• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Georgia special election heads to runoff as Ossoff earns 48% of vote

Status
Not open for further replies.
S

Sephzilla

Member
Georgia OssoffGAF, I'm lending you my energy

giphy.gif
 
RPGCrazied

RPGCrazied

Member
TheCochese said:
The results are going to come, and be what they are going to be no matter if you worry about it or not.

We're the ones who have to be repped by this person, so relax.
Click to expand...

Yeah, I know. But if he wins it would change everything. That means even places like Georgia could flip, why not other places? Texas? Alabama?
 
T

The Wart

Member
Earthbound64 said:
If it's within polling error of 50-50, then yes, it's a dead heat.

If H0 is "GuyA has 50 or greater" and Ha is "GuyB has 50 or greater," a 52-48 poll with a polling error >2 wouldn't give you evidence to reject the alternative.

Do people still not understand how polls work?




EDT.
Click to expand...

Er, no? Your logic only works if all polling errors are at least 2 pts. If there is a chance that the polls are closer than two points then it is better to be up 2 than down 2.

You seem to be applying statical hypothesis-testing logic but that's not really appropriate here.
 
E

Earthbound64

Banned
The Wart said:
Er, no? Your logic only works if all polling errors are at least 2 pts. If there is a chance that the polls are closer than two points then it is better to be up 2 than down 2.
Click to expand...

Yes, I'm treating it as if the polling errors are >2%
Are they more accurate than that? Since, if they're not, what I said applies.


The Wart said:
You seem to be applying statical hypothesis-testing logic but that's not really appropriate here.
Click to expand...

Why is it "not really appropriate here"?
 
B

borghe

Loves the Greater Toronto Area
br3wnor said:
This is true and it bodes well for 2018 either way, BUT an actual win would help put some pressure on Senator's voting on the AHCA bill next week. If Ossoff can outperform expectations and win by more than a sliver, it would really help in defeating the AHCA. As it is, Trump now has a 72% approval rating with Republicans. Have Ossoff get a nice win here and the moderate Republicans might have to really think hard about if they want to go down w/ the Trump Train or start trying to save their own asses heading into 2018.

On that level, I think this election could have a real effect w/ an Ossoff win.
Click to expand...

this is what I'm hoping for. If Ossoff wins, the Senate (well, more correctly the individual GOP senators) can't pull their typical 50 vote bullshit having their two most embattled members voting against. If he wins, more than 2 senators know they'll have to vote against AHCA, and they'll fight to either be the two that vote against, or just outright kill the bill.
 
Z

Zyae

Member
Earthbound64 said:
If it's within polling error of 50-50, then yes, it's a dead heat.

If H0 is "GuyA has 50 or greater" and Ha is "GuyB has 50 or greater," a 52-48 poll with a polling error >2 wouldn't give you evidence to reject the alternative.

Do people still not understand how polls work?




EDT.
Click to expand...


If it's 48-52 and the polling error is 2% it can either be

47-53
49-51

If they are 50-50 and the polling error is 2%

49-51
51-59

Clearly you'd rather be up 2% with a 2% polling error than even or down.


Obviously a lot more scenarios than this.
 
S

Steve Youngblood

Member
RPGCrazied said:
Yeah, I know. But if he wins it would change everything. That means even places like Georgia could flip, why not other places? Texas? Alabama?
Click to expand...

Don't get me wrong. I totally get that there are definitely implications that the result -- whatever it may be -- will send. I don't to sound like I'm saying the outcome isn't important. But having said that, it would probably serve you well just to calm down a little. Ossoff could lose. If he does, while it might mean that Trump isn't as unpopular as we'd like him to be, it could also just mean that a traditionally red district is staying red at least a little while longer. If Ossoff wins, it's definitely a blow for Trump's agenda, but it isn't necessarily a definitive turning point from which the GOP can not recover. At the end of the day, this is ultimately just one Congressional district.

I get that this stuff is exciting for political junkies. But at the same time, I really wish that we could stop completely buying into news cycles where every shiny new thing is The New Most Important Thing Event Ever in the History of Politics!!!!!
 
L

Lord Frieza

Member
Zyae said:
If it's 48-52 and the polling error is 2% it can either be

47-53
49-51

If they are 50-50 and the polling error is 2%

49-51
51-59

Clearly you'd rather be up 2% with a 2% polling error than even or down.


