Republicans' Biggest Alarm Bell

Republicans have not gotten a lot of great news in recent weeks. Beyond Senator McCain's considerable problems, and the long-standing political reality that Republicans will sustain robust losses in both the House and the Senate, things have actually gotten worse in all three battlefields in the last week, particularly on the U.S. Senate map. The Senate picture right now looks awful for Republicans. Republican-held seats in Virginia and New Mexico are surely gone, with Colorado, New Hampshire, and North Carolina approaching that territory. Alaska will be lost if Ted Stevens is convicted, and the Republican incumbents seeking re-election in Minnesota and Oregon are down at this moment. Mississippi now looks 50-50 for the GOP to maintain a hold, and even Kentucky is neck-and-neck, with Mitch McConnell fighting for political life.

However, not all of this should have been totally unexpected. Each of these races always looked potentially bad if things broke just right for the Democrats. To be sure, few expected a year ago that North Carolina would be where it is now -- with Senator Elizabeth Dole down several points to a formally-unknown state senator -- but the potential was there. Ditto Mississippi, Minnesota, Oregon, and Alaska.

Yet, one state which was on absolutely no one's map of potentially close races was Georgia. To be totally fair, no savvy politico made any arguments that Georgia, where first-term Senator Saxby Chambliss is seeking another term, could possibly be close.

This made complete sense. Chambliss has been a good fundraiser and Democrats had totally failed in recruiting a top-flight, or even second-tier challenger to face off against the Senator; though, to be fair, the Democratic bench is not terribly deep at all to begin given Georgia's rightward movement. Indeed, since Chambliss ousted Democratic Senator Max Cleland in an famously acrimonious 2002 race by 53-to-46, Georgia has lurched almost completely to the right. Where Georgia had a Democratic governor and two Dem senators in 2002, the state now has a popular two-term Republican in the governor's mansion, and two uber conservatives in the U.S. Senate. John Kerry lost the Peach State very badly, by almost 18 percentage points.

In short, as Georgia had finally become a deep red state, Democrats did not really try to fight Chambliss. Many Democrats desperately wanted to take out Chambliss for his attacks on Cleland, but they were completely frustrated given the state's new dynamics and the lack of any bench. Until this month, Chambliss appeared headed to a very easy re-election. With all due respect to the eventual nominee, former State Rep. Jim Martin, he was not a top-notch candidate, and he was given zero chance to oust Chambliss. Indeed, he lost the first round of the primary before winning an easy run-off over a candidate who would have been even more of a sacrifical lamb against Chambliss.

In just two short weeks, all of this has gone out the window. In an astonishing turn of events, the Georgia senate race is now tied. And this is not a conclusion gleaned from one single poll: six recent polls have found nearly the same results in this contest. This is so incredible, we need to take a look at each of them.

Poll #1: Survey USA (poll conducted 9/28 and 9/29):

Chamblis 46
Martin 44

McCain 52
Obama 44

The poll had a lot of worthwhile data in the crosstabs. Fascinatingly, it had Martin ahead 47-45 (+2 points) among men, but Chambliss leading with women 48-42 (+6) (two findsings, quite honestly, I don't buy). Just about all of the age groups were even-steven: 47-47 among those aged 18 to 34, 44-43 (+1) Martin with the 35-49 set, and Chambliss edging Martin 46-45 (+1) with with the 50-64 group. Chambliss got his overall lead by winning 65+ people by 53-39 (+14).

Perhaps the most interesting number was the following: among the small subset that had already voted (Georgia has early voting), Martin was ahead 61-36 (+25), while only down 47-43 among likely voters. Yes, this was a small part of the sample (a mere nine percent), but given that the polls tightened suddenly in the midst of the economic crisis, this seems to show that clearly Martin has the momentum in the contest, at least at time this poll was conducted two weeks ago, and the people going to the polls right now are breaking heavily for him. This is bad news for Chambliss, but at least it allows him time to regain his footing before November 4.

In terms of the presidential breakdown, needless to say that McCain's high single digit lead is not a good sign for the Republican, who should be blowing out Obama in the state.

