Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[TempAcc]

Surely those polls pointing a 90% chance of a hillary win were all completely honest and used sound methodology, right?

[Criptfeind]

It's a pretty far cry from saying "People polled say that they think Clinton will win" (what you first linked too) to saying "Clinton has a 90% chance to win the election."

Edit: Actually also it's a pretty far cry from "Clinton has a 90% chance to win the election" to "Anyone who doesn't think Clinton has a 90% chance to win the election is either a white supremacist or a conspiracy theorist."

[TempAcc]

Hyperbole is a thing that exists.

[Criptfeind]

That doesn't make your example any less irrelevant though.

[penguinofhonor]

You know, people say you only have a 5% chance to roll a 20 on a twenty-sided dice, but I just rolled a twenty on the first try! How could I when everyone says you're basically guaranteed not to get a twenty? There must be some media conspiracy pushing this view.

[Zangi]

Technically, Clinton won something... the Popular Vote.  So everyone is a winner, yay!

It means jack-all when the winning that counts is the Electoral College.

[MetalSlimeHunt]

It doesn't count much for Clinton, but it does count in that Trump has less of a mandate than perhaps any other President. Even Ford was supported by procedure and by not being Nixon (and, for the record, had decent approval ratings for a good while).

[martinuzz]

The US think tank AMTI (Asian Maritime Transparancy Initiative), part of the Center for Strategic and International Studies has concluded, based on new sattelite images, that China has installed advanced weapons systems on all 7 of the islands it has created in the South Chinese sea. It was already know the islands were fitted with airstrips and radar systems, but added to that list are now advanced anti-air and anti-missile systems.

So it looks like, while the US and Russia have been quarreling over anti-missile shields, and how they will invalidate MAD, China went ahead and started building one.

Meanwhile, China says that "when someone rolls their muscles in front of you, you grab a catapult", arguing the islands' militarisation is purely defensive.

http://www.volkskrant.nl/buitenland/china-plaatst-zwaar-geschut-op-kunstmatige-eilanden-in-zuid-chinese-zee~a4434935/

https://amti.csis.org/chinas-new-spratly-island-defenses/

EDIT: hmm the amti article calls it CIWS, not anti-missile. Perhaps it's not anti-nuke then.

[PTTG??]

Ok so let's build a better political party. Among other things, we need to have strong internal regulations, because the entire 2016 election was the product of deeply malfunctioning party apparatus.

So. For one thing, we need to make sure that the party itself is a strong, functional democracy. No applause-o-meter appointments. No god-damned superdelegates. Our goal is to get a candidate who will best benefit the country and the world into office, and to ensure that that candidate has a strong backing.

Actual policy decisions are actually kind of irrelevant. It's more important that the party representatives use sound judgement than that they agree with me. After all, I'm wrong sometimes. So, what does that party look like?

[Neonivek]

I do LOVE how America uses Statistics

Essentially if Hillary had a 51% chance to win, it ACTUALLY means she has a 100% chance to win.

In no way does a 90% chance of Hillary Victory mean that she can possibly lose

[LoSboccacc]

You know, people say you only have a 5% chance to roll a 20 on a twenty-sided dice, but I just rolled a twenty on the first try! How could I when everyone says you're basically guaranteed not to get a twenty? There must be some media conspiracy pushing this view.

so it seems you have no idea what a confidence interval is or how sample size relates to it all. an outlier result doesn't mean you rolled a 20 on the first try, but you rolled a 20 for 20 times in a row, and that's why they claim it's either a rigged election, rigged polls, or social factors (i.e. people ashamed saying they'd vote for Trump) - it's not just a product of chance.

[martinuzz]

I do LOVE how America uses Statistics

Essentially if Hillary had a 51% chance to win, it ACTUALLY means she has a 100% chance to win.

In no way does a 90% chance of Hillary Victory mean that she can possibly lose

Half of the country voted Hillary, the other half voted Trump. The third half is left confused.

[Guardian G.I.]

Stupid tangentially related stuff: remember how Sergarr went full LCS many pages back? There's a person who's actually just like that IRL - Keith Olbermann from GQ. He's been triggered by Trump incredibly hard:



"RUSSIAN SCUM!!!"

[Harry Baldman]

so it seems you have no idea what a confidence interval is or how sample size relates to it all. an outlier result doesn't mean you rolled a 20 on the first try, but you rolled a 20 for 20 times in a row, and that's why they claim it's either a rigged election, rigged polls, or social factors (i.e. people ashamed saying they'd vote for Trump) - it's not just a product of chance.

It does not mean that at all. The chance for somebody to win the election was the proportion of possible results as calculated from the data that went their way. For Clinton that was 80-90% of projected real results, with 10-20% of projected real results showing a relatively unlikely Trump surge on Election Day and predicting exactly the kind of sampling error that would give rise to something like this. At the end of the day a good chunk of Trump's voters were missed by the majority of the polls. You can argue why that would be, but the fact remains that they were missed, and the confidence interval having a specific percentage value assigned to it accounts for exactly the possibility that the majority of your gathered data does not reflect the real distribution, which is never impossible and in this case wasn't even what you would call a sure enough thing in a biology paper under certain conservative estimates.

See, you're applying deterministic hindsight to a probabilistic attempt at foresight. A throw of the dice is also hypothetically a perfectly predictable process in which you, knowing the full initial conditions and having adequate tools, could easily predict the side they'd land on. The fact that it's practically impossible to gather these measurements in a real situation is why probabilities and confidence intervals exist.

Of course, improving methodology in light of this result is probably solid advice nevertheless, and has already very clearly occurred to the pollsters and statisticians involved.

[Neonivek]

Once again we are going back to "90% means 100%!"