2004 User Predictions - Discussion

[cskendrick]

I just don't see it.

From where I'm sitting, the Zogby results line up with about a 55%+ approval rating for Bush, which is way over the 8-poll moving average (at 51.50, even filtering 1-stdev outliers) as of the latest Gallup poll.

Let's say Zogby has a superior handle on the situation at the moment: If so, and this holds true in November,  Bush sweeps Red States and pockets IA, NM, WI, OR, MN, MI, ME, WA and PA, as well.

Final score in ye olde EC:  370-168. Not a sweep, but good enough for more government work.

I'd be inclined to accept Zogby if the Gallup poll had shown a significant upward bump in Bush's approval ratings; as it stands, I cannot help but notice that the sample was taken in the midst of the short-lived scandal surrounding Kerry last week. There may be a distortion in the results; a follow-up sample seems prudent.

What am I saying? It's nine months to Election Day. Of course there will be follow-up!

[cskendrick]

Breakout of Southern States in the 2004 presidential election (as of the latest poll data)

Bush continues to be strong in the South, and support for him in the core of reliably Republican states isn't seriously threatened in my model until his nationwide approval ratings dip under 45%. (At the moment, the average of the latest eight major polls is 51.5%.)

Reliably Republican
1. Texas
2. Mississippi
3. Lousiana
4. Kentucky (border state)
5. Arkansas
6. South Carolina
7. Georgia
8. Alabama
9. North Carolina
10. West Virginia (border state)

Leans Republican
11. Virginia
12. Tennessee                  
13. Missouri (border state)

Leans Democrat
14. Florida                          

Reliably Democrat
15. Maryland (border state)

[cskendrick]

Saratoga - Thanks for the welcome aboard.

[cskendrick]

Florida leans Bush, Arkansas isn't solid Bush.

I'm working on a liberal/conservative model; I think Arkansas is as conservative a state as Georgia or South Carolina, and Florida is as liberal a state as Minnesota or Oregon...rather, about as closely divided as those states.

A weakness of my model is that conservatism and Republicanism (likewise, liberalism and Democratism) are imperfectly correlated.

And fixing the leaks is why I'm here.

Arkansas - Why are the Democrats relatively strong there? The Clintons? A strong party organization?

Florida - I don't really need to ask the reverse question, given whose brother is governor there.

[cskendrick]

Thank you, everybody.

Here's what I did on short notice: I swapped 5% of the electorate from one side to the other, depending on the straw poll of suggestions here.

Figuring it prudent, I boosted the conservative count in Texas, as well...not that I needed to.

I might contest the suggestion that Tennessee is reliably Republican, given its Democratic governor. But two years ago, South Carolina had a Dem governor, too, so I will concede the point.

Adjusted to More Republican: Texas, Tennesee, Florida

Adjusted to More Democrat: Louisiana, Arkansas, West Virginia.

Breakdown of Southern State, version 2.0

Reliably Republican
1. Texas
2. Mississippi
3. Kentucky (border state)
4. South Carolina
5. Georgia
6. Alabama

Leans Republican
7. North Carolina
8. Tennessee
9. Louisiana
10. Arkansas

Leans Democrat
11. Virginia
12. Florida
13. West Virginia (border state)
14. Missouri (border state)

Reliably Democrat
15. Maryland (border state)

You know, this makes a lot more sense.

Thanks, folks!

NOTE: The Pew report came out today; the moving average of nationwide Bush approval ratings (which I use to drive my predictions) now stands at 50.50. This information is incorporated here:

[cskendrick]

Regarding Missouri -

I guess I'd better show you, rather than tell you.

All of this is driven by an 8-poll moving average of national approval ratings for President Bush.

Each state has its own conservative/liberal predisposition. For Missouri, this is what you get

MISSOURI scenarios (as of 2/20/2004)

Bush ratings     Status
57.2%+            Strongly Republican
54.3-57.1%       Republican
51.5-54.2%       Leans Republican
48.5-51.4%       Leans Democrat
45.5-48.4%       Democrat
45.4%-            Strongly Democrat

As of the latest polls, the moving average is 50.5%.

Whatever it actually will be come Election Day is a matter for the voters to decide.

