The readers at fivethirtyeight.com don't need to be told that the election results from 2008 matched Nate's pre-election poll aggregations in almost every state. But it's also interesting to look at some of the discrepancies:
The biggest discrepancy is in Hawaii, which Obama visited a few days before the election to go to his grandmother's funeral. Then there's Nevada--I don't know what was going on there, but probably Nate does. Beyond that, there's a systematic pattern that Obama did better than the polls in Deomocratic states and worse than the polls in Republican states. Does this represent a real pattern of voters--perhaps people reverting to their more predictable positions at the last minute, with Vermonters moving to the Democrats and residents of Wyoming going the other way? Or maybe it's an artifact of the poll aggregation, with the predictions being pushed too close to 50%, on average? I don't know. It would be interesting to do similar aggregations of state polls from 2000 and 2004, and also to look at national polls, to see if this pattern was occurring in earlier years.
The discrepancies had very little impact on the election forecast because most of the problems were in landslide states. If the poll aggregation predicted Obama would get 60% in Delaware and he actually got 62%, it doesn't really matter anyway. But it could be interesting to study such systematic discrepancies with the goal of improving the methods for the future.
50 comments
It looks like racism played and continues to play a huge factor in the vote tally discrepancies, and how Obama is viewed today. Race would be one of the least relevant factors in an election for Hawaii and Vermont voters, and one of if not the most important factors for some of the states that voted below the bar and well to the left on this graph.
I think the "Bradley Effect" was discounted and laid to rest on a national basis, but something similar to a "Bradley Effect" may have played out in Oklahoma.
How much polling was done in some of those outlier states? Obviously there was plenty from NV. But rest of the outliers are what you described as "landslide states" which presumably didn't have a wealth of polling data.
Maybe people in landslide states are just more likely to show up to the polls if they know they are on the winning side.
A landslide state won't have as high a turnout as a battleground. Not only does it feel better to vote for a winner, but you will likely have more friends encouraging you to vote for the winner.
I'm sure there are other possible reasons people can come up with as well.
I had the same pattern with my predictions--most of the close states I came close to nailing (I also missed Nevada by quite a bit, it wasn't actually close in the end) but was far off in states where the margin was expected to be large but turned out to be larger.
At some point someone, maybe Nate, showed a correlation between Obama's results and whether or not he ever visited the state, even if it was only during the primary campaign.
Polls exist to create news. It's not news if a blowout is going to be a bigger blowout. If I were trying to perfect my polling techniques, I might sacrifice some precision at the extremes in order to focus on the sweet-spot around 50/50.
I just love that Hawaii appears to be saying "HI" from above the graph.
"Deomocratic"
FFS. I don't want to harp, but please do at least a simple spellcheck...
Mark, there is no Bradley effect. Every example of the so-called Bradley effect has been easily explained by other factors...usually improper polling. There is no reason enough people would lie in a survey and say they'll vote Obama and not in the booth. McCain had plenty of pluses to him that racists could have supported.
I think it will come down to polling issues. In the extremities, there wasn't enough. In the center (NV), the turnout models were off.
Could there perhaps be a slightly bigger version of this graph? I can't find some states, including my own state, Oregon, lost in the tangle.
Me on 12/14/08:
My quick check indicates that in "contested" states with significant polling Nate's projections were most off in McCain's favor in Nevada and New Mexico. There's nothing nearly as far off in Obama's favor among sufficiently polled "contested" states. Iowa is the furthest off in Obama's favor unless you think West Virginia, which had by far the biggest drop-off in voters (over 5%) from 2004 to 2008, was "contested." In Iowa, Selzer had a polling triumph in about as difficult to figure an environment as can be imagined in January and, ten month later, was the pollster furthest off regarding the margin of victory. For what it's worth, she was the closest in Michigan.
That's very interesting. It appears the real data approximates a polynomial curve of a cubic function much better than the linear trendline being used to show X=Y in this graph. Is there a third order function that could better be used to best predict this data?
Juvanya "There is no Bradley Effect".
I sort of agree. That's why I said "Something similar to a Bradley Effect."
Race places a huge role in how people vote in states like Alabama & Oklahoma. I think it may have played a role in some of these polling verses voting discrepancies.
Was it late appeals to "White Pride" in some areas? I received many such appeals via email here in the Deep South.
