Federal Review Composite Poll: Bush 328, Kerry 210

[phillybri]

http://www.federalreview.com/compositepoll.htm



A friend of mine runs this composite poll through his blog. He's wondering if there's any holes in his methodology...

[ewagner]

1. Wisconsin "leaning" to the Repubs????

Over my cold, lifeless body!!!!! :grr:

[ComerPerro]

36. No shit

This representation looks pretty ill-informed to me.

[theboss]

2. Bush's advantage/Kerry's advantage

Bush's advantage is that there is a shitload of states he simply does not have to campaign in.

Kerry's advantage is that the states he is strong in have a huge number of electoral votes.

There was an interactive map on Edwards site that was fun to play with.

[neverborn]

3. What???!?

Is this the Diebolded poll? :x

[NewJeffCT]

4. New Hampshire?

I can't see Kerry & Dean's neighbor going Red. aren't AZ and NM toss ups at this time?

[MAlibdem]

40. NH def could go red

Did last time and NH and MA are not friendly. We piss on them, they see us as snobs, it's fun!

[eumesmo]

46. NM is, AR isn't

[MissB]

5. The idea that Oregon will lean red is funny (nt)

[ibegurpard]

31. Oregon is dangerous territory for us right now.

The economy there sucks and the legislature and the Governor have been pretty useless. It was WAY too close in 2000. However, the fact that it hasn't gone R presidentially since the 80's makes me wonder why it would be put in the lean R category. If anything I would classify it as a lean D, which is by no means a safe bet but a more accurate assessment, IMHO.

[phillybri]

32. Kick

:kick:

[Josh]

55. Nader took 5.1% in Oregon -

the margin would've been a little different with him not there.

[TheFarseer]

6. This is just depressing

Florida and Missouri aren't even close. Ohio and Pennsylvania going to Bush. I think something extraordinary has to happen for Kerry to win. It seems no matter what happens, people are still going to vote Bush. I'm hoping we can win Texas if I'm the only one in the state that remembers to vote. Keep your fingers crossed.

[Richardo]

7. I have a hard time believing Ohio, PA and WV are leaning Bush...

[BigBigBear]

14. I agree

wtf ?

[eumesmo]

47. Look at the popular vote.

Bush is shown as up about between 2 and 3 points. If Kerry gains 4 points or so, that would shift things around. Their sight has a list of the battleground states, and many of them are close.

[Hunter_1253]

8. PA a Bush state...never

Pennsylvania, I doubt, will be a Bush state. Gore carried it in the last election, and there’s a Democrat as Governor to help Kerry campaign in the state.

[Freddie Stubbs]

20. The last poll conducted here had Bush with a comfortable margin

The state went for Gore in 2000 and Clinton in 1996 despite having a Republican governor. And it went for Bush in 1988 when we had a Democratic governor.

PA is a swing state. It has a Democratic Governor and a Republican controlled legislature.

Who will carry the Keystone State? It's too early to tell, but both candidates certainly have a shot at it.

[Zynx]

37. Bush did not have a "comfortable" margin.

[Freddie Stubbs]

45. Poll: Bush edges ahead in Pennsylvania

(04-21) 09:27 PDT HARRISBURG, Pa. (AP) --

President Bush has edged ahead of Democrat John Kerry in Pennsylvania, according to a poll of voters in the critical swing state won by Al Gore in 2000.

In a three-way race, Republican Bush is supported by 45 percent of the state's voters, compared with 39 percent for Democrat Kerry and 8 percent for independent candidate Ralph Nader, according to the poll released Wednesday by the Quinnipiac University Polling Institute. The rest were undecided.

Without Nader in the race, Bush still holds a slight 46-42 lead over Kerry, according to the Connecticut-based institute.

