Electoral College Breakdown 2004: April 28th Update

I Wouldn't Touch It With a 10 Foot Poll

Magic Town

I have a weakness for Jimmy Stewart movies. Sit me down in front of "The Man Who Shot Liberty Valence" and I am a very happy camper. Every Christmas, I am the sentimental sap who does not run screaming at TBS's eight millionth airing of "It's a Wonderful Life". And as a budding conservative, while growing up I loved the idealism and triumph of old-fashioned patriotism and virtue over abusive government and corruption in "Mr. Smith Goes To Washington".

A less known Stewart movie was about polling. In "Magic Town", Stewart played a pollster who uncovers a method to do remarkably accurate public opinion polls. His "mathematical miracle" was merely the discovery of "Grandview", a small heartland town so representative of the country in aggregate that just by polling this one town, he could be more accurate than Gallup, at a fraction of the cost and in significantly less time.

The problems started when newspaper editor Jane Wyman discovered his secret and made it the news. Suddenly self-aware, the townspeople started taking on an aura of responsibility, knowing the nation was listening to them. They realized that they were not just providing a measure of the nation, but also influencing the nation. They started researching issues, going so far as to provide reference libraries to be consulted before being polled.

They changed.

Within short order, the people of Grandview became the butt of jokes nationwide for their preposterous views. Where once they were perfectly representative of the nation, they ended up with views so different that people wondered if they were really in the United States at all.

Political polling first began, as mentioned in a previous article's quiz, during the election of Warren Harding. Literary Digest conducted the poll by sending out postcard to voters in the six states considered to be most competitive. The names were selected from phone books and from automobile registration. The poll successfully predicted Harding's victory, and polling was here to stay. The Literary Digest repeated their poll in subsequent elections, always getting the answer right, straight through the election of Franklin Delano Roosevelt in 1932. The Literary Digest may not have found as cheap or fast an answer as Jimmy Stewart, but they had the accuracy of "Magic Town" down pat.

The poll in the run-up to the 1936 election was their most ambitious ever. Over ten million surveys were mailed out, with over two million returned. In these days of polls of less than 1000 voters, it is hard to fathom a poll with a sample size of 2,000,000. Given the voting population, such a poll had a margin of error of less than a tenth of a percent! The poll predicted a 57%-43% election, as compared to the 61%-37% one that actually happened. The margin of error was demolished, especially when one considers that the poll predicted the landslide victor to be Alfred Landon, and not FDR.

In retrospect, what went wrong for the Literary Digest pollsters is easy to understand. Their ballots were mailed to people with phones; not many people had phones and those who did were richer than most. Their ballots were mailed to people with cars; not many people had cars and those who did were richer than most. It is safe to say that among the voters who owned cars or had a phone, Landon scored a significant victory. When the Literary Digest first started, it is very possible that the affluent's views were representative of the general public's views to a sufficient degree that by polling them, relatively accurate results were obtained. If that was the case, however, by 1936 it no longer was. If Grandview had changed here, or if the nation had, the result was the same. Those polled no longer represented those who would vote.

By the year 2000, polling techniques were significantly more advanced. In addition, the general wealth of the nation had changed to where even our poor tended to have phones. Pollster Scott Rasmussen decided to search for his own Grandview with an automated polling system, which would use recorded scripts and would tally the replies automatically. This method was considerably cheaper than methods that involved people doing the calling (or going door to door), and considerably faster than polling by mail.

But to really have found "Magic Town", he needed to not just be faster and cheaper, but to be as accurate as all of the other pollsters. To do so, he used all the techniques that have evolved over the years. He used techniques for controlling the sample by measuring demographics, to try to ensure that the sample remained representative of the nation as a whole. Since the people who show up on Election Day are different than the pool of all registered voters, he used some methodology for determining who were the likely voters.

Mr. Rasmussen put out a ton of polls during the 2000 campaign. Many were right in line with what other polling companies were putting out. He was beating them in cost. He was beating them in frequency. But in the end, he had President Bush leading by 6 points in the popular vote-- and it was not a one time 'outlier', as he had shown the lead to be that large for several days in his tracking poll. It turned out that he was not polling Grandview, but rather its slightly more Republican neighboring town of GrandOldPartyView.