But this sure is a lot of fun.

[cskendrick]

The election goes to the House of Representatives

What a nightmare scenario.  Does anyone have any thoughts about the possibility of another result that gives Bush the electoral majority but has the Democrats winning the popular vote?

Well, that's just it; this time around, it looks like Bush is the one who is at risk of winding up with the popular vote but losing in the Electoral College.

From where I'm sitting, Bush has to have just under a 52% approval rating to win.

This is doesn't sound like too demanding a hurdle to clear, but his average poll ratings at the moment are under that.

Here are the last eight that I have a hold of:

Gallup   2/7/04   52
   ABC   2/10/04   50
Gallup   2/11/04   51
   CBS   2/14/04   50
   Pew   2/15/04   48
Harris   2/15/04   51
Gallup   2/17/04   51
   Fox   2/18/04   48

Note: I use the day before the end date of a survey as the timestamp.

Worst case scenario (for purposes of Constitutional law and general civil unrest): There is a (highly unlikely) change of control in the House of Representatives AND there is a tie in the Electoral College.

Yikes.

On the other hand, I'd be content with taking back Congress. Divided government is good.

[cskendrick]

The election goes to the House of Representatives

What a nightmare scenario.  Does anyone have any thoughts about the possibility of another result that gives Bush the electoral majority but has the Democrats winning the popular vote?

It's very early to consider that possibility. Furthermore, it would be historic. I believe the popular vote has never been lost by an incumbent who had no challenge for the nomination, and who had a net gain for his party in the House over the two preceding House elections.

Those are some significant qualifiers.

Is there anyone other than George W. Bush who is in that box?

[cskendrick]

My Nailbiters

Conditions - Bush's approval ratings going into the election are right on the cusp, votes are decided by less than 0.5% in each case.

In the case of Ohio (remember, we're going out of our way to cause a crisis here) it's a tie.

Not a statistical dead heat; a tie.

1. Ohio
2. Nevada
3. Iowa
4. New Hampshire
5. Missouri
6. New Mexico

The good news is that reality is NEVER so capricious.

I mean, when's the last time that the deciding state's votes were so close that there were...recounts...court appearances...hanging chads...

[cskendrick]

cskendrick

where do you get your data?

The job approval data is publicly available; I use www.pollingreport.com as my source of first resort. It's easy and the tables I am interested in are free.

The rating of each state by a "Conservative Percentage") is derived from the voting history of the last three presidential elections (1992 1996 2000), extrapolated forward to 2004.

The sensitivity of each state to a change in Bush's approval ratings is updated using the by-party breakout of Bush numbers. The greater the "conservative" factor, the more the state's predicted electoral behavior converges on Republicans' views of Mr. Bush, and vice-versa.

It works pretty well, I think. Right now, Bush's approval ratings are under what he needs to secure re-election, and my assessments of the states are based on that, not any predicted approval rating.

If I have to place a bet down, I'd say this one's going to be a close, dirty expensive contest...and I'd be saying nothing new or interesting.

Not sure if I answered the question, but it's late and I've got one more post to reply to before turning in.

Until next time...

[cskendrick]

The election goes to the House of Representatives

What a nightmare scenario.  Does anyone have any thoughts about the possibility of another result that gives Bush the electoral majority but has the Democrats winning the popular vote?

It's very early to consider that possibility. Furthermore, it would be historic. I believe the popular vote has never been lost by an incumbent who had no challenge for the nomination, and who had a net gain for his party in the House over the two preceding House elections.

Those are some significant qualifiers.

Is there anyone other than George W. Bush who is in that box?

GWB is in excellent company, since 1860 (GOP vs DEM) the situation has happened five other times:
Lincoln 1864
T Roosevelt 1904
F Roosevelt 1936
Eisenhower 1956
Reagan 1984

Every incumbent in that period with both those other factors against lost reelection (4 times).

Ah. It's not as rare a combo as I originally thought.

I'd be careful about historical determinism in politics, though.

The Democrats got exactly what they deserved in 2002 for assuming that historical trends would be friendly.

Human beings make or break their own destiny.

But I don't have to tell Republicans that.