In a poll taken two weeks earlier by SUSA, Chambliss had led 53-36 (+17 Chambliss).

Poll #2: Research 2000 for Daily Kos (9/29, 9/30, and 10/1):

Chambliss 45
Martin 44

McCain 50
Obama 43

Daily Kos commissioned this poll right after the SUSA numbers were released in order to see if they were an outlier. The results are almost a mirror image of SUSA's findings. For the most part, the poll had an identical composition of respondents, though R2K's poll was 27 percent black, whereas SUSA was 25 percent; the state's overall black population is around 29 percent currently, so one could argue that both slightly under-sampled that segment of the population a tiny bit.

Daily Kos, as usual, provided richly-interesting cross tabs for us to dissect. First, it found Chambliss' favorable-unfavorable split at 54/42 (+12), and Martin's at 51/34 (+17). Martin thus has near-equal favorables as the incumbent, but his unfavorable is perhaps not surprisingly considerably lower because he more unknown in the state (though, he was the Democrats' unsuccessful nominee for lieutenant governor in 2006). Party splits kind of canceled each other out: Chambliss was at 71/24 (+47) among Republicans, while Martin scored 75/15 (+60) among Democrats. Martin might be enjoying a bit more strong support from his base at this time. However, among key independents, Martin was at 53/29 or +14, a solid score, while Chambliss' favorables among these key group rounded out at 51/45, or +6.

In terms of the head-to-head numbers, Chambliss led 48-41 (+7) among men, while Martin won women 47-42 (+5). Dems break 78-13 (+65) to Martin, Republicans 77-11 (+66) to the incumbent, and indies go 46-42 (+4) to the challenger. So, according to R2K, Chambliss is doing about equal among his base with Martin, but Martin has a decent edge with independents/unaffiliated voters. Compared with SUSA, I think these numbers are probably more accurate: it makes perfect sense that Martin would be winning women while losing men. The key number here is obviously indies, which Martin has to win by a solid amount in order to pull off the upset.

Not surprisingly, the poll finds whites breaking to the Republican 63-26 (+37), but Martin wins blacks 83-8, and Hispanics, who make up a small but perhaps crucial five percent of the electorate, by a wide 69-11 (+58). Here we see where Martin can most easily find his path to victory: on Obama's coattails. The higher the black turnout, the more votes Martin gets. That's a time-and-tested formula for Democrats, and it more certainly applies here. Martin will almost definitely clear 90 percent of this group, and while Hispanics are a small part of the state, his blow-out lead with them could prove pivotal.

Finally are the age groups, which are similarly tight here: Among 18-29 year olds, Martin leads 48-42 (+6) (which is likely how it really is, so SUSA is probably wrong here), it's tied 44-44 among those 30-44, Chambliss leads 46-43 (+3) in the 45-to-59 bracket, and with people over 60, the incumbent has the edge by 47-42 (+5).

While the ultimate findings are basically the same as SUSA, I think these numbers are more logical, particularly among age groups and the among the sexes.

Poll #3: Strategic Vision (10/5, 10/6, and 10/7):

Chambliss 47
Martin 44

McCain 50
Obama 43

While there are no cross tabs to look at, the point is that the numbers are the same. For the third straight poll, Martin is at 44 percent, and Chambliss is below 50 percent. The poll does note that Chambliss' favorable/unfavorable split is a mediocre 49/41 (+8). If true, this would most likely be the product of voter unrest with the economy.

Allow me to point out that this is a Republican polling outfit, so the fact that they found the race so tight is quite telling. Strategic Vision is not a bad outfit, but unsurprisingly many of their polls often have a Republican lean.

In their poll one month earlier, Strategic Vision had the race at 54-36 (+18 Chambliss). Therefore, they confirm that fast tightening of this contest since mid-September.

Poll #4: Rasmussen (10/7):

Chambliss 50
Martin 44

McCain 54
Obama 45

Again, Martin is at 44 percent, but Rasmussen puts Chambliss at an even 50 percent, several points above every other recent poll. This may be an outlier, and I am guessing that it is for reasons I will outline.