This is not meant as a slam on Mr. Rasmussen, not in the least. His poll results, this election cycle, are so clearly different than last that it is obvious he has tried to learn from his 2000 experience and has probably made some changes. Many sniff at his technology, saying that an automated poll cannot be as accurate as a conventional poll. SurveyUSA disagrees. They have been very accurate of late, and their accuracy will continue just as long as their ability to figure out who a likely or certain voter remains. Just when one thinks they have found Grandview, the people of Grandview change.

Where Mr. Rasmussen was in 2000 trying to build a reputation, pollster John Zogby was cementing his. If one was to survey those who follow polls to some degree, which pollster was most accurate in 2000, more times than not one would get as an answer "Zogby"; this is despite the fact that he was not (Harris was most accurate on the national level, and Mr. Zogby's state polls were hit or miss). While his reputation may be inflated beyond where it should be, he is still a very solid pollster. It is possible to be both overrated and still quite good. And on the race that mattered, Mr. Zogby was famously very close, as he had been in a few earlier high visibility races.

For conducting his polls, Mr. Zogby uses more conventional methods of having people calling and reading from a script, rather than using an automated system. Unique to a Zogby poll, however, is how stringent a mathematical model is used to color the results. He scales over-represented demographics down. He adjusts for which areas he projects to have higher turnout. The raw data and the presented data in a Zogby poll can be markedly different. It can make him extremely accurate when his turnout model successfully predicts what the actual turnout will be like in various areas and among various demographics.

But when his predicted turnout model is off the mark? Then he is wrong. Take 2002 as an example. "I blew Illinois. I blew Colorado and Georgia. And never in my life did I get New Hampshire wrong," he said, "but I blew that too." His map said he was in Grandview, but his directions had taken him somewhere else. In 2000, he knew Democrat turnout was going to be higher than most pollsters' turnout models were figuring. In 2002, he missed that Republicans had tuned their get out the vote efforts to where their turnout would be higher.

Both Mr. Rasmussen's problems in 2000 and Mr. Zogby's in 2002 were likely due to the fact that the universe of who they considered to be likely voters was not actually representative of who were really likely to vote. They made assumptions, and the assumptions were blended in with the results, and the total error was not merely that which random chance dictated but that plus the amount by which their assumptions were wrong. Other pollsters try to avoid this bias point by using other means for determining who is a likely voter. Gallup, for example, simply asks if a respondent is likely to vote or not. This works in the years where those who say they will show up, show up, and it fails in the years where those who show up are significantly different than those who say they will. Just because someone says they are from Grandview does not make it so.

Still others avoid the potential addition of a biasing influence or additional error by simply going with a poll of registered voters, or of adults. This does not get away from the core problem, however, in that usually the people who end up voting have a different makeup than the entire universe of potential voters-- which is the primary reason that likely voter polls were developed in the first place.

Just as there is no such thing as a perfectly accurate poll, there is also no "Magic Town", other than in the movies. While I do not dispute the fact that, overall, likely voter polls give results closer to the actual election results, I prefer registered voter polls. They may show the race being closer or wider than it actually is, but I do not have to guess as to what sort of bias is being introduced in paring it down to a likely voter. I can use my own judgment, and my own analysis based on the results of other questions, to figure out what the complexion of the race actually is, especially if they give the demographic breakdowns and the answers to other questions such as favorability.

The citizens of Grandview realized that they were not just stating the opinion of the nation, but that they could move the opinion of the nation (at least until they became so unrepresentative that they became a travesty of a sham of a mockery- a travshamockery). The determination of who is a likely voter gives that same molding ability, to some degree, to the pollsters. I would rather they not have it.

This Week's Polling Updates Overview

Currently, Bush leads in the ECB by a very slight 216-210 tally, and by 258-217 if slight advantage states are factored in.

	Florida
Electoral Votes: 27
2000 Result
Bush 48.85%
Gore 48.84%

Background: Despite the best efforts of the results-oriented Florida Supreme Court, Bush held on to win the state in 2000, just as every recount conducted afterwards validated. Did you know that since 1948, though, that only three times has Florida gone for the Democrat candidate? Johnson got 51%, Carter got 52%, and Clinton (2nd term) got 48% (with Perot taking 9%). More times than not, the Republican has come closer to 60%. Why Bush underperformed here to such a degree is something his campaign must rectify.