I might think that the discrepancy for Alaska is likely the same one that showed in the Senate race, where Ted Stephens came surprisingly close, more so than expected. I would agree with some of the speculation I've heard that one of the factors was that Alaska's polls were open well after McCain had delivered his concession speech, and even after Obama's victory speech; there was around a full hour after the latter, during which the Democratic voters, (for both the Presidential and Senate races) which historically are the most likely to stay home, are likely to have decided at that point that if they hadn't voted yet, there was no point anyway. Meanwhile, for their Republican counterparts, they were likely not as inclined to stay home, or possibly may have already voted enough anyway.
Unfortunately, I doubt that there was any exit polling done that factored in the time of the voting present in the state, so I have my doubts if we could really know. However, if such data doesn't exist already, at least a tendency could perhaps be found if exit pollers took efforts to kep track of what times people voted, and looked for a possible partisan shift there.
Of course, an alternative source for the discrepancy may simply have been due to Sarah Palin, upon the state's undecided voters, especially those who may have not kept up with the election; simply put, if you haven't been paying attention, but you ARE aware of your governor's celebrity status, that may contribute to your final decision.
Meanwhile, for the "outer south" states (like TN, KY, and AR) I would not be surprised if the "Bradley Effect" may have, in fact, appeared there. This would contrast from the "Deep South" (specifically MS, AL, GA, and SC) where this effect would certainly be as likely to show up, but would likely be counter-acted by the increase in African-American voter turnout, given that all four of those states are among the top six in African-American population, with between approximately 25.9-36.3%. (http://en.wikipedia.org/wiki/List_of_U.S._states_by_African-American_population)
Meanwhile, with Arkansas, that percentage is far lower, at around 15.6%. Louisianna also shows the same trend, in spite of being, according to Census 2000, ranked #2 for African-American population. However, I'm hypothesizing that Census 2000's numbers are no longer valid, due to the effects of Hurricane Katrina, which has substantially reduced the population of New Orleans, the state's highest concentration of African-Americans. I'd guess that perhaps its percentage has been reduced to be closer to that of Arkansas, which would be the reason for the similar shift; similar might be said for the other two states Obama under-performed Kerry, Tennessee and West Virginia.
Finally, I'd note that a lot of the polls in the more Republican states more closely seemed to reflect the 2004 results than the 2008 results. I'm not quite sure what to make of that myself, but it is rather curious; Utah was Bush's biggest margin in 2004, and wound up being polled as the highest margin for McCain. Similarly, excepting Alaska and the Outer South states, there seems to be a correlation between how well Bush did in those states in 2004, and how much polls under-estimated McCain's performance. That might be interesting to look into.
Turnout is certainly affected by knowing your state is going to be close or not and when the outcome is clear, the winner of that state benefits.
Another reason to support NPVIC!!
I expect a pro-Obama cellphone effect may have disproportionally affected NV, while a pro-McCain poor military polling effect affected MD.
Possible speculation of a "whitey tape" is most likely responsible for the large delta's in OK and WY while having an opposite affect in VT and HI due to the reverse Bradley affect.
That's my professional opinion.
A minory inaccuracy in the piece:
Obama visited his grandmother on her deathbed about 10 days before the election (over the weekend of the 24th of October).
She passed away a day before the election, on the 3rd of November (continental US time).
There was no funeral - Obama and his sister scattered the ashes in a ceremony on December 23rd. (Source: http://en.wikipedia.org/wiki/Madelyn_Dunham#Death)
Actually, Obama went to Hawaii as president-elect to attend his grandmother's funeral. His visit to Hawaii immediately before the election was to say his good-byes to his grandmother who died just a few days later on the day before the election.
Unless I missed it, 538 has never done a comparison between Nate's model and the predictions of pollsters. The 538 model is supposed to provide a significant "alpha", outperforming models without the clever covariates etc. So, how did you guys do? Was it worth it?
Unless I missed it, 538 has never done a comparison between Nate's model and the predictions of pollsters. The 538 model is supposed to provide a significant "alpha", outperforming models without the clever covariates etc. So, how did you guys do? Was it worth it?
As I understand it, most of what the model did was overkill. Most of the regression and factoring in older polls didn't do much better than just averaging the past two weeks of pills before election day did. Most states were well enough polled that there was enough information there.