In March, Bush had 44 percent, Kerry 40 percent and Nader 7 percent. Without Nader in the question, Kerry and Bush ran even -- with Kerry at 45 percent and Bush at 44 percent.

more: http://www.sfgate.com/cgi-bin/article.cgi?f=/news/archive/2004/04/21/politics1227EDT0619.DTL

These results echo another poll that was released last week.

[cmayer]

54. Reject the premise.

Look at the polls from Rassmusen and others. They don't match up with this projection.

Kerry leads in Iowa and Michigan. It's a toss-up, advantage Kerry, in Florida, Ohio, Oregon, Wisconsin and Pennsylvania (some show even in PA.)

It's dead even in West Virginia in one poll, down in a couple of others. Battlegrounds where Kerry is running a little behind are Arizona, New Mexico and Missouri.

Not looking too bad if you ask me. I don't think the public has absorbed all of the bad news yet. It takes a week or two to sink in.

[devinsgram]

26. Right on there,

Pa will never go to *.

[aquart]

9. Alaska? Nevada? I dunno. Is George really beloved in those places?

Is this based on current polling? Or tradition? New Mexico is red?

This poll shows a satisfied electorate. What the hell is this guy's methodology? Darts?

[Richardo]

16. There was a story on NPR yesterday about how betrayed Nevadans feel

...because of Smirky's flip flop and fast tracking of Yucca Mountain Nuclear Waste Dump.

[Leilani]

44. Alaska is safely Republican!

What are you smoking?

[Padraig18]

10. PA, WI, IA and OR leaning BUSH???

:wtf: His methodology is clearly fucked up, if he thinks THAT...

[Leilani]

11. That map is not looking too promising!

Just a few weeks ago, the numbers looked different.

[HooptieWagon]

12. Florida has been a 46-46 tie for the past two months

I don't see how it can be considered "advantage" Bush*. And California, called "advantage" Kerry, is more than likely pretty solidly Democratic. I'm not sure where your friend is getting his numbers from, RNC maybe?

[newyawker99]

21. Hi HooptieWagon!!

Welcome to DU!! :toast:

[Ready4Change]

23. Which is why it's red. A little vote fixing by Jeb...

Just like 2000, except with touch screen voting there'll be no troublesome recounts to fret over.

[qanda]

24. If it's within the margin of error

It should probably say too close to call. And how is a tie broken? This map is very flawed.

[moondust]

13. IA had Kerry up by 10 points in recent polls!

n/t

[DemNoir]

15. Big omission

What, no Weekly Reader poll? Now we are turning to amateur pollsters to get our daily dose of negativity. Will this ever end?

[phillybri]

17. He's not making the info up...

Check out his methodology...

http://www.federalreview.com/methods.htm

[DemNoir]

18. Limiting Factor

The problem with combining these numbers is that the result can only be as good as the weakest of the polls included.

[eumesmo]

48. This is true

You can't just take polls at face value. Different pollsters use very different methodologies and have different track records. People tend to lump them together as if they were indistinguishable. You see this on T.V. where the pundits will quote the latest poll without regard to who put it out.

[muriel_volestrangler]

19. Do you really trust a poll that invites you

to discuss it at Free Republic? Look at the bottom of the page linked in the original post.

[moondust]

22. HaHaHaHa! Good eye!

n/t

[phillybri]

25. It's also going to include THIS link...

...here at DU.

To his credit, the guy is trying to keep this thing fair in spite of being a conservative.

[tom_paine]

27. I'm guessing there are some holes

If he is relying on Federal and Corporate TV Pravda data, it's no wonder his results are what they are.

We now swim in a Soviet-style Sea of Bushveik Lies.

[DarthReagan]

28. Does This Idiot Make Any Sense at All?

I know the rules and that I am liable to be banned, but I promise not to disrupt, call names, or pimp conservative philosophy. Hence, my first post is an honest admission of who I am. I created the poll that phillybri posted and hope to come here to respond to questions about the methods. I'm trying to get a sense of where the EV stands and where the national polling stands. When I see different polls with different results, I wonder what it means, and I've attempted to construct a way to at least make a little sense of the differences. If my methods are biased toward Bush, then I have failed. That's why I'd like your comments.