In the first ECB of 2000, Florida was listed as a battleground with a slight advantage to Gore. This time around, it is starting with a slight advantage for Bush. Florida has 6 Democrat Representatives and 18 Republicans. Both chambers of the state legislature are controlled by the Republicans. Republicans control most of the executive branch. However, both Senate seats are held by Democrats. As of Dec. 1, 2003, the state registration was 41.9% Democrat and 38.6% Republican.

Date	Polling Company	Link	Type	MOE	Republican	Democrat	Margin
4/29/03	Mason-Dixon	Link	RV	5%	Bush	53%	Unnamed Democrat	38%	Bush +15
12/3/03	Schroth & Associates	Link	800 RV	3.5%	Bush	43%	Unnamed Democrat	37%	Bush +6
1/15/04	Rasmussen Reports	Link	LV	5%	Bush	47%	Unnamed Democrat	45%	Bush +2
2/27/04	Research 2000	Link	500 LV	4%	Bush	47%	Kerry	42%	Bush +5
3/4/04	American Research Group	Link	600 LV	4%	Bush	44%	Kerry	45%	Kerry +1
3/4/04	Schroth & Associates	Link	800 RV	3.5%	Bush	43%	Kerry	49%	Kerry +6
3/14/04	Rasmussen	Link	400 LV	5%	Bush	45%	Kerry	48%	Kerry +3
4/1/04	Mason-Dixon	Link	625 RV	4%	Bush	51%	Kerry	43%	Bush +8
4/13/04	Rasmussen Reports	Link	500 LV	5%	Bush	46%	Kerry	47%	Kerry +1
4/21/04	American Research Group	Link	600 LV	4%	Bush	46%	Kerry	45%	Bush +1

Punditry: No real change. For such an exciting state, Florida is being boringly predictable so far this year. The Mason-Dixon poll is looking as if it was a bit optimistic for the President. Tossup.

	F		Iowa
Electoral Votes: 7
2000 Result
Gore 48%
Bush 46%

Background: Iowa is a state of streaks, going 1-5-4 over the last 10. Clinton would likely have lost his first campaign against Bush had Perot not been a factor. The state is generally close, with the only surprisingly large margin coming when Dukakis beat Bush by 10 points.

Iowa rated a slight advantage to Bush in the first ECB of 2000. This time, it rates a slight advantage to the Democrats. Other positions in Iowa are mixed. The Republicans hold 4 of the 5 House seats, and the Senate seats are split. The Republicans control both chambers of the state legislature, but the Democrats hold all major executive offices except for Auditor. Republicans hold a 32% to 29% advantage in registration.

Date	Polling Company	Link	Type	MOE	Republican	Democrat	Margin
9/12/03	Des Moines Register	Link	803 Adults	3.5%	Bush	41%	Unnamed Democrat	41%	Even
10/30/03	Research 2000	Link	LV	4%	Bush	46%	Unnamed Democrat	47%	Dem +1
1/5/04	Research 2000	Link	LV	4%	Bush	50%	Unnamed Democrat	42%	Bush +8
2/11/04	Selzer & Co.	NA	RV	3.4%	Bush	42%	Kerry	49%	Kerry +7
3/23/04	Rasmussen	Link	500 LV	4.5%	Bush	41%	Kerry	51%	Kerry +10
4/21/04	American Research Group	Link	600 LV	4%	Bush	46%	Kerry	47%	Kerry +1

This was a very interesting poll in that it is amazingly even across the board. The sample had equal numbers of Republicans and Democrats, which is appropriate in Iowa considering the parties have similar registration numbers. The candidates are within 1 point among independents. Each candidate gets 6% of the other candidate's party. Both get comparable numbers from within their own party. This state is as close as can be. Given that Kerry had previously had a lead, I will designate it as a Slight Advantage for Kerry, although a good argument could be made to call it a complete tossup.

	F		New Hampshire
Electoral Votes: 4
2000 Result
Bush 48%
Gore 47%

Background: Clinton won twice here, and the second time would have been even without Perot. The first time, he likely would have lost.

