Rubicon

Electoral College Wrapup

Sometimes cleverness is not enough. Or, as critic Rex Reed Gore Vidal once said about the Cockettes, "Lack of talent is not enough." Case in point: my photo here of Alberto Giacometti's sculpture "The Nose," from the Hirshhorn Museum in Washington, was meant (all too cleverly) to evoke Pinocchio, which would immediately call to mind George Bush, who has been lying about so much the last few years that he would certainly be found out and punished severely, perhaps by losing the election. Yes, sometimes cleverness is not enough.

The averaging methodology I used for these electoral college updates, however, worked all too well, predicting on Monday a big win for Bush. As you can see, the final projection was wrong in only one state, Wisconsin, which went for Kerry. (I am, perhaps prematurely, calling Iowa for Bush on his 13,000-vote lead.)

My last futile attempt on Monday to allocate undecided voters, however, failed to take account of the fact that they refused to allocate themselves as I preferred. As a result, that map unfortunately did not turn out to be part of the reality-based community, much like the final Zogby projections.

There is much to be learned from the ups and downs of this campaign—though after the fact I must say the EC updates were finally little more than Fun with Numbers, and the Common Wisdom could have come just as close, with a lot less trouble. We are all fascinated with the horserace, it's true, and candidates need to know which strategies are clicking with the voters and which aren't, so conducting and analyzing polls is certainly a valuable skill. If you want to know more about it, Mark Blumenthal's your man, over at The Mystery Pollster. He posted an excellent preliminary wrapup this morning, and his tutorials on the fine points of the artful science are better than anything I ever saw in my years with the Census Bureau.

Thanks to all of you who have been following these updates and sending me your suggestions and ideas over the months. Today it's back to the hard work of dealing with the issues, which are even more important now that we are faced with another four years of Bush's catastrophic successes. Since you've been reading the updates, you may be interested in some of my earlier posts on other subjects (with guest posts now and again by John Shelton Lawrence):

• nonpolling aspects of the election
  
• ruminations on the American empire—this is Rubicon, after all
  
• the Iraq war
  
• religion, particularly as it is related to government
  
• terrorism
  
• or the other subjects listed under "Categories" in the left column

I welcome your comments and emails on anything you see here. For now, it's time to figure out what we should be doing in the next year or two to ensure a different result for the next election. That's clearly a cooperative effort, already begun at many of the blogs listed over to the right.

So take a short break, friends and comrades. Tomorrow it's back to the barricades!

Electoral College Update 35

Bush 296, Kerry 242 electoral votes. In this latest projection, based on averages of all state-by-state polls within the last 7 days, Wisconsin has shifted to Bush since Saturday. Several states are extremely close, however, so that in actuality the race is a dead heat. The outcome depends entirely on which party is more successful getting its voters to the polls.

The chart shows projections since March 18. Click for a larger view.

In the alternate calculation of swing states, the result is the same.  Among the 19 swing states in 2000, Iowa (7 EVs), New Mexico (5), and Wisconsin (10) are on track to change from Democratic to Republican, while New Hampshire (4) is moving from Republican to Democratic. This map represents the raw averages of polls within the last 7 days.

Many states remain extremely close. In six, the gap is less than 2 percent: Pennsylvania, where Kerry leads; and Florida, Iowa, New Mexico, Ohio, and Wisconsin, where Bush leads. These now appear to be the final battleground states. If either candidate wins all six, he will be elected 311–227.

In an election that includes an incumbent, the majority of poll respondents who remain undecided just before an election usually vote for the challenger. That majority is typically between 65% and 85%. When undecided voters are allocated in this way, the map looks very different, and Kerry wins. Here I deduct 2% for Nader, Badnarik, and other candidates, then allocate the remaining undecided voters 30% to Bush and 70% to Kerry. Both of these calculations probably understate the actual swing to Kerry, particularly since Nader is not on the ballot in Ohio or Pennsylvania. With this allocation, Iowa, New Mexico, Ohio, and Wisconsin move into the Kerry column by a miniscule margin, and Florida is tied. The result is Kerry 282 284, Bush 227, tied 27. This map represents the results with undecided voters allocated.