[cskendrick]

Thank you, everybody.

Here's what I did on short notice: I swapped 5% of the electorate from one side to the other, depending on the straw poll of suggestions here.

Figuring it prudent, I boosted the conservative count in Texas, as well...not that I needed to.

I might contest the suggestion that Tennessee is reliably Republican, given its Democratic governor. But two years ago, South Carolina had a Dem governor, too, so I will concede the point.

Adjusted to More Republican: Texas, Tennesee, Florida

Adjusted to More Democrat: Louisiana, Arkansas, West Virginia.

Breakdown of Southern State, version 2.0

Reliably Republican
1. Texas
2. Mississippi
3. Kentucky (border state)
4. South Carolina
5. Georgia
6. Alabama

Leans Republican
7. North Carolina
8. Tennessee
9. Louisiana
10. Arkansas

Leans Democrat
11. Virginia
12. Florida
13. West Virginia (border state)
14. Missouri (border state)

Reliably Democrat
15. Maryland (border state)

You know, this makes a lot more sense.

Thanks, folks!

NOTE: The Pew report came out today; the moving average of nationwide Bush approval ratings (which I use to drive my predictions) now stands at 50.50. This information is incorporated here:

You can't look at Govs races as an indicator for the direction a state votes. You have to look at the past presidential elections, and factor those together. West Va would be considered more democratic, however I believe that they will go Bush this time.

Florida on the other hand has ALWAYS been solid GOP. This notion about how it leans democratic must be coming from the close election of 2000. FL went GOP in 1992, and would have went GOP in 1996 had Perot not been running. Clinton did not win but by a few points. I think with all that in mind you have to consider 2000 as more a fluke for being so close. Since 2000, more people in FL are now registered GOP than Dem, the first time in the state! FL doesn't lean GOP, it will be reliably GOP in 2004.

If you look at the LA Gov race, you had basically identical candidates for Gov. It was an off year so the GOP expected to do worse than normal. Had you had a John "F" Kerry run in LA, GOP would have won big time. The Dems win in the south with a "moderate" dem, or at least in Edwards case, a "perceived" moderate.

First off: I am not looking at gubernatorial races as a driver for my model, I am not sure what you are getting at, and I am incorporating the last few presidential elections.

WEST VIRGINIA - Darn it. I just knew I shouldn't have gotten in the business of manual adjustments yesterday.

Yesterday, I took a suggestion to tweak my predictions for West Virginia...now I hear the exact opposite from you.

Ruling: I'm moving WV back to the original settings and keeping it there!

FLORIDA: This adjustment I will keep; I boosted Florida's "Conservative Quotient" (that has a nice ring to it) already, per the consensus suggestion yesterday.

All Bush has to do is keep his approval ratings from falling any lower and he'll win the Sunshine State. That's not going to be difficult now, is it?

LOUISIANA: I stand by this one. Bush clears Lousiana even if his nationwide ratings fall below 49%.

What I don't get is that in your first paragraph, you tell me not to concern myself with gubernatorial races, and here you create a fantasy football league situation in which a candidate from Massachusetts runs for Lousiana governor. Of course he'd be killed. He's not from Louisiana!

And Trent Lott would be killed if he ran for governor in Michigan; things that Mississippians are willing to accept or overlook, the Michiganders won't.

That's why I don't get into these fantasy football discussions, save to make light of them.

[cskendrick]

The florida race was so close because of the bring 5 campaign, where african-americans were encouraged to bring five friends and relatives with them when they voted.

Whilst it was highly sucessful for those who were allowed to vote...

Five real friends or five imaginary friends?

There are allegations that Katherine Harris used her office as FL Secretary of State to give Republicans a boost in the 2000 election.

Specifically, that she engaged in selective purging of the voter rolls on the eve of the 2000 election.

It's a testable hypothesis.

I'd like to hear more about how we can test for real versus imaginary voters.

Sounds like a useful confidence test in instances where the mechanism of tallying votes is suspect, invisible...or proprietary.

[cskendrick]

The florida race was so close because of the bring 5 campaign, where african-americans were encouraged to bring five friends and relatives with them when they voted.

Whilst it was highly sucessful for those who were allowed to vote...

