July 9, 2008
The State of the Race
Part One
In the month since the end of the Democratic nominating contest, the Presidential race has settled into a fairly narrow band in Obama-McCain polling results. The two daily tracking polls, Gallup and Rasmussen, have consistently shown an Obama lead. The Rasmussen results have been more stable , with the Obama bounce moving as high as a 7 point lead after the last primary and Hillary Clinton's departure from the race, and never less than 3. Gallup has shown a lead for Obama of as much as 7, and a few days when the race was tied.
On Tuesday, Rasmussen had Obama with a 6 point lead, and Gallup showed a 2 point Obama edge (within the margin of error of the survey). The three most recent national polling results (CNN, Democracy Corps, a Democratic polling outfit, and Time) all showed 4 or 5 point Obama leads. The average of the two daily tracking polls and the 3 latest national polls gives Obama a 4.2% lead.
In essence, there is a lot of confirmation of a stable, reasonably close race; Obama with a lead in the 4-5% range. Nate Silver on his fivehtiryeight.com site (538 is the number of Electoral College votes) is in the same ballpark, believing Obama may have as much as a six point lead at the moment, though he suggests this overstates things a bit, since in most recent Presidential races the trailing candidate closed a bit near the end.
Three possible outcomes
The various betting sites suggest that Obama is given about a two thirds chance of winning, McCain about one third. This seems about right to me, since I think the three possible outcomes are a close McCain win (perhaps without a popular vote plurality), a close Obama win, and a blowout Obama win, all about equal possibilities.
The good news for John McCain is that he is not behind by 15-18 points, as Bob Dole was in 1996 against Bill Clinton at this stage of the race. Given the current political environment, he should be. President Bush has a 30% job approval rating. Over 80% of Americans think the country is headed in the wrong direction. Barack Obama skipped out of the federal funding system for the general election, and will likely raise two to three times the $84 million provided to John McCain.
On the left, unions, pro choice and environmental groups, MoveOn.org, and 527 groups funded by the likes of George Soros, Peter Lewis and the Sandlers, will far outspend similar groups on the right (when Obama said he needed to opt out of the federal financing system to level the playing field against the spending of right wing 527s, he was lying). Obama will have a huge money advantage across the board.
The RNC currently has more cash on hand than the DNC, but this may soon change as well, as the Obama campaign is conducting many joint Obama/DNC fundraising events. The Obama campaign has been very systematic and organized in registering new voters, and in its campaign messaging. The McCain campaign has been unfocused as far as its message, largely wasting the three months after wrapping up the nomination, and is going through another change in the campaign management team and structure.
Another big Obama advantage (as the candidate with the "change" agenda) is the state of the economy, with high gas and food prices, rising unemployment, the housing industry collapse, and a very weak stock market. Overseas, two wars continue in Iraq and Afghanistan (Iraq markedly improving for our side, Afghanistan stable at best) and there is the possibility of a new conflict with Iran over its nearly completed nuclear weapons program.
The closeness of the race suggests that many Americans are not yet sold on Barack Obama. The Illinois Senator had his worst weeks of the campaign when the Reverend Wright videos surfaced and then again after the Reverend went off at the National Press Club. Throw in Obama's remarks in San Francisco about rural/small town and working class voters clinging to guns and God , and it is likely that some doubts about Obama which developed during this period have lingered on.
Three key states
So too, while Obama has improved his standing in the polls since the end of the nominating race, his numbers remain weak in three important battleground states, Ohio, Michigan and Florida, with 58 Electoral College votes among them. In 2004, Bush won Florida by 5%, Ohio by 2%, and lost Michigan by just over 3%, while winning nationally by 2.4%. If one looks at the state vote versus the national vote, Florida was a slightly stronger than average Republican state, Ohio just about average, and Michigan a much stronger Democratic state than the national average. If Obama is ahead by 4% at the moment nationally, then in Michigan he is running a bit weaker than his national number (whereas Kerry ran 6% better), in Florida he is running as much as 9-10% weaker than his national number (Kerry 2% weaker), and in Ohio, about the same as his national lead.