Rasmussen finds the incumbent's personal favorable split at 57/37 (+20), and Martin's at 45/38 (+7). There is no way Chambliss is at +20 in his personal score, and Martin, who is much more unknown in the state up to this point is a mere +7. It does not make sense, and runs counter to common political sense. Yes, I acknowledge that Martin did run for LG in 2006, but I do not think that makes him a fraction as known or as visible as incumbent U.S. Senator Chambliss. In this author's amateur opinion, Chambliss does not have a 57 percent approval score right now. That runs counter to every other poll and campaign anecdote.

It's interesting to note that the last Rasmussen poll of the race, conducted on September 16, found the race at 50-43. Thus, this poll finds basically zero change between September and October. This too undermines the complete credibility of this poll. While I do not dispute the possibility that Martin's support has remained unchanged, it is difficult to believe that in the middle of the economic crisis, Chambliss enjoys an identical level of support as one month ago. No chance. I think Rasmussen is one of the very best in the business, but this is an outlier poll..

Also note that Rasmussen finds higher support for McCain than the others. Maybe that's true. After all, this is Georgia we are talking about. But I don't think so, at least now, so I kind of dismiss this poll a bit.

Poll #5: Insider Advantage (10/9):

Chambliss 45
Martin 45

McCain 49
Obama 46

This is the freshest poll, coming out on Friday. This poll provides absolutely no other data, but that's okay. We see basically mirror results as those in every other recent poll.

Of the five polls, four have Martin at 44 percent support, and one has him at 45 percent. That's pretty darn stable. Chambliss is at 46, 45, 47, 50, and 45, for an average of 46.6. Subtract the Rasmussen outlier, and his average dips slightly to 45.75.

What's the conclusion of all of this? Duh, the race, at least over the last two weeks, is just about totally tied up, a stunning development.

Clearly, the economic collapse has had an enormous impact on this particular contest. Both Survey USA and Strategic Vision found the contest closing a whooping 15 points between September and October, with again nearly identical numbers in the process.

These numbers are so shocking, again, because no on earth minus Jim Martin and his wife (if even them) ever thought this race was winnable. Martin was off of every person's map, including this author's. That he could come up like this to the point that he is tied up with an incumbent in Georgia in the second week of October is nothing short of a political miracle.

Georgia/Kentucky parallels

I would go as far as to say that Martin's current strength is even more impressive than Bruce Lunsford's close position with Mitch McConnell in Kentucky. Like in Georgia, the latest batch of recent polls out of the Bluegrass State has found the race basically tied, after McConnell too had led by wide margins throughout the summer. Absolutely, this is also a stunning result this late in the game, and I think it is pretty amazing that Lunsford has been able to keep it at all close. The difference with Georgia, I think, is that Martin was a much more unheralded candidate than Lunsford. While Lunsford has lost several past attempts at statewide office, he is still very well known in Kentucky, so he started with a greater advantage than Martin. Furthermore, whereas Lunsford is wealthy and has the ability to self-fund his campaign, Martin is not rich, and has not had the luxary of dipping into his pockets to match his well-funded opponent. As a result, while I think we are talking in fine degrees, I find Martin's resurgence slightly more impressive than Lunsford's. Indeed, Kentucky has always been on peoples' maps, if for no other reason than Democrats have been eager (albeit mostly frustrated) to knock of the GOP Senate leader in revenge for Tom Daschle's 2004 defeat. Conversely, while Dems wanted revenge on Chambliss for his defeat of Max Cleland in 2002, no one felt that event was possible even in their wildest dreams.

No measured Obama coattails. Yet.

Anyway, going back to the Georiga race, it is important to note that these poll numbers do not reflect an Obama coattail-impact, which itself is likely to boost Martin both in early voting and on election day. As we noted, at least several of these polls included samples of black voters around equal to the state's approximate 27 percnet black population. In other words, these polls' results are based on the assumption that black turnout will approximately equal the percentage of black voters in Georgia.