Obviously a lot more scenarios than this.
Click to expand...

This is a weird hill to die on bruh
 
E

Earthstrike

Member
Earthbound64 said:
If it's within polling error of 50-50, then yes, it's a dead heat.

If H0 is "GuyA has 50 or greater" and Ha is "GuyB has 50 or greater," a 52-48 poll with a polling error >2 wouldn't give you evidence to reject the alternative.

Do people still not understand how polls work?




EDT.
Click to expand...

"dead heat" has no well defined mathematical meaning. On a wholly technical level, if you have two populations each with some proportion of x and y, and the first population you measure 52%x-48%y, and the second you meansure 50%x and 50%y, and if both sampling methods were measured at the 95% CI, then the 52%x population will have more x a larger proportion of the time than the other population.

confidence intervals do not imply uniform probability of all results.within it. To top it off, even if they did, that would actually imply a measurement above 50 is more likely to be above 50 for any give CI.
 
E

Earthbound64

Banned
Zyae said:
If it's 48-52 and the polling error is 2% it can either be

47-53
49-51

If they are 50-50 and the polling error is 2%

49-51
51-59

Clearly you'd rather be up 2% with a 2% polling error than even or down.


Obviously a lot more scenarios than this.
Click to expand...

It's not a matter of "scenario selection" - because that's not what poll results tell you.
The poll results tell you that it's not certain that it isn't 50-50, and that's all they tell you.


Earthstrike said:
"dead heat" has no well defined mathematical meaning. On a wholly technical level, if you have two populations each with some proportion of x and y, and the first population you measure 52%x-48%y, and the second you meansure 50%x and 50%y, and if both sampling methods were measured at the 95% CI, then the 52%x population will have more x a larger proportion of the time than the other population.
Click to expand...

The results aren't "populations," though - they're polling results.
 
RPGCrazied

RPGCrazied

Member
Steve Youngblood said:
Don't get me wrong. I totally get that there are definitely implications that the result -- whatever it may be -- will send. I don't to sound like I'm saying the outcome isn't important. But having said that, it would probably serve you well just to calm down a little. Ossoff could lose. If he does, while it might mean that Trump isn't as unpopular as we'd like him to be, it could also just mean that a traditionally red district is staying red at least a little while longer. If Ossoff wins, it's definitely a blow for Trump's agenda, but it isn't necessarily a definitive turning point from which the GOP can not recover. At the end of the day, this is ultimately just one Congressional district.

I get that this stuff is exciting for political junkies. But at the same time, I really wish that we could stop completely buying into news cycles where every shiny new thing is The New Most Important Thing Event Ever in the History of Politics!!!!!
Click to expand...

I can't help it. I know it does nothing but to continue to dwell on it. I just want to send a message to Trump that a blue wave is coming and don't expect much support coming soon. I'm just tired of all the corruption and lying in this White House. We need more Democrats in power to stop this. At least cripple this Presidency even more.
 
F

Fancyarcher

Member
Has anyone heard about turnout being lower then expected? Somebody on another forum claimed that, just asking. I'm assuming all of the early bigger votes came from the past few days, correct?
 
E

Earthstrike

Member
Earthbound64 said:
It's not a matter of "scenario selection" - because that's not what poll results tell you.
The poll results tell you that it's not certain that it isn't 50-50, and that's all they tell you.




The results aren't "populations," though - they're polling results.
Click to expand...

Huh? The polls are attempts to measure a population. This sounds like a way to sidestep my main point with nonsense. It should be more than clear how the post I made maps onto the current discussion.

I'll give you the benefit of the doubt.an try to explain a bit about the basis of statistics as it is applicable in the case of polling. The population in this case is people who will vote for either a or b. Since, at a given point in time, everyone is either in a state of a, b, or neither. We can use a poll to attempt to get an idea of what that population looks like. This is what polls are for. in my example I was talking about measurements made against two populations whose actual distributions were unknown.
 
D

Deleted member 231381

Unconfirmed Member
Earthbound64 said:
If it's within polling error of 50-50, then yes, it's a dead heat.

If H0 is "GuyA has 50 or greater" and Ha is "GuyB has 50 or greater," a 52-48 poll with a polling error >2 wouldn't give you evidence to reject the alternative.