Johnson won here. Other than that, Republican wins back through Dewey beating Truman. The elder Bush crushed the man who Kerry was Lt. Governor

Both last time and this time, New Hampshire started as a slight advantage for Bush. When looking at the other offices, it is hard to understand just why this state is not more firmly in his control. The Republicans hold all the House seats, both Senate seats, control both chambers of the state legislature, and hold all major executive branch offices, while having a 37%-26% registration advantage.

Date	Polling Company	Link	Type	MOE	Republican	Democrat	Margin
4/27/03	Franklin Pierce College	Link	600 RV	4%	Bush	42%	Unnamed Democrat	42%	Even
9/11/03	American Research Group	Link	RV	4%	Bush	50%	Unnamed Democrat	36%	Bush +14
10/16/03	Research 2000	Link	RV	4%	Bush	51%	Kerry	39%	Bush +12
12/10/03	American Research Group	NA	RV	4%	Bush	50%	Unnamed Democrat	38%	Bush +12
10/16/03	Research 2000	Link	RV	4%	Bush	51%	Kerry	39%	Bush +12
1/20/04	Associated Press	Link	600 LV	4%	Bush	55%	Kerry	40%	Bush +15
2/20/04	University of New Hampshire	Link	511 LV	4%	Bush	38%	Kerry	53%	Kerry +15
3/18/04	American Research Group	Link	463 RV, Nader an option	4.6%	Bush	45%	Kerry	39%	Bush +6
3/18/04	American Research Group	Link	463 RV, Nader not an option	4.6%	Bush	47%	Kerry	45%	Bush +2
4/1/04	American Research Group	Link	600 LV, Nader an option	4%	Bush	48%	Kerry	43%	Bush +5
4/1/04	American Research Group	Link	600 LV, Nader an option	4%	Bush	48%	Kerry	45%	Bush +3
4/22/04	Rasmussen	Link	500 LV	4.5%	Bush	45%	Kerry	47%	Kerry +2

Punditry: Rasmussen's recent tendency of being a few points less favorable to the President than other state polls continues. Tossup. Interestingly, despite the Kerry advantage according to Rasmussen, Bush polls better on the economy. Go figure.

	California
Electoral Votes: 55
2000 Result
Gore 53%
Bush 42%

Background: On a three election streak for the Democrats, California has a reputation as a liberal bastion. While Gore did handle Bush easily in 2000, the fact is that the reputation may not fit the data on the Presidential level. Only three candidates have broken 53% in California since the 1964 landslide. Al Gore last time, homestate icon Ronald Reagan in his re-election campaign but not his first election, and Richard Nixon in his re-election campaign but not his first successful Presidential campaign.

Date	Polling Company	Link	Type	MOE	Republican	Democrat	Margin
8/16/03	Field	NA	RV	4%	Bush	42%	Unnamed Democrat	47%	Dem +5
8/16/03	Public Policy Institute	NA	LV	3%	Bush	40%	Unnamed Democrat	45%	Dem +5
1/3/04	Public Policy Institute	Link	LV	3%	Bush	45%	Unnamed Democrat	45%	Tied
1/13/04	Field	NA	RV	3.4%	Bush	46%	Unnamed Democrat	47%	Dem +1
1/18/04	Rasmussen	NA	LV	4%	Bush	41%	Unnamed Democrat	46%	Dem +5
2/13/04	Knowledge Networks	Link	RV	4.1%	Bush	38%	Kerry	42%	Dem +4
2/16/04	Public Policy Institute	Link	1,103 LV	3%	Bush	37%	Kerry	54%	Dem +17
2/22/04	LA Times	Link	1,521 RV	3%	Bush	40%	Kerry	53%	Dem +13
2/27/04	Knowledge Networks	Link	505 RV	3.8%	Bush	38%	Kerry	43%	Kerry +5
3/11/04	Rasmussen	Link	455 LV	5%	Bush	44%	Kerry	53%	Kerry +9
4/17/04	Rasmussen	Link	502 LV	5%	Bush	40%	Kerry	51%	Kerry +11
4/21/04	L.A. Times	Link	1,265 LV, Nader an option	3%	Bush	39%	Kerry	49%	Kerry +10
4/21/04	L.A. Times	Link	1,265 LV, Nader not an option	3%	Bush	41%	Kerry	53%	Kerry +12

Punditry: Same basic result as the LA Times poll. Same basic result as the 2000 election. Same designation as before: Strong for Kerry.