So too, while Obama has improved his standing in the polls since the end of the nominating race, his numbers remain weak in three important battleground states, Ohio, Michigan and Florida, with 58 Electoral College votes among them. In 2004, Bush won Florida by 5%, Ohio by 2%, and lost Michigan by just over 3%, while winning nationally by 2.4%. If one looks at the state vote versus the national vote, Florida was a slightly stronger than average Republican state, Ohio just about average, and Michigan a much stronger Democratic state than the national average. If Obama is ahead by 4% at the moment nationally, then in Michigan he is running a bit weaker than his national number (whereas Kerry ran 6% better), in Florida he is running as much as 9-10% weaker than his national number (Kerry 2% weaker), and in Ohio, about the same as his national lead.
Pollsters and their samples
Scott Rasmussen has noted that there has been much greater volatility in his poll numbers this year than occurred in the Bush Kerry race in 2004. The total swing from the biggest daily McCain lead to the biggest daily Obama lead has been 17%. With no incumbent President or Vice President running for the first time in 80 years*, voters perceptions on the two candidates, and in particular the less well known Obama, have moved up and down. Polling is also more difficult this year for a number of reasons.
In the primaries, Barack Obama tended to over-poll in primaries -- his actual results were not as strong for him as final polls in various states (when this occurs with African American candidates, it is often called the Bradley effect, for the former Los Angeles Mayor Tom Bradley who was defeated in his race for Governor of California, though he had led in all the polls.). There were exceptions, especially Indiana and North Carolina, where Obama did better than the final polls in these states, but there appears to be some evidence this year of voters telling the pollsters what they think they want to hear, or put another way, providing the more politically correct response ("I am for Obama, and therefore I am not a racist").
Given the weakness of the Republican Party this cycle, pollsters need to gauge the right mix of Republicans, Democrats, and independents in their sample, either nationally or in a state, or miss badly. Independents tend to be closer to 50-50 in their profile, but both Democrats and Republicans tend to be very supportive of their Party's nominee (as much as 85-15 splits). If Republicans are under-sampled by 10%, and Democrats over-sampled by 10% in the same survey, it could produce a 14% swing in the overall poll result. The recent national poll results that seem to have missed the mark by a wide margin: Newsweek, and the Los Angeles Times/Bloomberg survey, both had a very heavy mix of self identified Democrats and many fewer self-identified Republicans. Of course, Newsweek has been wrong so often and by so much that new explanations are needed with virtually every one of their surveys.
Rasmussen has had a very stable mix of about 9-10% more Democrats than Republicans all year. So the changes in his results over time reflect how those identified with each party have changed their preferences over time.
After Clinton dropped out of the race, some Democrats "came home", and party loyalty numbers increased. Finally, there may be many new voters this year, as there were in 2004, when turnout grew by over 20 million compared to 2000.Many of the new voters this year are likely to be under age 30 or African American. Both groups favor Obama by large margins, and in each case, it will be difficult to estimate turnout. With the under age 30 group, many of whom do not have land line phones, survey results may be inaccurate. It is hard to know whether the possible "over-polling" by Obama (the Bradley effect) will be matched or overwhelmed by the possible underestimation and under-sampling of new Obama voters turning out. Again, Rasmussen, has tried to gauge the shift by increasing the share of African Americans in his surveys in many states, 36% for instance in Mississippi.
After Clinton dropped out of the race, some Democrats "came home", and party loyalty numbers increased. Finally, there may be many new voters this year, as there were in 2004, when turnout grew by over 20 million compared to 2000.Many of the new voters this year are likely to be under age 30 or African American. Both groups favor Obama by large margins, and in each case, it will be difficult to estimate turnout. With the under age 30 group, many of whom do not have land line phones, survey results may be inaccurate. It is hard to know whether the possible "over-polling" by Obama (the Bradley effect) will be matched or overwhelmed by the possible underestimation and under-sampling of new Obama voters turning out. Again, Rasmussen, has tried to gauge the shift by increasing the share of African Americans in his surveys in many states, 36% for instance in Mississippi.