Five real friends or five imaginary friends?

There are allegations that Katherine Harris used her office as FL Secretary of State to give Republicans a boost in the 2000 election.

Specifically, that she engaged in selective purging of the voter rolls on the eve of the 2000 election.

It's a testable hypothesis.

I'd like to hear more about how we can test for real versus imaginary voters.

Sounds like a useful confidence test in instances where the mechanism of tallying votes is suspect, invisible...or proprietary.

There is proof of dead people voting in Missouri.

Some would say that there is proof of living, eligible voters being denied the vote in Florida.

If the acts are equally reprehensible, then we can resolve the moral weight by a simple comparison of frequencies: (a) how many dead people voted in MO, and for which party? versus (b) how many living people were WRONGLY denied the vote in FL, and for which party would they have voted for?

Is this a game you are comfortable playing?

[cskendrick]

The florida race was so close because of the bring 5 campaign, where african-americans were encouraged to bring five friends and relatives with them when they voted.

Whilst it was highly sucessful for those who were allowed to vote...

Five real friends or five imaginary friends?

There are allegations that Katherine Harris used her office as FL Secretary of State to give Republicans a boost in the 2000 election.

Specifically, that she engaged in selective purging of the voter rolls on the eve of the 2000 election.

It's a testable hypothesis.

I'd like to hear more about how we can test for real versus imaginary voters.

Sounds like a useful confidence test in instances where the mechanism of tallying votes is suspect, invisible...or proprietary.

Purging criminals is a good policy.

At no point do I criticize that position.

I think that purging eligible voters is bad.

I think reasonable, freedom-loving people the world over would agree with me.

And I am sure that you do, as well.

[cskendrick]

Cskendrick... opebo is the most consistantly ammoral person on this site.
He likes gerrymandering and low turnouts, and also thinks that Dubya should fake Bin Laden's capture to get re-elected...

What interesting views!

I'll have to ask about them later on.

[cskendrick]

The florida race was so close because of the bring 5 campaign, where african-americans were encouraged to bring five friends and relatives with them when they voted.

Whilst it was highly sucessful for those who were allowed to vote...

Five real friends or five imaginary friends?

There are allegations that Katherine Harris used her office as FL Secretary of State to give Republicans a boost in the 2000 election.

Specifically, that she engaged in selective purging of the voter rolls on the eve of the 2000 election.

It's a testable hypothesis.

I'd like to hear more about how we can test for real versus imaginary voters.

Sounds like a useful confidence test in instances where the mechanism of tallying votes is suspect, invisible...or proprietary.

There is proof of dead people voting in Missouri.

Some would say that there is proof of living, eligible voters being denied the vote in Florida.

If the acts are equally reprehensible, then we can resolve the moral weight by a simple comparison of frequencies: (a) how many dead people voted in MO, and for which party? versus (b) how many living people were WRONGLY denied the vote in FL, and for which party would they have voted for?

Is this a game you are comfortable playing?

The dead people voting in Missouri were of course voting Democrat, in inner city districts.  The Democrats are way ahead of the Republicans in the arena of voter fraud.  I was never prouder of Kit Bond than after the 2000 debacle when Ashcroft was 'defeated' - the worthy man nearly burst a blood vessel on the stump complaining of the Democrat shenanigans in the City of St. Louis.  

Maybe it's just me, but it seems to me that tampering with elections a terrible thing, even when Republicans do it.

[cskendrick]

well the fact is, Republicans won Florida by 537 votes, if even only 25,269 of those 50,000 voted Democrat then they would have won Florida and the election, considering the majority of the 50,000 were Democrats, that wouldn't have been too hard

Well worth it to make sure no felons voted.  I for one don't want criminals voting.

Katherine Harris' partisan zeal cost thousands of ELIGIBLE voters their franchise.

Tampering with elections is bad news, even when Republicans do it.

[cskendrick]

but we have already established that a large proportion of these people were NOT criminals, they just had similar names to criminals. How would you feel if the majority of these had been Republicans and Al Gore had won because of this, I bet you would change your mind then.

Besides they were not Felons, they were EX-felons, they had committed a crime and served their debt to society, by denying these people the right to vote you treat them like second class citizens.