Tomorrow is the only poll that counts, and it's clear from these maps that the election is still up for grabs, especially in these six swing states, so be sure to vote. And if you're still undecided, I do have a suggestion: Kerry is an excellent choice, far superior to that other fellow.

The following table shows the current lead in each state, based on averages of state polls conducted within the last 7 days. Raw averages are shown in the left columns, and results with undecided voters allocated are in the right columns.

Electoral College Update 34

Bush 286, Kerry 252 electoral votes. In this latest projection, based on averages of all state-by-state polls within the last 7 days, Hawaii, Minnesota, and Wisconsin have shifted to Kerry in the last two days, and Ohio has shifted to Bush. You'll find the latest data here.

In the alternate calculation of swing states, the result is the same.  Among the 19 swing states in 2000, Iowa (7 EVs) and New Mexico (5) are on track to change from Democratic to Republican, while New Hampshire (4) is moving from Republican to Democratic.

Many states remain extremely close. In five, the gap is less than 2 percent: Minnesota and Wisconsin, where Kerry leads; and Florida, Iowa, and Ohio, where Bush leads. These now appear to be the final battleground states. If either candidate wins all five, he will be elected 306–232.

Click here for a spreadsheet showing the state-by-state leads. The projection will be updated one more time, on Monday, November 1—check late in the evening Pacific Time for new data. Click here for methodology.

Electoral College Update 33

Bush 290, Kerry 248 electoral votes. In this latest projection, based on averages of all state-by-state polls within the last 14 days, Hawaii, Minnesota, New Mexico, and Wisconsin have shifted to Bush since last week. You'll find the latest data here.

In the alternate calculation of swing states, the result is Bush 286, Kerry 252. Despite the slight lead Bush holds in Hawaii polls, Kerry is still likely to carry the state, so this may be a more accurate projection.  Among the 19 swing states in 2000, Iowa (7 EVs), Minnesota (10), New Mexico (5), and Wisconsin (10) are on track to change from Democratic to Republican, while Ohio (20) and New Hampshire (4) are moving from Republican to Democratic.

Many states remain extremely close. In five, the gap is less than 2 percent: Ohio, where Kerry leads; and Florida, Hawaii, Minnesota, and Wisconsin, where Bush leads. These four large states (not including Hawaii) are likely to be where the election is won or lost. If Kerry wins all four, he will be elected 299-239; if Bush wins all four, he will be elected 306-232.

Professional polls tend to be very reliable, assuming proper sampling, question design, interview technique, and other methodological niceties. Nevertheless, polls conducted on the same day and with the same population can yield different results. Furthermore, political polls ask respondents to say how they would vote today; by the time of the election, they may change their minds, so the data, though correct today, may be incorrect tomorrow. If you'd like to read more about these often perplexing issues, here are three good places to start:

• Richard Morin has an interesting overview of the polling business and its complexities in today's Washington Post.
  
• Mark Blumenthal, at his Mystery Pollster blog, answers nearly every question you can think of, and he writes in a cogent, understandable style. His recent pieces include an essay on tracking polls and a thorough, six-part explanation of how likely voters are defined. Notice the FAQs in the upper right margin.
  
• Ruy Teixeira, at the Donkey Rising blog, examines recent polls and explains the meaning of their internals and their biases. In the process, he provides tutorials on how to read polls and insight into the election process. His latest essay argues that this will be a high-turnout election, and that is good news for Democrats.

Click here for a spreadsheet showing the state-by-state leads. The projection will be updated two more times, on Saturday, October 30, and Monday, November 1—check late in the evening Pacific Time for new data. Click here for methodology.

Electoral College Update 32

Kerry 277, Bush 261 electoral votes. In this latest projection, based on averages of all state-by-state polls conducted within the last 14 days, Ohio and Wisconsin have moved into the Kerry column since last week. Kerry leads for the first time in six weeks and is within 0.3 percent in Florida. You'll find the latest data here.

The alternate calculation of swing states shows the same result. Among the 19 swing states in 2000, Iowa (7 EVs) is on track to change from Democratic to Republican, while Ohio (20) and New Hampshire (4) are moving from Republican to Democratic.

