reported, I have no fucking idea what that number is and how I should feel about the election.

Basically UK polls make a shitton of assumptions US pollsters don't have to, and US pollsters miss badly enough as it is. And so it's probably about as likely that May gets a 100 seat majority tomorrow as she loses seats, but the tails are very, very long so you may get to laugh. But I hope you don't because he's a dirty commie.

slugsrbad wrote:Bracing myself for nothing substantially new, but remembering that it will now be all under oath which will go underlooked by the masses.

Social contact is a particularly effective way of reducing anti-gay sentiment because gay and lesbian identity is independent of the socio-economic, ethnic, racial, religious and regional divisions that separate Americans on other issues. While coming out is more complicated in certain parts of the country and within certain communities, gay and lesbian people are members of every social class, ethnicity, religion and race. White Americans are much more likely to have a close friend or family member who is gay than black, even though black Americans vastly outnumber gays and lesbian people. (I know these groups are not mutually exclusive.) Since the early 1990s, Americans collectively met and welcomed many more gay and lesbian people into their families and social circles.

thephan wrote:Pac, are you forecasting?

slugsrbad wrote:I'm very excited for this to devolve into Russia vs. Hillary

pacino wrote:

thephan wrote:Pac, are you forecasting?

no, what i wrote is based on a 'source close to the White House' stated today. Yesterday Trump's lawyer used the testimony to say Trump was vindicated.

pacino wrote:

slugsrbad wrote:I'm very excited for this to devolve into Russia vs. Hillary

Burr obliged you