There's no such thing as an 'ex-felon'.  You're a felon for the rest of  your life - lucky to be out of prison, but still marked.

And there's no such thing as a LEGITIMATE stripping of an eligible person's voting rights.

In fact, it's a crime to do so, several times over.

Even when Republicans do it.

[cskendrick]

not a bad analysis.  I still think Wisconsin will go with Kerry, and that Iowa is a tossup.  Ohio is also going to be close and I suspect WV will go to Kerry.  Nevada may also go to Kerry.  I think your percentages will be close +/- 2%, which pretty much tells you nothing.

My pick for the closest race is Ohio.

As for the popular vote/electoral vote situation, I think Bush is in much greater danger of being "Gored" (gets the PV, loses in the EC) than is being considered.

From where I'm sitting, he needs about 52% approval ratings (not exactly a high bar to clear) to win, though the inevitable arrival of Ralph Nader to the party may change that.

[cskendrick]

You must be logged in to read this quote.

True
After all, the states that Dubya's economic [mis]managment have hit hardest have a tendency to be small...

Which states are those?  The states in worst shape economically are mostly big, and mostly Dem.  Such as CA, WA, MI, IL.. also a few medium ones like OR and MA.  Its interesting how poorly the West Coast has done for a long time now - even before Bush was president.  
The only Republican states in anything close to bad shape are the Carolinas and they're not in any doubt going for Bush.  WV and PA are doing fine, better than OH, which is not that bad.
But the overarching truth is the economy isn't that bad anyway.

All politics is local, but all economics is personal.

And across time and the set of countries, the worst thing you can have for regime instability is high growth with most of the people getting very little of the proceeds.

A sustained disparity between market performance and net job creation (new jobs less net growth in the workforce) is not only bad news for the incumbent administration...it's bad news for the country at large.

[cskendrick]

Hah! A textile worker who loses his job and gets one at Walmart isn't going to be happy about it.

Textile workers didn't make much anyway - not a significantly better job than WalMart.  Besides, my point was that the Carolinas are so strongly Republican that even some unemployment-induced voting will not shift them.

Not that I know anything about the Carolinas (see logo to upper left) but my mom is a diehard Republican and hangs out only with diehard Republicans.

None of them are voting Bush this time around, and now that Dean's no longer the nightmare alternative, they're contemplating either (a) voting Kerry/Edwards (they'd DEFINITELY vote Edwards at top billing) or (b) sitting this one out.

My mom's reasons: Bush promised to restore honesty to the White House....then he lied.

But that's anecdotal, and impossible to verify independently.

What I can validate is that if Bush's numbers drop to about 49%, he'll lose North Carolina no matter who the Democratic candidate is.

He'll lose South Carolina at about 47%.

Figure out the odds around those.

[cskendrick]

not a bad analysis.  I still think Wisconsin will go with Kerry, and that Iowa is a tossup.  Ohio is also going to be close and I suspect WV will go to Kerry.  Nevada may also go to Kerry.  I think your percentages will be close +/- 2%, which pretty much tells you nothing.

My pick for the closest race is Ohio.

As for the popular vote/electoral vote situation, I think Bush is in much greater danger of being "Gored" (gets the PV, loses in the EC) than is being considered.

From where I'm sitting, he needs about 52% approval ratings (not exactly a high bar to clear) to win, though the inevitable arrival of Ralph Nader to the party may change that.

I don't understand your popular vote analysis. It seems to me that Bush can win handily without even getting 50% of the PV.

....
             
SCENARIO 1 Bush loses 0.9% of total popular vote, loses FL, NH, OH, MO, and NV
PV: Dems 52.0% Reps 47.0%
EV: Dems 327    Reps 211

SCENARIO 2 Bush gains 0.68% of total popular vote, loses FL and NH
PV: Dems 50.4% Reps 48.6%
EV: Dems 291    Reps 247

SCENARIO 3 Bush gains 1.93% of total popular vote, holds FL and NH, gains IA, NM, and WS
PV: Reps 49.8%  Dems 49.2%
EV: Reps 300      Dems 238

So according to this method, Bush gets a 300 EV win without even getting a majority of the PV (though beating the Dem's PV).

