What's new
By:

Obama Might Lose This

Stoked said:
I like it because I feel they do a fair job of avoiding bias. The provide articles, videos, polls, charts and what not from the far left, the far right and everything in between. That way I can look at all of it and decide for myself. Always good to hear others thoughts and arguments when a good faith effort is made to logically and rationally make them.

I also feel, pure uninformed opinion here, that an average of all the polls would land somewhere closer to the truth of the publics opinion.
Click to expand...

If the regression analysis is well done, it acomplished two things:
1) Takes different sample sizes into account. A poll of 3000 people gets more wieght than a poll of 1000 people.
2) It reduces the size of the error margin.

Taking an average of averages accomplishes neither of those.
 
One Brow said:
If the regression analysis is well done, it acomplished two things:
1) Takes different sample sizes into account. A poll of 3000 people gets more wieght than a poll of 1000 people.
2) It reduces the size of the error margin.

Taking an average of averages accomplishes neither of those.
Click to expand...

I never said it did. I just prefer to look at what all the polls are saying. Just a personal preference.

You are also just focusing on the polls side of it. RCP has some very interesting articles plus links to RCWorld, RCScience, RCReligion...
 
Stoked said:
I never said it did. I just prefer to look at what all the polls are saying. Just a personal preference.

You are also just focusing on the polls side of it. RCP has some very interesting articles plus links to RCWorld, RCScience, RCReligion...
Click to expand...

I have no complaints about RCP as a site. I'm just pointing out why sirkickyass feels a regression analysis gives a better indication of what the polls are saying than an average of averages gives. I certainly don't want to disturb your personal preference.
 
PKM said:
I voted for Bush in '00 and had no candidate in '04 and didn't vote. Also didn't vote in '08.
I've had people give me crap about not voting. I think that's dumb. If there is no one worth voting for, I won't.
Click to expand...

Agreed. I can understand why some see it as apathy, but if there's no real choice you accomplish nothing by pretending to make a choice.

There are no viable options in this election.

I think Trout talked me into showing up and writing in a candidate for pres. pretty much to be counted as someone who didn't support either of our major parties.

I heard a funny Bob Hope joke recently. It's gonna be paraphrased.

You've heard they saying, "One can't fool all of the people all of the time?" Well, that's why we have two political parties.
Click to expand...
 
One Brow said:
I have no complaints about RCP as a site. I'm just pointing out why sirkickyass feels a regression analysis gives a better indication of what the polls are saying than an average of averages gives. I certainly don't want to disturb your personal preference.
Click to expand...

You might very well be right. To be honest if RCP used that method instead of averaging the polls then that is what I would be comfortable with. I really like that site and they average polls so that is how I get my info.
 
Gameface said:
Agreed. I can understand why some see it as apathy, but if there's no real choice you accomplish nothing by pretending to make a choice.

There are no viable options in this election.

I think Trout talked me into showing up and writing in a candidate for pres. pretty much to be counted as someone who didn't support either of our major parties.

I heard a funny Bob Hope joke recently. It's gonna be paraphrased.
Click to expand...

My take on voting is that there is more on the ballot than president. Usually there are local initiatives, city council, mayor, senate, house reps at the state and federal level...

Write in Big Bird for president. The real change will start when we place people with morals and the ability to listen and compromise in as mayors, school boards council members, state level reps, Senators...

We have to start somewhere.
 
Stoked said:
My take on voting is that there is more on the ballot than president. Usually there are local initiatives, city council, mayor, senate, house reps at the state and federal level...

Write in Big Bird for president. The real change will start when we place people with morals and the ability to listen and compromise in as mayors, school boards council members, state level reps, Senators...

We have to start somewhere.
Click to expand...

I agree. I would also like to see a bit more local power (but that's just a whole 'nother huge discussion).

As for me not voting this way, I was living in a new state, or multiple states, each year so I never got a good grasp on the local needs/politicians. Otherwise, I would have been voting as you outlined above.
 
I need to educate myself on what issues are on the local ballot. Right now I can't think of a single initiative, judge, city council member, or anything else local that I even have a clue about.
 
Gameface said:
I need to educate myself on what issues are on the local ballot. Right now I can't think of a single initiative, judge, city council member, or anything else local that I even have a clue about.
Click to expand...

In Utah here are some more local votes:

Love v. Matheson
Hatch v. Howell
Utah Governor
 
Sirkickyass said:
This is the thrust of my question. Why do you prefer a simple average over regression analysis?

I've worked enough with stata to seriously respect the relational insights it gives.
Click to expand...

I'm interested in why you think it is more relevant.