Anyway, a couple of responses to what has been posted here. I am aware that an Iowa poll shows a Kerry lead of 10 points. The poll is from Rasmussen who uses automated polling (not real people), thus he can't tell if he is getting a voter of someone's 3 year old. His polls in 2000 (then, called Portrait of America) were the most incorrect. At the end, he had a lead for Gore of 9 points, but throughout the campaign, he constantly showed a big Bush lead. Don't know what moved it at the end, but I just don't trust him. However, that means I've probably got Iowa slotted wrong, so I may go back to using some of his numbers (to some extent).

My methodology is heavily based on numbers and not what I think will happen. Do I think Bush will win Pennsylvania? It's surely a tossup, but the numbers aren't saying that now. As we get closer to the election, this thing should give an idea of real trends. We'll see.

I'll try to address your comments about this poll. And if you see anything that suggests my presidential preference seems to be influencing the results, please point that out.

[muriel_volestrangler]

38. I appreciate your posting here

and withdraw my slightly snarky remark above, now there's a link to here too.

Is your definition of a 'recent' state poll the same as a recent nationwide poll? If so, then I would have thought the state poll ought to have more weight than the Base Margin/Composite result figure - being based on one poll, rather than one poll plus your best guess as to the changing situation particular to the state. If the state poll has a larger margin of error than the nationwide one, I can see that should be taken into account.

Have you posted anywhere the differences between the base margin/composite results and state polls? If we saw those (historical as well as current), we might be able to make suggestions about improvements (someone might see a pattern).

After you have found a difference in a state between the 2 results, would you take that difference into account in future weeks, even though the state poll becomes out of date? eg if in one week the composite-derived lead for a candidate is 4%, but a state poll shows 7% (so you'd show it as 5.5%), would you, in future weeks, just use the composite result, whatever it does, or would you use the composite result plus 1.5%?

[DarthReagan]

42. Thanks for the comments

The national polls that are used are the ones from the immediately preceeding week, as well as the composite result from a week ago. So, there are elements of 2 weeks of national polls in each composite result. This should help to even out some weekly swings in order to try to show a real trend.

State polls are used until the are stale, but I don't know what that means. Since state polls are not conducted as often, I can't update them regularly. Right now, I have no polls older than early March in the poll. But if a new one comes out, I'll replace the old one or average...really, it's just a guess. If a new NJ poll shows Kerry up 10, I won't average that with the one showing Bush leading. That's what I call "punditry". If several state polls come out and show NJ really is close, then I'll give recent state polls more weight. In fact, I'll increase the weight of state polls as we get closer to the election.

I havne't posted a lot of my data out of equal parts laziness and an html aversion. I'd need to code new tables and things. I have considered doing it. I hope that answers you questions.

Thanks for taking this seriously and helping me out.

[eumesmo]

49. You're right about Rasmussen

The non-poll stuff on his site is often interesting, but it's hard to take the actual polls seriously. Certainly they should be given less weight than other polls with a better track record. Not taking into account who is doing the polling is an error made by many prognosticators. Not all polls are equal.

Enough soap box. Time for a question. You wrote, "If a new NJ poll shows Kerry up 10, I won't average that with the one showing Bush leading. That's what I call 'punditry'."

I didn't what you were wanting to say here.

[DarthReagan]

51. I lean toward Kerry

I guess my point is that averaging the polls doesn't necessarily seem appropriate and I'll let my judgment trump the numbers. I don't believe for a minute that Bush will win NJ, so when the poll says that, I don't buy it and don't give it too much weight. So, when I see the opposite -- Rasmussen has Kerry up huge in NJ -- I don't just average them, I look at the results and decide what I think is right. In this case, I'd almost dismiss the poll for Bush and, even though I don't trust Rasmussen, I'm more inclined to believe that based on history.