Most of the graph can be explained simply by the 'losers stay home' effect. It's easier to answer your phone than go to the polls. Nevada probably reflects a mistake in the assumed make up of the electorate. I haven't checked the pre-election polls, but the Nevada electorate was 77% White in 2004, but only 69% White in 2008. If you weighted your poll by the 2004 demographics, you would've underestimated Obama's vote by roughly 4 or 5%, even if you accurately reflected the respondents' preferences.
I agree with David - this data could be substantially explained by the fact that there was considerably less polling done in the "landslide states," for obvious reasons, and that the polling aggregate was therefore based on less data.
My two cents. Closer states mean more polls, more data means less error. Silver's model made use of demographic regressions using national data, so there may be a tendency towards the mean somewhere in there. There also may be an inclination of pollsters to publish polls that make the contest look closer than it actually is, because polling that state would become more valuable.
@ Karl-- I think a 3rd order function would be inappropriate here. Nate's goal is a 1-to-1 correspondence between his predicted Obama vote and the actual Obama vote, and the X=Y line is a nice reference to show discrepancies.
Mark - Be clearer with your terminology. A "Bradley Effect" would be where whites say they'll vote for a non-white, but don't; clearly that didn't happen here. Race being a factor in the choices of the electorate is well-documented, but also easily predicted by polling data.
The VT/WY anomaly is much more easily explained by a lack of polling. The outcome of those races were never in doubt, so no one was willing to spend too much money to find out exactly where the candidates. Less data means less statistical precision, which is clearly what we see here.
I can tell y'alls what happened in Nevada... all the Cali based Obama supporters, sure that Cali was a deadlock, either called, or hit the pavement in Nevada.
The discrepancy would have been much higher if Nevada wasn't also bordering Utah.
HI is obviously a homestate effect.
Obama did awesome. let's just leave it at that......
Poor dodd
Raj is right. The Obama campaign's dirty little secret was that they paid for a week in Vegas for hundreds (maybe thousands) of us California college students so that we could spend the days knocking on doors for him. The Clark County (where Vegas and 80% of Nevada's population is) democratic party voter turnout was something like 90-95%. There was such an enormous GOTV effort that just about every democrat got out to the polls.
I looked at this a while ago, though with pre-certified data.
The only trend I was able to find, was that Obama did worse than expected in states with lots of young people. See http://stochasticdemocracy.blogspot.com/2008/11/pollster-inaccuracy.html
But my data is available at http://stochasticdemocracy.blogspot.com/search/label/Judgement%20Day , so, people are free to do their own regressions.
Maybe I am patting myself on the back a bit since i was there, but I think Nevada had a lot to do with the huge influx of Obama volunteers from CA. I saw some stat somewhere (here?) about the huge pro-Obama disparity as far as voters being contacted face-to-face in NV.
The influx of CA Obama workers certainly played a very large part in the unexpectedly large win in NV. (LV is closer to 75% of the pop. than 80% and closer to 70% of registered voters) Barack and Michelle also visited here a zillion times, which helped (after 5x more than McCain).
Speaking as a resident of Maryland, which voted slightly less for President Obama than predicted - the expectation that it would go for Obama was extremely high, and thus there was a dearth of polling after the February 12 primary. If I recall correctly, there may have been about 4 or 5 polls between the middle of February and the end of October.
Contrast that with Indiana, Virginia and North Carolina, states that hadn't voted for a Democratic Presidential candidate since 1964, 1964 and 1976, respectively. The perception that these states would be competitive created the demand for poll after poll, with multiple polls being published weekly. All those formerly solid red states all voted very close to the final polls.
More data was available for the latter three states mentioned above than was available for Maryland, thus there was a greater probability of accuracy for those three states.
I think there is a simple explanation and a few people sort of dance around it. Sure, some from CA went to NV and that helped with the GOTV. I think this is part of a larger explanation that gives us the results seen in the graph. Quite simply, Obama spent money in states he thought he could win. In doing so, he put boots on the ground, calls over the wires, letters in the mail, adds on TV and radio and internet, etc. Even in states where he had a decent lead, he still spent some money. Where he didn't think he could win, he did none of this (and it is likely that McCain spent in these states, though not a lot because he simply didn't have the cash).
