Hillary's Latest Electability Argument Is Based On the Clinton "New Math"
Clinton strategist Howard Wolfson held a conference call yesterday and was quoted in the Washington Post as saying that “current Electoral College vote predictions -- based on an aggregate of public polling -- show Clinton beating McCain and Barack Obama losing to the Arizona senator.” The Clinton team is now using these predictions to try to persuade superdelegates to support her bid for the nomination.
From what I can gather, the Clinton campaign seems to have come up with a new metric for measuring success in the primary. First, you use the results of polls taken in each state for the two possible November match-ups, Clinton vs. McCain and Obama vs. McCain. You then project the winner-take-all electoral college votes that each would receive assuming these polls represented the actual results in November. The overall electoral vote “winner” is thereby argued to be more electable.
Well, this sounded questionable to me, but I decided to look at the polling data. This was easy to do since Greg Sargent provided a link at TPM Election Central to the documents presented by the Clinton campaign to the superdelegates. After looking at the polling data presented by the Clinton campaign, it appears to me that the Clinton analysis is on shaky ground.
First of all, the polls included in the Clinton data were taken by 15 or so different pollsters over a several month period. We have already seen during the primaries that some pollsters were better than others at accurately predicting the actual election results. It is therefore valid to question whether the polls selected by Clinton are reliable and whether the methodologies are consistent. For example, I noticed that Clinton used polling data from May 12-20 that show McCain leading Obama in both Virgina (+8%) and Ohio (+4%). However, recent Survey USA polls from May 18 show Obama to be ahead of McCain in Virginia (+7%) and Ohio (+9%). These two states alone account for 33 electoral votes so the disagreement in the Clinton polls and Survey USA polls can result in as much as a 66-vote swing in the match-up between Obama and McCain. Which polls are accurate? What accounts for the significant differences? We cannot know, and that is the problem with the Clinton methodology. More significantly, it demonstrates that drastically different electoral vote totals can be obtained depending on which polls are used.
It is also impossible to project electoral votes so far in advance in the swing states where the polling results are statistically tied or fluctuating back and forth. For example, Michigan and Missouri are too close to call (+/- 2-3%) for either Clinton or Obama in the Clinton polling data so how does one hypothetically assign their 28 electoral votes? Well, we see that Hillary errs on the side of Hillary. In her count, she is awarded both Michigan and Missouri even though she is tied with McCain in the Michigan poll and ahead by just 2% in the Missouri poll. Obama, on the other hand, is denied both states because of small 1% and 3% deficits, respectively. It is difficult to understand how the superdelegates are supposed to accept that Obama would lose Michigan whereas Hillary would win it based on a 1% difference in their polling match up with McCain in May. More likely, such polls would suggest that Clinton and Obama have statistically equal chance of winning or losing the state. In fact, these swing states have the potential to go either way up until the election depending on how the candidates perform and how well their messages are received.
As I was looking at Clinton’s polling data, I was also struck by the fact that many of Clinton’s loudest arguments made during the primaries could be directly refuted by her own data. For example, the Clinton campaign made a huge deal about winning the Texas primary and, in fact, the fate of Hillary’s campaign had even been staked to it (by Bill). Yet, the polling data shows her to be even further behind than Obama in a match up with McCain (-15% and –13%, respectively). The Clinton campaign also argued that Hillary’s double-digit victory in Pennsylvania was reason enough to prove her electability over Obama, and yet the polling data shows that both she and Obama can beat McCain in Pennsylvania. And in Indiana, where Hillary’s marginal victory “broke the tie”, the polling data shows that both she and Obama are significantly behind McCain (-11% and –8%, respectively). It just shows that the Clinton campaign uses election results and polling data in one way when it serves their interests and in another way when it doesn’t.
But this is all beside the point. Polling data DOES NOT MATTER in determining the Democratic nominee. There is a process by which the Democratic nominee is selected and it is based on delegate count. There are the pledged delegates from the primaries and caucuses and then there are the superdelegates that can be courted. The pledged delegate counts are the most representative of the people’s will since they reflect the votes that were actually cast in the elections. Yet the argument that the Clinton campaign seems to be making to the superdelegates is that (selective) polls from each state should somehow override the actual election results.