No Bush bias here, I believe the poll showing Kerry leads is the most trustworthy. My God, it's New Freakin' Jersey. To average these numbers would be to way overstate Bush's chances here. It will take a lot more polls for me to believe otherwise.

[randr]

29. I would throw FL, NV, and AR into the pink zone

and I see that if this is anywhere near accurate the Reps are far more vulnerable than the Dems, assuming the states with a weak lead are up in the air.

[DarthReagan]

30. Republican Vulnerability

You are right that any state that is light pink or light blue should be considered a tossup. Although I don't like saying tossup -- it seems like a cop out, so I just give the prediction as of the date I have the numbers.

Of the tossups, Gore won WA, OR, NM, WI and IA last time. Bush won OH, WV, AR and NH. The numbers pulled MO and FL out of the tossup category, but we all know they still are tossups.

A slight movement in the national polls in Kerry's favor will change this map significantly, as could state polls in FL, WI, OR, NM, PA or any other battleground. This is just a snapshot of what's available today.

[boneygrey]

34. Methodology

How did you arrive at splitting Nader votes 65% to Kerry. Also you said you used punditry in some decisions. Can you elaborate.

[DarthReagan]

35. Nader vote to Kerry, etc

I looked at exit polls, where VNS didn't have the sense to ask Nader and Buchanan voters who they'd have voted for if Nader and Buchanan weren't on the ballot. Morons.

Anyway, by seeing the breakdown of voters by sex, party, income, etc., and how they voted among the parties, I came up with some not too solid numbers that made it appear that Nader and Buchanan voters most likely would have stayed home. So, if Nader wasn't on the ballot, those folks wouldn't have voted for Gore, they wouldn't voted at all.

Well, that certainly would have tilted things toward Bush, so I just allocated more the the Dems than the Reps. So, instead of using an even split baseline for Florida, I acutally use a number showing a Gore win.

Did the same with Buchanan, going 65/35 for Bush, though I bet Buchanan's voters would have stayed home too, shooting things up in the back yard with their militia buddies.

Punditry? Well, here's an example. A few weeks ago, there was a Michigan poll showing Bush leading by 2. There was a Rasmussen poll (whom I don't trust because of his recorded phone calls), which showed a Kerry lead by 4 in MI. While I trusted the old fashioned methodology of the Bush +2 poll, I just couldn't believe that Bush had a 2 point lead in Michigan, so I considered the Rasmussen poll and recorded the vote as +2 for Kerry in MI, then the composite polling numbers took over. I've ignored a Rasmussen Iowa poll showing Kerry by 10, but I'll probably use that in the next poll, unless another comes out.

[mondo joe]

41. How about some purple states?

I'd suggest that where states are a toss up - and you might select a percentage split to indicate a toss up - you just go purple.

[DarthReagan]

43. Purple would work, but

I considered just designating tossups, but decided that was a copout. I'm just going to call them each week and see how they change.

[LibertyorDeath]

33. Send it to TruthIsAll he will show it for the shit it is. IMO

[Deja Q]

39. When it comes to repukes, RED is an apt color...

For they are the new commies.

I hope it's false, that red is disturbing.

[agincourt]

50. Kerrys lost 58 electoral votes,

since last week. Keep posting the map. I don't know if we're in that bad of shape but it will show trends. Too many sheeple want to abolish the 20th century I guess.

[DarthReagan]

52. Trends, it is

That's what I really expect my analysis to show. I don't know that on November 1 I'll have the right states slotted in the right columns, but I bet that if a trend appears in my analysis, it will appear in the final vote.

[Terran]

53. Missouri leans, but I don't see it as a Bush shoe-in

Missouri was one of those close contests in 2000, and things are incredibly shitty, economy-wise, right now here--we hold the record for lost manufacturing jobs.

All but one of our state-wide officials are Dems, even though we have two repug Senators. It could go either way.