If nate regressed these results with ad spend and other resources in state in the last 2-3 weeks, I think we might be able to explain some of the outliers.
As someone mentioned above, Nevada's demographics have changed a lot in recent years - including the lower white population.
At the peak of the growth, about 6,000 people moved to Clark County every month, and it is notoriously difficult to measure the demographics of recent arrivals. The information from their old drivers' licenses is about it. Making assumptions about their political preferences is that much harder.
For example, sometime in 2007-2008, Washoe County (where Reno's located) turned blue unexpectedly. Other Western states have experienced big changes, but none have been as profound as Nevada's.
The 2008 cycle revealed much about the new Nevadans, but I suspect they'll have some surprises in store in 2010...
Tim Quigley said...
If nate regressed these results with ad spend and other resources in state in the last 2-3 weeks, I think we might be able to explain some of the outliers.
Tim,
That might work in a lot of states, but there are many that it wouldn't work.
Example - to reach Northern Virginia, the ad $$ would have had to be spent on DC area media, some of the more expensive in the US. Those ad $$ also reached very, very safe (for Obama) DC voters, and the very, very safe Maryland suburban counties of Prince Georges and Montgomery. The DC and (above named) Maryland jurisdictions have about 40% of the DC metro area population, and probably the same % of voters.
Also consider that the Philly market reaches into, and covers the lower half of New Jersey (the northern half is covered by the NYC market). Pennsylvania was fought over, NJ was not, but southern NJ received a lot of ad dollars because they were spent in the Philly market.
Indiana was heavily fought over, but NW Indiana is covered by the Chicago market, and almost nothing was spent there, as Illinois was safe, and Lake County in Indiana was safe, but Porter and LaPorte Counties in Indiana were more up in the air. In the rest of Indiana, the Fort Wayne market bleeds into Ohio (no loss as both states were contested), but Evansville bleeds into Kentucky (loss, as Kentucky was a given for McLamebrain), and Terre Haute bleeds into Illinois (granted, downstate, but Illinois was safe). Also, the Louisville market includes a major population center in southern Indiana (Jeffersonville, New Albany, Clarksville, etc.), but little money was spent in Kentucky, therefore little was spent in a way to reach major portions of southern Indiana.
As I said, such an analysis might work in a lot of states, but there are enough anomalies to raise a lot of questions about how to do the methodology.
Just to add to what I wrote above - since the entire state of Maryland was considered safe for Obama, the Baltimore metro area had almost no advertisements for Obama or McCain.
We were saturated with ads for Maryland CD-1 (Kratovil vs. Harris), but the rest of the Maryland Congressional districts were very safe for the incumbents. As a result, there was almost no advertising seen in the Baltimore market for the races in CDs 2, 3, 6 and 7 (CDs 4, 5 and 8 are entirely in the DC market).
I may be able to provide some insight into one of your outliers - Maryland.
On election day, I personally waited in line for 2 hours and 45 minutes to cast my ballot (and this in a state that was never a battleground). I have no reason to doubt thqat many, many others who intended to vote on their way home from work, saw the lines and just gave up.
The only reason I stood in line that long was because their was a question on the ballot amending the state constitution to allow Early Voting. Thank god it passed.
I live in Hawaii, and expected a very high turnout for Obama. The whole state seemed rabid for him in a way that was not represented in the polls. I doubt his pre-election visit had much to do with it; the local coverage was quite respectful and minimal. It seemed a lot of folks who normally don't vote turned out, in part because he presented a sort of "new wave" progressive aura, and also being a favorite son. To say that Hawaiians are supportive of our own is an understatement; recall the American Idol competitor who won week after week, although she was clearly out of her league. Whole companies gave over their phone systems for phone banking, and many callers did not know anything about the show, only that one of our own was competing and needed support. They crashed the state's phone system repeatedly so that hours for calling had to be extended. The campaigning for Obama all along was almost oppressive, even though the state was so blue as to be deep navy. The whole state together really is a very small town in feeling, and it seemed everyone knew someone who had known "Barry". In any event, the high turnout was not a surprise.
At the Princeton Election Consortium, I did this analysis back in November. My general interpretation was that likely-voter screens underestimate the relative amount of enthusiasm for the state's dominant party. This is a relative measure and could be explained in several different ways. See my post for more.