How McCain could win
With all this by way of introduction, how does the underdog McCain pull off an upset? If one starts with the 31 states won by George Bush in 2004, totaling 286 Electoral College votes, three of these red states, totaling 21 Electoral College votes, seem to be leaning toward Obama at the moment. From strongest to weakest of the three for Obama, they are Iowa (7), New Mexico (5) and Colorado (9). In each state, Obama is ahead in every recent survey, though Colorado is very close in the most recent survey (2% Obama lead). If no other state shifted , these three would be enough to elect Obama. Obama is ahead in most surveys in Ohio (21) and Virginia (13), and in one recent poll in Indiana (11), and Montana (3). The Virginia results and the Indiana survey are only 1-2 point leads for Obama, in reality a tie. In Ohio, Rasmussen shows McCain ahead by 1, while other surveys give Obama a larger lead. Ohio was one of the states , Pennsylvania another, where Obama badly underperformed the final poll results in the Democratic primary, losing each state by 10% ,when the final poll average was only a 5% Clinton victory. Rasmussen has Obama up 5 in a recent Montana poll, perhaps explaining the Obama visit to the traditionally Republican state (at least in Presidential elections) on July 4th. The Obama team is also targeting other red states: Missouri (11), North Carolina (15), Georgia (15), Florida (27), North Dakota (3), Alaska (3), and Nevada (5). So far, McCain leads in all of these states that have had recent polls, though several are close.
McCain for his part is targeting 4 Kerry (blue) states: Pennsylvania (21), Michigan (17), New Hampshire (4),and Wisconsin (10). Michigan seems the best prospect of the group, then Pennsylvania. It is unclear whether any major effort will be made in New Jersey (15), Minnesota (10), or Oregon (7). If McCain were to win any of these three states, they will be gravy on top of a solid national win (north of 300 Electoral College votes) . Similarly, Obama does not need to win Georgia or Indiana or North Carolina to win, and if he wins these states, it probably means he will win big (325-400 Electoral College votes).
Put another way, if McCain can win in Michigan, then he starts at 303 Electoral College votes, before deducting for states Obama will win. If McCain can win Michigan and hold Ohio, Virginia and Indiana, that is likely his best shot at victory. Then he could withstand the loss of New Mexico, Colorado, and Iowa. He could lose another few small states (Nevada, Montana) and still win. If McCain wins Virginia, he will likely also win North Carolina and the rest of the South. If he wins Indiana, he will likely also win Missouri. If he wins Ohio, Florida is safe. In essence, all of this is based on rank ordering the states from most favorable for McCain to least favorable (Nate Silver has a quantitative model for a tipping point methodology, a similar concept).
Many paths to victory for Obama
The reason Obama is the favorite is that he has many more paths to victory, since he is targeting many more red states, than McCain has blue states to target. A national lead of 4-5% for Obama moves the playing field about 7% overall from where it was in 2004 when Bush won by 2.4%. If the playing field moved 7% the other way, McCain would be ahead by nearly 10% nationally, and he would be targeting many more blue states and Obama would have far fewer targets among the red states. Put simply, no candidate who loses the popular vote by 10% will even come close in the Electoral College. It is also highly unlikely that a candidate who loses by 5% in the national popular vote can win an Electoral College majority.
But a smaller deficit -- 3% or less -- could leave the door open to a repeat of the 2000 result. Had 60,000 voters shifted from Bush to Kerry in Ohio, Kerry would have won the Electoral College in 2004 and lost the popular vote by 3 million (over 2% margin). My own view is that such a split result is far more likely to favor McCain than Obama this year.
Obama will have lots of states with big margin wins (wasted votes in other words) -- New England, New York, Maryland, DC, California, Illinois -- and will suffer narrower losses in many Southern states than was the case for John Kerry, due to heavy turnout of African American voters. If McCain can win enough of the tossup states, most likely by a small margin, he can win. However, he will have to close the national popular vote gap from the current 4-5% level to have a chance, or alternatively, the national poll numbers would need to overstate Obama's lead by a few points. Both of these are possible, but no national campaign can count on the latter being the case.
Tomorrow, I will explore what McCain can do on his part to win, and where Obama may be vulnerable.
*Thanks to John Zimmerman, who pointed out that in 1952, Harry Truman stuck around for a while as possible Democratic candidate (this was before 22nd amendment) , before dropping out. So 1952 was last time neither party had a President or VP on general election ticket, but 1928 was last time neither President nor VP made any effort to even run..
Richard Baehr is chief political correspondent of American Thinker.