	New York
Electoral Votes: 31
2000 Result
Gore 60%
Bush 35%

Background: From 1960 onward, Republicans have carried the Empire State only three times. Nixon beat McGovern, Reagan beat Carter, and Reagan beat Mondale. Even Dukakis won here.

Date	Polling Company	Link	Type	MOE	Republican	Democrat	Margin
4/03	Marist	Link	RV	4%	Bush	32%	Unnamed Democrat	39%	Dem +7
9/23/03	Marist	Link	RV	4%	Bush	32%	Unnamed Democrat	48%	Dem +16
10/28/03	Quinnipiac	NA	RV	4%	Bush	42%	Kerry	50%	Dem +8
11/19/03	Zogby	Link	LV	4%	Bush	41%	Kerry	46%	Dem +5
1/7/04	Marist	Link	617 RV	4%	Bush	34%	Unnamed Democrat	36%	Dem +2
4/12/04	Quinnipiac	Link	1,279 RV, Nader an option	2.7%	Bush	35%	Kerry	49%	Kerry +14
4/12/04	Quinnipiac	Link	1,279 RV, Nader not an option	2.7%	Bush	36%	Kerry	53%	Kerry +17
4/15/04	Marist	Link	602 RV	4%	Bush	38%	Kerry	56%	Kerry +18
4/22/04	Siena Researh Institute	Link	625 RV	3.9%	Bush	32%	Kerry	51%	Kerry +19

	Oregon
Electoral Votes: 7
2000 Result
Gore 46.96%
Bush 46.52%

Background: The last 10 elections have gone 1-5-4. Without Perot, Clinton would likely have lost his initial run here though. From 1968 on, the only Democrat to break 50% here was Michael Dukakis; Reagan and Nixon each broke 50% in their re-elect.

In early 2000, Oregon was polling strongly for Bush. The left coast influence eventually took hold and turned it into a very even state, and it starts this year as a tossup. Democrats hold 4 out of 5 Representative seats, while the Senate seats are divided, just as control of the state legislative chambers is split. The Democrats hold most of the executive branch positions. Democrats hold a 3% lead in registration, 39%-36%.

Date	Polling Company	Link	Type	MOE	Republican	Democrat	Margin
3/5/04	Hibbits	Link	?	?	Bush	40%	Kerry	45%	Kerry +5
4/7/04	University of Oregon	Link	440 RV	4.7%	Bush	47%	Kerry	45%	Bush +2
4/25/04	Rasmussen	Link	500 LV	4.5%	Bush	45%	Kerry	46%	Kerry +1

Punditry: From Bush +2 to Kerry +1, which means that the categorization of Tossup holds steady.

	F		Pennsylvania
Electoral Votes: 21
2000 Result
Gore 51%
Bush 46%

Background: Democrats have won this state 6 of the last 10 elections, with the first of Clinton's wins being attributable to Ross Perot being on the ballot (19%). Typically, the races in the Keystone state have been close. Pennsylvania started as leaning Bush last time, but has drifted to where it has a slight advantage for the Democrats. Republicans hold a 12-7 advantage in the numbers of Representatives, and hold both Senate seats. They also hold both chambers of the state legislature. The Democrats hold the major executive branch positions except for Attorney General, and have a significant registration advantage (48%-42%).