Do people still not understand how polls work?
Click to expand...

somewhere in the Welsh mountains, the Pedant Symbol flashes

a cry is heard

"To the Pedantmobile!"

nananananannanana

Equally, though, while a 52%-48% poll with a MoE of 3 percentage points within the 95% confidence level doesn't give you enough evidence to disprove that the actual result is 50%-50%, it also doesn't give you enough evidence to disprove that the actual result is 54%-46%. In the absence of any other information, we can conclude that the most probable actual result is... 52%-48% (since the MoE is derived from the CLT and the LLN, and the MoE is therefore not evenly distributed but instead has a normal distribution about the sample mean). So the most accurate statement is not 'it is a dead heat', it is 'we cannot be certain it is not a dead heat'.
 
E

Earthbound64

Banned
Earthstrike said:
Huh? The polls are attempts to measure a population. This sounds like a way to sidestep my main point with nonsense.
Click to expand...

No, it was a way to clarify what was being done.


Earthstrike said:
The population in this case is people who will vote for either a or b. Since, at a given point in time, everyone is either in a state of a, b, or neither. We can use a poll to attempt to get an idea of what that population looks like. This is what polls are for. in my example I was talking about measurements made against two populations whose actual distributions were unknown.
Click to expand...

Right, and the polls tell us that the H0 is within the range of error for the poll, and so we can't reject Ha.
It's possible Ha can be rejected, sure. But the poll can't tell us that.



Crab said:
Equally, though, while a 52%-48% poll with a MoE of 3 percentage points within the 95% confidence level doesn't give you enough evidence to disprove that the actual result is 50%-50%, it also doesn't give you enough evidence to disprove that the actual result is 54%-46%. In the absence of any other information, we can conclude that the most probable actual result is... 52%-48% (since the MoE is derived from the CLT and the LLN, and the MoE is therefore not evenly distributed but instead has a normal distribution about the sample mean). So the most accurate statement is not 'it is a dead heat', it is 'we cannot be certain it is not a dead heat'.
Click to expand...

This is pretty much what I was getting at, yes.
Apologies if I stated it poorly.
 
T

The Wart

Member
Earthbound64 said:
It's not a matter of "scenario selection" - because that's not what poll results tell you.
The poll results tell you that it's not certain that it isn't 50-50, and that's all they tell you.
Click to expand...

Ah now we're getting to the issue of epistemic vs ontological uncertainty!

There may be some sense in which the situation is *in reality* truly, or at least effectively, 50-50. This is ontological.

But in terms of what our best guess is based on the information we have available, we should guess that Ossoff has an at least slightly >50% chance. That's epistemic.

I'm not convinced the ontological view is that useful. After all, if we live in a deterministic universe, then the outcome is ontologically either 100% one way or the other way. This is assuming that quantum indeterminacy doesn't affect votes.
 
D

Deleted member 231381

Unconfirmed Member
The Wart said:
Ah now we're getting to the issue of epistemic vs ontological uncertainty!

There may be some sense in which the situation is *in reality* truly, or at least effectively, 50-50. This is ontological.

But in terms of what our best guess is based on the information we have available, we should guess that Ossoff has an at least slightly >50% chance. That's epistemic.

I'm not convinced the ontological view is that useful. After all, if we live in a deterministic universe, then the outcome is ontologically either 100% one way or the other way. This is assuming that quantum indeterminacy doesn't affect votes.
Click to expand...

A man after my own heart.
 
E

Earthbound64

Banned
The Wart said:
After all, if we live in a deterministic universe, then the outcome is ontologically either 100% one way or the other way. This is assuming that quantum indeterminacy doesn't affect votes.
Click to expand...

Of course. As you said, the result is either going to be 100% one way or the other (assuming 3rd-parties don't exist, of course).
Just, until all the votes are in and counted, we don't know which.
 
D

Davidion

Member
The Wart said:
Ah now we're getting to the issue of epistemic vs ontological uncertainty!

There may be some sense in which the situation is *in reality* truly, or at least effectively, 50-50. This is ontological.

But in terms of what our best guess is based on the information we have available, we should guess that Ossoff has an at least slightly >50% chance. That's epistemic.