I could almost understand their logic if they were using recent polls to show that Democratic voters in certain states no longer favored Obama even though their primary or caucus results gave him the victory. Unfortunately, they aren’t using logic but, rather, Rove-inspired manipulation. They are basing their argument on polls that pit the two Democratic candidates against a candidate, McCain, who they have not yet campaigned against. The American people have not been able to hear the Democratic candidates debate McCain and contrast their positions and approaches. In fact, most Americans know very little about McCain at this point except that he is a “maverick”. So while more than 20 Democratic debates were held to help the Democratic voters determine their nominee, Clinton would have the superdelegates believe that polls taken in March, April, and May during a heated and competitive Democratic primary accurately reflect the general election results that will be produced in November. This is ridiculous, especially when such polls were taken with both Democratic candidates still actively campaigning, thereby making it very likely that their supporters’ allegiances would influence the polling numbers.
Such projected electoral vote predictions are especially unrealistic since they assume a static voter sentiment between now and November. Even after the nominees are chosen, the polls do not necessarily reflect the final election results. Al Gore was behind, 55%-39%, to Bush in the Aug. 11-12 Gallup poll leading up to Democratic convention, but he actually ended up winning the popular vote. And in the 2006 Virginia senate race, George Allen enjoyed double-digit leads over Jim Webb and was so confident that he was even considering a 2008 presidential bid…but we all know about his infamous “Macaca” remark in August. The momentum of a political race can change in either direction during a campaign, but especially as the election date draws near and the intensity increases. And from what we’ve heard, McCain is a loose cannon!
At some point, though, we have to look at the bigger picture. How do we rationalize the destruction of our democratic process, whereby suspect polls taken more than five months before the general election are used to override the delegate counts obtained in sanctioned state primaries and caucuses? How do we even know that these polls reflect the true intent of Democratic and Republican voters? There is absolutely no way to corroborate that someone will actually vote as they claim in the polls, especially given the rogue Limbaugh followers who have been creating mischief throughout the primaries. If such polls were truly to be used to determine our nominee, they should be verifiable in some way to rule out rigging or fraud. But how is that possible? While Clinton supporters may feel at this point that the ends justifies the means, I wish they would think about the long-term implications not only for the Democratic party but also for the country. After our recent stolen elections, do we really want to legitimize a non-transparent process that would make it all the easier for corrupt parties to “select” a president?
From what I can gather, the Clinton campaign seems to have come up with a new metric for measuring success in the primary. First, you use the results of polls taken in each state for the two possible November match-ups, Clinton vs. McCain and Obama vs. McCain. You then project the winner-take-all electoral college votes that each would receive assuming these polls represented the actual results in November. The overall electoral vote “winner” is thereby argued to be more electable.
Well, this sounded questionable to me, but I decided to look at the polling data. This was easy to do since Greg Sargent provided a link at TPM Election Central to the documents presented by the Clinton campaign to the superdelegates. After looking at the polling data presented by the Clinton campaign, it appears to me that the Clinton analysis is on shaky ground.
First of all, the polls included in the Clinton data were taken by 15 or so different pollsters over a several month period. We have already seen during the primaries that some pollsters were better than others at accurately predicting the actual election results. It is therefore valid to question whether the polls selected by Clinton are reliable and whether the methodologies are consistent. For example, I noticed that Clinton used polling data from May 12-20 that show McCain leading Obama in both Virgina (+8%) and Ohio (+4%). However, recent Survey USA polls from May 18 show Obama to be ahead of McCain in Virginia (+7%) and Ohio (+9%). These two states alone account for 33 electoral votes so the disagreement in the Clinton polls and Survey USA polls can result in as much as a 66-vote swing in the match-up between Obama and McCain. Which polls are accurate? What accounts for the significant differences? We cannot know, and that is the problem with the Clinton methodology. More significantly, it demonstrates that drastically different electoral vote totals can be obtained depending on which polls are used.