I've been treating Bush's job approval ratings as a proxy for the nationwide popular vote.

Some states have much higher concentrations of Bush supporters than others; which I took to mean that Bush needs greater than 50% approval ratings to win this time around.

I need to modify predictions based on Nader bleed-off, and I am waiting on some poll figures to come out before doing that.

At CURRENT poll ratings, these are my predictions, by state (moving average 50.8% Bush job approval)

State   % Bush
WY   61.5%
ID   60.2%
UT   59.6%
TX   57.7%
SD   57.4%
ND   57.2%
OK   55.4%
MT   55.2%
AK   54.7%
NE   54.6%
MS   54.4%
KY   53.9%
SC   53.4%
GA   53.1%
IN   52.9%
AL   52.8%
NC   52.7%
WV   52.3%
TN   52.3%
LA   51.6%
KS   51.6%
AR   51.4%
VA   50.2%
FL   50.1%

CO   49.9%
AZ   49.9%
MO   49.5%
NV   49.2%
OH   49.1%
IA   48.8%
NH   48.8%
NM   48.6%
WI   48.1%
OR   47.9%
MN   47.5%
MI   47.1%
ME   46.8%
WA   46.6%
PA   46.5%
VT   45.0%
IL   44.9%
CA   44.4%
DE   43.3%
NJ   43.1%
HI   43.0%
MD   42.8%
CT   41.8%
NY   41.7%
MA   39.9%
RI   38.7%
DC   29.4%

Pubs 219 Dems 319

If Bush gets his average approval ratings to 51.9%, he wins....not exactly a challenging goal.

We're close to the inflection point here, so miniscule changes in overall popularity carry significant weight.

[cskendrick]

Two excellent posts Cskendrick!
Not much for me to add... except this: the most recent "matchup" poll for SC I could find (done last autumn) had Bush below the National average.
Food for thought.

Re: Props

Thanks! I'll take what I can get!

This is a tough room!

I find it difficult to imagine South Carolina turning on Bush, but then again I never thought I'd hear my mom saying that she intended to do so.

[cskendrick]

I tend to agree that a scenario where Bush is "Gored" is unlikely but you have to admit your analysis, while logical is simplistic.  The way it "could" happen is if some of the Bush states go for Bush, by wider margins than in 2000 (Florida would be a likely candidate), some Gore states go for Kerry (or Edwards) by a lesser margin than in 2000 (California is often cited as a possibility), but Kerry (or Edwards) manages to eke out a victory in, say Ohio, or NH & MO or some other combination which results in an electoral victory.  I don't see it, but it "could" happen.

It could happen, but if you believe that FL will go to Bush, it's almost imperative that the Democrat take OH to make it a competitive EV race. If the PV stays close, that would mean that the Democrat manages to come away with a margin in OH that is skewed way off the 2000 results, i.e., wins a significant number of Ohio Bush voters from 2000. (Bush had a 1% margin over Gore+Nader in OH in 2000, so the Democrat needs to shift 0.5% of total voters AWAY from Bush PLUS win all the 2.5% of Nader voters in that state.

Other states present even more of a challenge to the Dems in a tight race, as the Bush margins over Gore+Nader in 2000 are even higher.

I believe Bush can lose only if he fails to gain an additional 1% of the total national vote, to bring his total at least to 48.8%. The Dems, however, cannot win unless they get at least 50.4%, a gain of a full 2% of the total nationwide vote.

Re: Florida

Florida is a card the Republicans NEED, so it's best to assume that considerable effort will be exerted by the GOP to hold down the fort there.

Re: The resurrection of the Democratic rank and file

2000 was a moribund voter turnout for the Dems and quite frankly the party got exactly what it deserved for taking its electoral fortunes for granted.

Unperturbed, the Democratic leadership had to take a second dose of ASSumption Pie in 2002, counting on a historical trend that the President's party loses seats in the midterm, rather than making their own fate.

I think the Boys in Blue got the message finally: that this is a contest, and you must play hard to have any chance of winning.

I think it will be a close contest again, and so do the Republican campaign managers.