I analyzed this question in 2004 as well. There you can see a similar tendency, though it was stronger for Bush-favoring states than for Kerry-favoring states.
Sam Wang
Hi Sam,
First, big fan of your blog -- nice to see you poking your head in over here. =)
Second, I was just about to point out your post on this subject. Personally, I think that during a blow out, the losing candidate's supporters are less likely to show up -- there's some emotional value in showing up to be on the winning team (that whole "I was there" thing), but refusing to show up (and therefore distancing oneself from the result) might be a way to take the emotional edge off being on the losing side. (Think of it as a physical manifestation of attributional bias, e.g. personalize success, depersonalize failure.)
During the campaign, I came across a journal article that touched on the subject -- Brown (1993), "Nonlinear Transformation in a Landslide," AJPS 37:582-609). As I recall, it was focused on the topic of how an electorate in the midst of major transition behaves, but it had some data that seemed to match my hunch: Turnout is depressed overall during a blowout (the likelihood of being a decisive vote has dropped substantially), but turnout is depressed MORE on the losing side than on the winning side.
I remember this coming up in the last couple of weeks when some people were panicked that once people start thinking "landslide" they won't show up, so they were running around screaming at the top of their lungs that Obama could still lose -- technically true, but the NET effect is still a gain for the winning candidate.
Now, if you could get lopsided information -- convince the losers that it's a landslide but the winners that it's close, or vice versa -- you could make a big bump in the numbers, but I don't see anything in this election that would be an example of that. (Nevada might fit: You could think of GOTV efforts as being something along these lines, at least in the abstract sense, and the efforts there did seem to be rather asymmetrical according to Sean Quinn's field reporting.)
Not sure if this was mentioned, but, there's a very large Hawaiian population in Nevada. Not sure if the numbers are big enough to account for the effect you're seeing here but, probably worth mentioning.
Matt
Hi, Sam. I did this analysis back in November too and posted it on my other blog. I reposted it here because I thought the 538 community would also be interested in it.
The Obama campaign’s organizing efforts in NV were absolutely astounding. It was brute force of will. I’ve never seen anything like it.
My wife and I were witness to just a part of the effort conducted from the Obama state HQ in Honolulu. One key was the demon-like computerized phone call mgt system that tracked each NV voter relentlessly through the registration and voting process, offering help on the ground at every turn.
Over a 2-3 month period, dozens of Hawaii volunteers, ages 12 to late eighties, made HUNDREDS OF THOUSANDS of cell phone calls. As we made these calls and fed the results back into the system – and as all the other efforts discussed above were made, we watched the balance in NV steadily shift from several points in McCain’s favor to several points in Obama’s favor.
Truly, YES WE DID.
I'd say the discrepancies had more to do with the size of the sample. Vermont and Wyoming are two of the least populated states.
Everyone has missed two very significant factors for Nevada in the last election, that as a long-term resident of the state, is very clear to me. The Obama team did have a well-organised group on the ground in the state going door to door. This was a factor, but it should have been obvious to pollsters.
Nevada has a remarkable early voting system. In last November's general election, early voting accounted for 46.50% of the total votes cast. Early voting was set-up with plenty of voting machines, properly manned by both election officials and volunteers every day for two weeks prior to the general election at all major indoor shopping malls, plus several government buildings frequented by local citizens. There were also many mobile early voting systems set-up at many large neighborhood grocery store outdoor strip malls, and other locations heavily used by residents, for a few days each. The lines were generally non-existent or short through-out all of early voting. It was hard not to vote early in Nevada last November.
The other factor few pollsters consider is that Nevada has used an easy to operate electronic voting system with a secure paper trail since the 2002 mid-term elections. Rep. Dean Heller (R-NV) was then the Secretary of State tasked with setting up a secure electronic voting system, and he wisely used the preexisting government tech employee assets from the State Gaming Control agency to assure its integrity. What government agency, federal or state, is better suited to assure a hard accurate count from an electronic device than Nevada's Gaming Control Agency? They've been at it for a very long time, and it is generally considered to be both non-partisan and honest.