It is also impossible to project electoral votes so far in advance in the swing states where the polling results are statistically tied or fluctuating back and forth. For example, Michigan and Missouri are too close to call (+/- 2-3%) for either Clinton or Obama in the Clinton polling data so how does one hypothetically assign their 28 electoral votes? Well, we see that Hillary errs on the side of Hillary. In her count, she is awarded both Michigan and Missouri even though she is tied with McCain in the Michigan poll and ahead by just 2% in the Missouri poll. Obama, on the other hand, is denied both states because of small 1% and 3% deficits, respectively. It is difficult to understand how the superdelegates are supposed to accept that Obama would lose Michigan whereas Hillary would win it based on a 1% difference in their polling match up with McCain in May. More likely, such polls would suggest that Clinton and Obama have statistically equal chance of winning or losing the state. In fact, these swing states have the potential to go either way up until the election depending on how the candidates perform and how well their messages are received.
As I was looking at Clinton’s polling data, I was also struck by the fact that many of Clinton’s loudest arguments made during the primaries could be directly refuted by her own data. For example, the Clinton campaign made a huge deal about winning the Texas primary and, in fact, the fate of Hillary’s campaign had even been staked to it (by Bill). Yet, the polling data shows her to be even further behind than Obama in a match up with McCain (-15% and –13%, respectively). The Clinton campaign also argued that Hillary’s double-digit victory in Pennsylvania was reason enough to prove her electability over Obama, and yet the polling data shows that both she and Obama can beat McCain in Pennsylvania. And in Indiana, where Hillary’s marginal victory “broke the tie”, the polling data shows that both she and Obama are significantly behind McCain (-11% and –8%, respectively). It just shows that the Clinton campaign uses election results and polling data in one way when it serves their interests and in another way when it doesn’t.
But this is all beside the point. Polling data DOES NOT MATTER in determining the Democratic nominee. There is a process by which the Democratic nominee is selected and it is based on delegate count. There are the pledged delegates from the primaries and caucuses and then there are the superdelegates that can be courted. The pledged delegate counts are the most representative of the people’s will since they reflect the votes that were actually cast in the elections. Yet the argument that the Clinton campaign seems to be making to the superdelegates is that (selective) polls from each state should somehow override the actual election results.
I could almost understand their logic if they were using recent polls to show that Democratic voters in certain states no longer favored Obama even though their primary or caucus results gave him the victory. Unfortunately, they aren’t using logic but, rather, Rove-inspired manipulation. They are basing their argument on polls that pit the two Democratic candidates against a candidate, McCain, who they have not yet campaigned against. The American people have not been able to hear the Democratic candidates debate McCain and contrast their positions and approaches. In fact, most Americans know very little about McCain at this point except that he is a “maverick”. So while more than 20 Democratic debates were held to help the Democratic voters determine their nominee, Clinton would have the superdelegates believe that polls taken in March, April, and May during a heated and competitive Democratic primary accurately reflect the general election results that will be produced in November. This is ridiculous, especially when such polls were taken with both Democratic candidates still actively campaigning, thereby making it very likely that their supporters’ allegiances would influence the polling numbers.
Such projected electoral vote predictions are especially unrealistic since they assume a static voter sentiment between now and November. Even after the nominees are chosen, the polls do not necessarily reflect the final election results. Al Gore was behind, 55%-39%, to Bush in the Aug. 11-12 Gallup poll leading up to Democratic convention, but he actually ended up winning the popular vote. And in the 2006 Virginia senate race, George Allen enjoyed double-digit leads over Jim Webb and was so confident that he was even considering a 2008 presidential bid…but we all know about his infamous “Macaca” remark in August. The momentum of a political race can change in either direction during a campaign, but especially as the election date draws near and the intensity increases. And from what we’ve heard, McCain is a loose cannon!
At some point, though, we have to look at the bigger picture. How do we rationalize the destruction of our democratic process, whereby suspect polls taken more than five months before the general election are used to override the delegate counts obtained in sanctioned state primaries and caucuses? How do we even know that these polls reflect the true intent of Democratic and Republican voters? There is absolutely no way to corroborate that someone will actually vote as they claim in the polls, especially given the rogue Limbaugh followers who have been creating mischief throughout the primaries. If such polls were truly to be used to determine our nominee, they should be verifiable in some way to rule out rigging or fraud. But how is that possible? While Clinton supporters may feel at this point that the ends justifies the means, I wish they would think about the long-term implications not only for the Democratic party but also for the country. After our recent stolen elections, do we really want to legitimize a non-transparent process that would make it all the easier for corrupt parties to “select” a president?