National polls continue to differ from one another in confusing ways, especially when measuring likely voters. Jim Rutenberg has a good piece in the New York Times this week explaining some of the intricacies. Pollsters, he writes, "say the differences in their surveys only highlight the difficulties this year in determining who is going to vote, no small task at a time of unusually high voter interest and many new voter registrations. And how pollsters set about figuring that out, they say, can make all of the difference in how the results are presented on television and in newspapers."

In Salon, Tim Grieve describes the problem:

They are the numbers flying by on the TV screen, and if you're living and dying by the presidential race, no amount of Zoloft can even out the highs and the lows they bring. George Bush and John Kerry are tied in the brand-new NBC News/Wall Street Journal poll. Kerry is up by 3 in a Democracy Corps poll released Tuesday. He was down by 3 in the New York Times poll Tuesday morning, down by 5 in the Washington Post poll Tuesday night, tied in last night's Zogby tracking poll, and down by an alarming 8 in the CNN/USA Today/Gallup poll released over the weekend.

Fortunately, he then provides four helpful guidelines for making sense of it all:

Electoral College Update 31

Bush 291, Kerry 247 electoral votes. In this latest projection, based on averages of all state-by-state polls conducted within the last 30 days, New Hampshire has moved into the Kerry column. You'll find the latest data here.

The alternate calculation of swing states shows the same result. Among the 19 swing states in 2000, Iowa (7 EVs) and Wisconsin (10) are on track to change from Democratic to Republican, and New Hampshire (4) from Republican to Democratic.

Bush remains ahead, but the lead is melting fast. Many of the state polls included in these averages were conducted before any of the presidential debates, and more recent polls show a significant shift to Kerry that is not yet reflected fully in the averages. This is true even in states that are strongly for Bush. For example, the Wilson poll in Oklahoma showed Bush ahead in that state by 30 points on October 3. One week later, the lead was reduced to 12 points. Oklahoma will doubtless still go Republican, but potentially by a relatively small margin.

If Kerry's surge continues, the total EVs will change quickly. Other recent individual polls show Kerry getting close in states that have been considered solid Bush. The latest Missouri poll shows a gap of only 1.8%, Arkansas 1.6%, and Tennessee 0.9%.

Here's another way of looking at the averaged numbers in this projection: if Florida (where the Bush spread is 2.5%) switches to Kerry, the senator wins 274-264. If Ohio (also 2.5%) switches, Bush wins 271-267, but only one additional state (for example, Iowa, where the Bush spread is less than 1%) gives the election to Kerry. These three states, along with Nevada and Wisconsin, all show Kerry within 3 points of Bush in the averages.

Click here for a spreadsheet showing the state-by-state leads. This projection is updated weekly—check Thursday evening Pacific Time for new data. Click here for methodology.

Electoral College Update 30

Bush 295, Kerry 243 electoral votes. In this latest projection, based on averages of all recent state-by-state polls, there are no changes from last week. You'll find the latest data here.

The alternate calculation of swing states shows the same result. Among the 19 swing states in 2000, Iowa (7 EVs) and Wisconsin (10) are on track to change from Democratic to Republican. Though no states have changed columns since last week, almost all recent state polls show a strong shift to Kerry following last week's debate, reducing the gap in Bush states and increasing the gap in Kerry states. The race remains extremely close, and the debate tomorrow night may well be as significant a milestone.

The methodology I use tends to lag recent events, since I average all polls within the last 30 days. Websites that aggregate only the most recent polls for each state have tended to show the Kerry surge more strongly: electoral-vote.com has Kerry 253, Bush 264, 21 tied; and 2.004k.com has Kerry 275, Bush 250, 13 tied. The averaging method I use has the advantage of smoothing out temporary variations and single-poll inaccuracies, but it does give a poll that is several weeks old as much value as one released yesterday. Therefore, results from Rubicon should be compared with projections that rely on the single most recent poll. Each type has advantages and disadvantages in teasing out what the numbers mean.

Polling organizations differ from aggregators, of course, in that their projections are based only on their own polls. Rassmussen Reports, for example, runs a national poll as well as state surveys. The state polls now add up to Kerry 169, Bush 240, toss-up 129; but many states have not been polled in the last week, so this result shows the same kind of lag effect as my projection, diluting the post-debate Kerry surge. This fact is made clear by Rasmussen's national tracking poll, which has no such lag and shows Kerry 46.7, Bush 47.8; and the 16-state battleground poll is tied at 47–47.