I think that this is highly unlikely, given both anecdotal evidence and a general common belief. In the 2004 presidential election, black voters made up 25 percent of voters according to CNN's exit poll. This is slightly under the state's overall population breakdown. It would be stunning if we do not see a big a rise among black voters next month. An item on the Atlanta Journal Constitution web page on October 1 had a nugget of information that speaks to this belief:

As of Monday, 135,412 ballots had been cast in Georgia for the Nov. 4 general election, whether by absentee or early in-person voting.

Nearly 40 percent of those voters — 53,160 — have been African-American, according to the office of Secretary of State Karen Handel.

It is this kind of intensity, driven by the presidential campaign of Barack Obama, that has worried Georgia Republicans and stirred the hopes of Democrats in the state.

As of Aug. 31, African-Americans, the most reliable demographic in the Democratic base, made up 29 percent of those registered to vote. However, their participation rate in elections traditionally is several points lower.

After a full week of early voting statewide, the highest performing counties are: DeKalb, with 14,560 votes cast; Fulton, 10,599; Gwinnett, 7,952; Cobb, 7,021; Chatham, 4,771. All have significant minority populations.

This finding helps support SUSA's number which indicates that Martin is winning early voters by a wide margin, at least at the last of September. That 40 percent of early voters are black is a great sign for not just Barack Obama, but also Jim Martin. While I do not expect the final results to show that 40 percent of the voting electorate was black, it bodes well for a turnout much higher than 27 or 29 percent come November 4.

As we noted above, at least SUSA and R2k simply undersampled black voters. This, by itself, should favor Jim Martin a bit, and it does not even take into account possible super-turnout numbers like those that we're seeing in early voting.

The good news for Democrats. If the poll numbers are accurate -- and their near-total agreements seems to indicate that that is the case -- Martin is positioned to win as things are. People in Georgia are very angry about what has happened over the last few weeks, and as an incumbent tied to President Bush's party, Chambliss is feeling the brunt of that anger. While it usually hurts any challenger to not be as known as the incumbent, and this race is no exception given Martin's lack of funds in this race, that dyanamic may end up helping Martin here for the simple reason that voters are an anti-incumbent mood. As a result, Martin is now well-positioned to capitalize because he is not known enough to have high personal negatives.

Perhaps even more importantly, if Obama is able to not only drive out black turnout, but also actually remain close in the presidential contest, Martin should enjoy enormously potent coattails. I do not expect Obama to carry Georgia -- who does? -- even given his national surge over the past few weeks. But if the race is now tied without including either of these factors in the polling, then Martin will win. Indeed, a friend who knows people working for Obama on the ground in the southeast told me that the Obama campaign is dedicating resources to Georgia in order to benefit Jim Martin, knowing full well that Obama probably won't win the state himself. The Obama campaign clearly sees that Martin has a real shot now, and his dedication of resources to the race is certainly helpful to the challenger.

The bad news for Democrats. There are still 24 days left, which is plenty of time for Chambliss to steady himself. Things look bad for Republicans across the country right now, but Chambliss has a lot of time, and presumably plenty of money to blast Martin for his very liberal positions and try to drive up his opponent's negatives enough to make him unpalatable for the conservative Georgia electorate. This might not hurt the challenger among black and Hispanic voters, but it could deny him the slice of the white vote he needs to prevail. Republicans are at their low point right now, and if the economic can fade from peoples' attention even a litle bit, given Chambliss' decent personal ratings, he should be able to survive.

The bottom line. It's pretty simple: if black turnout hits record levels, and Chambliss is not able to personally rebound or better define Martin's liberal positions, Jim Martin will pull off likely the biggest upset of a cycle that is likely to be filled with upsets. The economy remaining on the front burner right now -- in place of a discussion of social issues which would be damaging to Martin -- would also be appreciably helpful to the challenger.

My personal guess is that Chambliss will squeak through in the end. Martin is just too liberal and too unknown to win a statewide race, and in Georgia of all places. That Martin is so close now is itself unbelieveable, and speaks well to the anger of the electorate, but I do not think he pulls off the impossible in the end. But hey, maybe Obama ends up losing by two points in Georgia, in which case Martin will likely be carried over the finish line. We'll soon see.