I'm not convinced the ontological view is that useful. After all, if we live in a deterministic universe, then the outcome is ontologically either 100% one way or the other way. This is assuming that quantum indeterminacy doesn't affect votes.
Click to expand...

giphy.gif
 
T

The Wart

Member
Crab said:
Equally, though, while a 52%-48% poll with a MoE of 3 percentage points within the 95% confidence level doesn't give you enough evidence to disprove that the actual result is 50%-50%, it also doesn't give you enough evidence to disprove that the actual result is 54%-46%. In the absence of any other information, we can conclude that the most probable actual result is... 52%-48% (since the MoE is derived from the CLT and the LLN, and the MoE is therefore not evenly distributed but instead has a normal distribution about the sample mean). So the most accurate statement is not 'it is a dead heat', it is 'we cannot be certain it is not a dead heat'.
Click to expand...

You have met your match pedantry man! I am... also pedantry man.

"We cannot be certain it is not a dead heat" is accurate, but based on the polling data it is more informative to say "it is probably not a dead heat". We don't need to accept or reject the hypothesis that the race is effectively a dead heat -- that is not a statement about probabilities or about the world, it is a decision one makes based on probabilities, in cases where you need some sort of firm judgement to guide your future behavior. But there's no need to make that sort of decision here; as the outcome proven definitively very soon. We should stick to the realm of estimated probabilities in this situation, as that contains the most information and not worry about "disproving" anything at this stage.

Source: I work in the biological and social sciences where overzealous application of hypothesis testing logic to noisy data is causing a massive crisis of confidence.
 
E

Earthbound64

Banned
T

TheCochese

Banned
RPGCrazied said:
Well, uh, did you think a Democrat could possibly win the district you live in?
Click to expand...

You're making the mistake of thinking that one Republican district is like another. The issues that have been raised here, the demographics, etc aren't likely to be the same in Alabama or Texas.
 
J

Jeronimo Jones

Member
I don't know if there's genuine confusion over the implications of an Ossoff win or if some people are just extremely high on optimism in opposition to Trump.

For the poster from the UK asking, this is still just one district (in suburban Atlanta) Georgia, not the whole state. There are currently 14 districts in Georgia, each with their own representatives.

I'm hopeful for an Ossoff win and I supported him in the primary and in the runoff, but the potential impacts of an Democratic win in previously Republican districts across the US and on the national stage may be being overstated.
 
S

Steve Youngblood

Member
I just want to note that my interest was never actually in trying to explicitly define what a "dead heat" is in as much as I was responding to what I thought was a post that was confusingly written.

Zyae said:
Its not a "dead heat", hes up by about 2 which is very close. Hes just as probable to win as she is basically.
Click to expand...

Let's break this apart.

Zyae said:
Its not a "dead heat", hes up by about 2
Click to expand...

OK. Point taken.

Zyae said:
which is very close. Hes just as probable to win as she is basically.
Click to expand...

Now I'm lost.
 
T

The Wart

Member
Earthbound64 said:
That's what I was arguing, yes.
If one doesn't feel it's a worthwhile point, then fair enough.
Click to expand...

There are definitely contexts where it makes sense, I think it's just inappropriate as a "default" mode of thinking about probability. As you may be able to tell this is a topic on which I have developed strong feelings.

BronsonLee said:
what on earth did I just walk into
Click to expand...

DON'T TAKE THIS AWAY FROM ME BRONSON
 
R

rjinaz

Member
Man this is going to be one hell of a long day. This early and the thread trying to pick a part the relevance of a 2% polling lead.

It's close enough it could go either way is all that matters. Hope slightly for the best but assume the worst.
 
A

AnotherDayAnotherDollar

Banned
I don't like either one of the candidates. I have also never seen a district election gather this much attention nationwide lol (then again I'm relatively new to politics). I do think living in the district should be a requirement to run in that district though.
 
Status
Not open for further replies.

Similar threads

Georgia's 6th Congressional District Special Election |OT| Round 2: Fight!
26 27 28 29 30
1K
198K
zethren replied
  • Locked
Reuters: In Georgia, a Democrat's 'Make Trump Furious' campaign rattles Republicans
0
2K
Mortemis replied
  • Locked
New Yorker: Can This Dem Win GA6? (Come for the article, stay for a mayor's racism)
27
5K
Viewt replied
  • Locked
GOP are becoming concerned about upcoming elections
2 3
110
15K
mcfrank replied
  • Locked
United States Congressional Special Elections
2
59
10K
Bonen no Max'd replied
Top Bottom