Zogby Interactive's battleground-state poll, however, is all Kerry, with the senator leading 322–216. Zogby shows only 3 of the 16 swing states going for Bush. For more details, click on the map.

Why do polls show such diverse results? It's not just the lag effect. Mark Blumenthal, writing as the Mystery Pollster, is a Democratic analyst who provides an expert's answer to that question. I recommend his other posts, too, for making sense of the numerical cacophony.

Guy Molyneux has an interesting perspective in The American Prospect on how voters respond in an election that includes an incumbent. He says that the spread is less important than whether the incumbent can remain over 50 percent, because most voters first decide how they feel about the incumbent, only later considering the challenger. As a result:

the incumbent’s percentage of the vote is a far better indicator of the state of the race than the spread. In fact, the percentage of the vote an incumbent president receives in surveys is an extraordinarily accurate predictor of the percentage he will receive on election day -- even though the survey results also include a pool of undecided voters. Hence the 50-percent rule: An incumbent who fails to poll above 50 percent is in grave jeopardy of losing his job.

Other studies have indicated that a challenger may capture up to 80 percent of undecided voters in the final days of a campaign. That's good news for Kerry, but not good enough to relax. The forces of Mammon may still have a trick or two up their sleeve.

This projection is updated weekly—check Thursday evening Pacific Time for new data.

Electoral College Update 29

Bush 295, Kerry 243 electoral votes. In this latest projection, based on averages of all recent state-by-state polls, Pennsylvania has moved back into the Kerry column. You'll find the latest data here.

The alternate calculation of swing states shows the same result. Among the 19 swing states in 2000, Iowa (7 EVs) and Wisconsin (10) are on track to change from Democratic to Republican. The race remains extremely close, and the debate tonight is likely to be a significant milestone.

National polls continue to vary widely. Among registered voters, the IBD/CSM/TIPP poll released Monday shows a tie, while Sunday's CNN/USA Today/Gallup report has Bush ahead by 13 points; among likely voters, IBD/CSM/TIPP has a 1-point lead for Kerry, and CNN/USA Today/Gallup an 8-point lead for Bush.

Steve Soto at The Left Coaster has been communicating with Gallup about their Republican-heavy sample, and in his latest scathing report, he reports that Gallup "used a 12% GOP bias in their likely voter screen. I kid you not." He continues:

Here is the text from the email I got from Gallup this morning outlining the party ID breakdown in their likely voter samples from their two most recent national polls:

Likely Voter Sample Party IDs – Poll of September 13-15
Reflected Bush Winning by 55%-42%

Total Sample: 767
GOP: 305 (40%)
Dem: 253 (33%)
Ind: 208 (28%)

Likely Voter Sample Party IDs – Poll of September 24-26
Reflected Bush Winning by 52%-44%

Total Sample: 758
GOP: 328 (43%)
Dem: 236 (31%)
Ind: 189 (25%)

Looking at this, again I have a simple question: how can anyone, especially USA Today and CNN, let alone the rest of the media take a Gallup national poll seriously when Gallup knowingly puts a poll out there for consumption with a 12% GOP bias in its likely voter sample that everyone knows does not exist in the country today or at any time in the last three presidential elections?

Yet this flawed poll showed a narrowing Bush lead from their similarly flawed poll of two weeks ago. So if a poll with an unsupportable GOP bias of 12% in its likely voter sample, shows an 8% Bush lead amongst likely voters when a poll they used two weeks ago with a 7% GOP bias showed a 13% Bush lead with likely voters, then how can anyone not conclude that Kerry is doing much better than Gallup would have you believe?

By presenting these polls with this kind of bias, and then ensuring through CNN and USA Today the farthest possible media saturation, why is Gallup not guilty of engaging in a political disinformation campaign?

In other words, Soto implies, it's not outlier data, it's liar data. Odd, of course, since Gallup is one of the oldest polling organizations around, and it has long been thought of as the gold standard. But the results are so bizarre that Ruy Teixeira has a novel idea:

Electoral College Update 28

Bush 316, Kerry 222 electoral votes. In this latest projection, based on averages of all recent state-by-state polls, Colorado, Iowa, New Hampshire, and Pennsylvania have moved into the Bush column. You'll find the latest data here.