Doesn't strata require qualifying subgroups before the survey? If so, how does Nate Silver's adjustements correct that after the fact? I looked into his methodology & it seems highly subjective. I don't doubt he's trying to minimize bias, but that he may not be able to when using already faulty polls. He has a ton of adjustments that seem similar to index adjusting but very subjective.

Not trying to be confrontational or anything. I don't know anything about polling methodology but relating it to quantifying chemicals has left more questions than answers.
 
franklin said:
I'm interested in why you think it is more relevant.

Doesn't strata require qualifying subgroups before the survey?

Not trying to be confrontational or anything. I don't know anything about polling methodology but relating it to quantifying chemicals has left more questions than answers.
Click to expand...

I'm not certain how to address the other parts of this beause this part has left me somewhat baffled.

It's possible we have a misread situation. I'm referencing stata, which is a regression analysis computer program: https://www.stata.com/

As for the methodology he's using I find most of it to either be obviously in best practices or above my personal ability to evaluate. Some of the index adjusting that you're referring to is empirical. House effects adjustments are based upon actual results vs. projected results of the poll across multiple election cycles. Other adjustments generally have a similarly empirical basis. Some of these are very sophisticaed. Silver had an article recently about measuring polls accuracy when they're going against the grain of conventional wisdom. As a general rule, I find it difficult to believe his model is significantly weighting non-statistically significant factors. Also, anytime you make your name in two different fields using statistical modeling at such a young age and with a high degree of accuracy in both fields my general assumption is that he's frankly better at it than I am.
 
Sirkickyass said:
As for the methodology he's using I find most of it to either be obviously in best practices or above my personal ability to evaluate. Some of the index adjusting that you're referring to is empirical. House effects adjustments are based upon actual results vs. projected results of the poll across multiple election cycles. Other adjustments generally have a similarly empirical basis. Some of these are very sophisticaed. Silver had an article recently about measuring polls accuracy when they're going against the grain of conventional wisdom. As a general rule, I find it difficult to believe his model is significantly weighting non-statistically significant factors. Also, anytime you make your name in two different fields using statistical modeling at such a young age and with a high degree of accuracy in both fields my general assumption is that he's frankly better at it than I am.
Click to expand...

Everybody keep an eye out for the asteroid that's headed our way.
 
Sirkickyass said:
I'm not certain how to address the other parts of this beause this part has left me somewhat baffled.

It's possible we have a misread situation. I'm referencing stata, which is a regression analysis computer program: https://www.stata.com/

As for the methodology he's using I find most of it to either be obviously in best practices or above my personal ability to evaluate. Some of the index adjusting that you're referring to is empirical. House effects adjustments are based upon actual results vs. projected results of the poll across multiple election cycles. Other adjustments generally have a similarly empirical basis. Some of these are very sophisticaed. Silver had an article recently about measuring polls accuracy when they're going against the grain of conventional wisdom. As a general rule, I find it difficult to believe his model is significantly weighting non-statistically significant factors. Also, anytime you make your name in two different fields using statistical modeling at such a young age and with a high degree of accuracy in both fields my general assumption is that he's frankly better at it than I am.
Click to expand...

The difference an R makes.
 
Of course the electoral map is what matters - and any way you slice it, it's an uphill climb for Romney.

I doubt PA, WI, MI and Ohio are truly toss up states. Things could change; but right now Obama wins all 4 of those states. In almost any other scenario, one swing state gives him at least 270 (with the exception of NH)

My prediction last week looks more and more like it's about to come true. Romney wins the popular vote and Obama wins the Electoral vote - buckle up for that one!
 
Scat said:
Looking at Stoked's favorite poll site is interesting. It has gone from predominantly blue just 3 weeks ago to predominantly red.

https://www.realclearpolitics.com/e...us/general_election_romney_vs_obama-1171.html
Click to expand...

Obama has a lead in electoral votes, 201 to 191.

https://www.realclearpolitics.com/epolls/2012/president/2012_elections_electoral_college_map.html

On their toss-ups, I hear Obama has basically conceded North Carolina, while Romney has basically conceded Wisconsin, Michigan, and Pennsylvania. That would make it 247-206.
 
Obama is winning or tied in Ohio in every major poll - some outside the margin of error. He's also trending slightly upwards while Romney is flat.

Anything can happen in the next week and a half; but if the election were today, I can't see Obama losing Ohio.
 
They both better get themselves on Leno, Letterman, Fallon, and Conan to finish out strong.
The one with the better late night appearances will win.
 
You must log in or register to reply here.

Share this page
Back
Top