Date	Polling Company	Link	Type	MOE	Republican	Democrat	Margin
5/13/03	Quinnipiac	Link	952 RV	3.2%	Bush	56%	Kerry	34%	Bush +22
10/9/03	Quinnipiac	Link	1,116 RV	3%	Bush	50%	Kerry	43%	Bush +7
11/23/03	Muhlenberg College	Link	430 RV	4.7%	Bush	47%	Kerry	40%	Bush +7
12/14/03	Quinnipiac	Link	1,092 RV	3%	Bush	50%	Kerry	42%	Bush +8
2/18/04	Quinnipiac	Ling	1,356 RV	2.7%	Bush	45%	Kerry	50%	Dem +5
2/22/04	Keystone Poll	Link	392 RV	4.9%	Bush	46%	Kerry	47%	Dem +1
3/3/04	Pennsylvania Public Mind	Link	1750 Adults	2.4%	Bush	45%	Kerry	47%	Kerry +2
3/11/04	Survey USA	Link	802 RV	3.5%	Bush	47%	Kerry	49%	Kerry +2
3/15/04	Qunnipiac	Link	1,022 RV (Nader not given as an option)	3	Bush	44%	Kerry	45%	Kerry +1
3/15/04	Qunnipiac	Link	1,022 RV (Nader given as an option)	3%	Bush	44%	Kerry	40%	Bush +4
3/16/04	Rasmussen	Link	500 LV	4.5%	Bush	44%	Kerry	45%	Kerry +1
3/29/04	Keystone Poll	Link	565 RV	4.1%	Bush	46%	Kerry	40%	Bush +6
4/19/04	Quinnipiac	Link	769 RV, Nader not an option	3.5%	Bush	46%	Kerry	42%	Bush +4
4/19/04	Quinnipiac	Link	769 RV, Nader an option	3.5%	Bush	45%	Kerry	39%	Bush +6
4/25/04	Pew Research	Link	867 RV	4%	Bush	42%	Kerry	42%	Push

Punditry: Madeline Albright's polling company shows a more favorable result for the Democrat than previous recent polls have, and I am dropping Pennsylvania back into the battleground range as a Slight Advantage for Bush. There are some odd internals to this poll. It has Kerry leading by 4 points among men, and Bush leading by 4 among women. It has 4% of Republicans voting for Nader, compared to 3% of Democrats and a whopping 12% of independents. It has Kerry winning in the Southeast by 15 and in the Southwest by 7, and Bush winning by 15 in the rest of the state.

Kerry E		F Bush
Safe	Strong	Lean	Slight	Tossup	Slight	Lean	Strong	Safe
DC (3)	DE (3)	MD (10) K48-B43 3/24/04	IA (7) K47-B46 4/21/04	NM (5) B46-K45 4/1/04	WI (10) B49-K45 3/31/04	VA (13) B48-WC33 12/3/03	SC (8) B52-UD36 7/28/03	AK (3)
HI (4)	VT (3) HD50-B38 10/2/03	MN (10) K50-B38 4/2/04	-	OH (20) B46-K45 4/2/04	AR (6) B47-K45 4/15/04	GA (15) B47-UD43 2/4/04	MS (6) B49-UD29 12/22/03	ND (3)
RI (4) K53-B31 2/7/04	ME (4) K51-B38 3/4/04	MI (17) K51-B41 4/4/04	-	FL (27) B46-K45 4/21/04	WV (5) B46-K41 4/15/04	NC (15) B51-K43 3/11/04	SD (3) B50-UD39 2/5/04	NE (5)
MA (12) K54-B32 4/5/04	IL (21) K47-B39 3/13/04	WA (11) K46-B41 4/5/04	-	NH (4) K47-B45 4/22/04	PA (21) B46-K46 4/25/04	NV (5) B49-K38 3/17/04	KY (8) B57-K41 2/16/04	WY (3)
-	CT (7) K52-B33 3/28/04	NJ (15) K51-B39 4/16/04	-	OR (7) K46-B45 4/25/04	-	AZ (10) B51-K42 3/18/04	KS (6) B57-K39 3/4/04	MT (3) B52-UD27 5/16/03
-	CA (55) K53-B41 4/21/04	-	-	-	-	TN (11) B52-K41 3/22/04	IN (11) B52-K37 3/24/04	TX (34) B54-K35 3/6/04
-	NY (31) K51-B32 4/22/04	-	-	-	-	MO (11) B49-K42 3/23/04	OK (7) B47-K35 4/1/04	ID (4) B55-K23 3/17/04
-	-	-	-	-	-	CO (9) B49-K44 4/14/04	LA (9) B52-K38 3/28/04	AL (9) B59-K27 3/18/04
-	-	-	-	-	-	-	-	UT (5) B66-K24 3/25/04
Totals
Kerry States	Battleground States	Bush States

23	124	63	7	63	42	89	58	69

210	112	216

Discuss ECB2004 On Free Republic

This week's quiz:Which President was the first one who was born in the United States of America?

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.