The alternate calculation of swing states shows the same result. Among the 19 swing states in 2000, Iowa (7 EVs), Pennsylvania (21), and Wisconsin (10) are on track to change from Democratic to Republican.

These shifts are somewhat discouraging, but the gaps are small and reversible. By my calculations, Bush leads by less than 3 percent in 6 states with a total of 70 electoral votes, while Kerry's lead is 3 percent or greater in every state where he is ahead. The 6 states now barely in the Bush column are Arkansas, Florida, Iowa, Nevada, New Hampshire, and Pennsylvania; if they were to turn around, Kerry would retake the lead 292 to 246. Furthermore, these are not the only states in play. There are 13 more states where one candidate or another has a slightly larger lead, 3.0 to 6.3 percent: 7 states with 83 EVs for Bush and 6 states with 58 EVs for Kerry. That's 212 electoral votes that could go either way. Despite the numbers at the top of this post, this remains an extremely close race.

Slight differences in methodology yield more heartening results. Both American Research Group (ARG) and Zogby International released results of extensive state polling this week, and both organizations found Kerry with a miniscule lead in electoral votes. ARG conducted about 30,000 interviews, 600 in every state and the District of Columbia, between 9/7 and 9/21. Here's ARG's summary:

How close is the race for president?

• George W. Bush is at 47% and John Kerry is at 46% in the weighted national popular vote.
  
• Bush leads outside the margin of error in 17 states with 133 electoral votes.
  
• Kerry leads outside the margin of error in 10 states with 132 electoral votes.
  
• Bush has any lead in 29 states with 253 electoral votes.
  
• Kerry has any lead in 20 states with 270 electoral votes.
  
• Bush and Kerry are tied in Wisconsin and West Virginia.
  
• Bush needs to defend small leads in 5 states - Colorado, Iowa, Nevada, New Hampshire, and Ohio.
  
• Kerry needs to defend small leads in 5 states - Maine, Florida, Minnesota, Oregon, and Pennsylvania.
  
• Among men nationwide, 51% say they would vote for Bush and 42% say they would vote for Kerry.
  
• Among women nationwide, 42% say they would vote for Bush and 50% say they would vote for Kerry.

That's Kerry by a whisker in a photo finish: 270 is enough to win.

Zogby conducted its usual biweekly poll from 9/13 through 9/17 in 16 "battleground" states (assuming all other states will vote for the same party as in 2000) and concluded that Kerry is ahead of Bush 297 to 241. Click on the map for more information.

Click here for methodology. This projection is updated weekly—check Thursday evening Pacific Time for new data.

Electoral College Update 27

Bush 275, Kerry 254 electoral votes. In this latest projection, based on averages of all recent state-by-state polls, Florida and Wisconsin have moved into the Bush column. You'll find the latest data here.

The alternate calculation of swing states shows the same result. Among the 19 swing states in 2000, New Hampshire (4 EVs) and Wisconsin (10 EVs) are on track to change parties, NH to Kerry, and WI to Bush. Colorado, a Bush state in 2000, is tied.

The race remains on a knife edge. By my calculations, Bush leads by about one-half of one percent in both Florida and Wisconsin, and there continues to be uncertainty about some polls' techniques for determining likely voters (LVs) and party identification. In fact, polls that measure registered voters (RVs) are almost certainly more reliable at this point in the race, though the LV polls tend to get more media play.

Ruy Teixeira has written two interesting pieces this week, one on the effect of these methodological issues on the state of the race and another on Kerry's gains among "persuadable voters." He also refers to a Democracy Corps report by James Carville and Stanley Greenberg that says:

1. The president has a lead of about 5 points.
   
2. A majority of the country still wants change.
   
3. Part of the stability of this race are the grave doubts voters feel about George Bush.
   
4. While Bush has gained on having plans for Iraq and it bringing more security, a stable majority says the war was not worth it.
   
5. And on the economy, voters by nearly two-to-one reject Bush’s assertion that the middle class is making gains.

Many of the results in this report demonstrate significant strengths for Kerry. (Click on the chart for details.)

Click here for methodology. This projection is updated weekly—check Thursday evening Pacific Time for new data.