These two factors contributed to a higher than historical norm voter turn-out for last November's election at 59.16%, and a larger voter turn-out was strongly in the Democratic Party's favor, because of the dissatisfaction with the Bush Presidency. (although higer voter turn-out might always be in the Democratic Party's favor)
Another factor that affected the vote less significantly was the deplorable Libertarian Party slate of Barr/Root. Nevada has a strong libertarian population, but they tend to be strong social libertarians as well as economic. Barr was unpalatable, and Root is a sports tout who lives in Las Vegas. A sports tout gets less respect in Nevada than a used car salesman, telemarketer or politician. None of these candidates placed third in the election. Barr/Root came in fourth.
看房子,買房子,建商自售,自售,台北新成屋,台北豪宅,新成屋,豪宅,美髮儀器,美髮,儀器,髮型,EMBA,MBA,學位,EMBA,專業認證,認證課程,博士學位,DBA,PHD,在職進修,碩士學位,推廣教育,DBA,進修課程,碩士學位,網路廣告,關鍵字廣告,關鍵字,課程介紹,學分班,文憑,牛樟芝,段木,牛樟菇,日式料理, 台北居酒屋,日本料理,結婚,婚宴場地,推車飲茶,港式點心,尾牙春酒,台北住宿,國內訂房,台北HOTEL,台北婚宴,飯店優惠,台北結婚,場地,住宿,訂房,HOTEL,飯店,造型系列,學位,牛樟芝,腦磷脂,磷脂絲胺酸,SEO,婚宴,捷運,學區,美髮,儀器,髮型,牛樟芝,腦磷脂,磷脂絲胺酸,看房子,買房子,建商自售,自售,房子,捷運,學區,台北新成屋,台北豪宅,新成屋,豪宅,學位,碩士學位,進修,在職進修, 課程,教育,學位,證照,mba,文憑,學分班,網路廣告,關鍵字廣告,關鍵字,SEO,关键词,网络广告,关键词广告,SEO,关键词,网络广告,关键词广告,SEO,台北住宿,國內訂房,台北HOTEL,台北婚宴,飯店優惠,住宿,訂房,HOTEL,飯店,婚宴,台北住宿,國內訂房,台北HOTEL,台北婚宴,飯店優惠,住宿,訂房,HOTEL,飯店,婚宴,台北住宿,國內訂房,台北HOTEL,台北婚宴,飯店優惠,住宿,訂房,HOTEL,飯店,婚宴,結婚,婚宴場地,推車飲茶,港式點心,尾牙春酒,台北結婚,場地,結婚,場地,推車飲茶,港式點心,尾牙春酒,台北結婚,婚宴場地,結婚,婚宴場地,推車飲茶,港式點心,尾牙春酒,台北結婚,場地,居酒屋,燒烤,美髮,儀器,髮型,美髮,儀器,髮型,美髮,儀器,髮型,美髮,儀器,髮型,小套房,小套房,進修,在職進修,留學,證照,MBA,EMBA,留學,MBA,EMBA,留學,進修,在職進修,牛樟芝,段木,牛樟菇,關鍵字排名,網路行銷,关键词排名,网络营销,網路行銷,關鍵字排名,关键词排名,网络营销,PMP,在職專班,研究所在職專班,碩士在職專班,PMP,證照,在職專班,研究所在職專班,碩士在職專班,SEO,廣告,關鍵字,關鍵字排名,網路行銷,網頁設計,網站設計,網站排名,搜尋引擎,網路廣告,SEO,廣告,關鍵字,關鍵字排名,網路行銷,網頁設計,網站設計,網站排名,搜尋引擎,網路廣告,SEO,廣告,關鍵字,關鍵字排名,網路行銷,網頁設計,網站設計,網站排名,搜尋引擎,網路廣告,SEO,廣告,關鍵字,關鍵字排名,網路行銷,網頁設計,網站設計,網站排名,搜尋引擎,網路廣告,EMBA,MBA,PMP
,在職進修,專案管理,出國留學,EMBA,MBA,PMP
,在職進修,專案管理,出國留學,EMBA,MBA,PMP
,在職進修,專案管理,出國留學,婚宴,婚宴,婚宴,婚宴
住宿,民宿,飯宿,旅遊,住宿,民宿,飯宿,旅遊,住宿,民宿,飯宿,旅遊,住宿,民宿,飯宿,旅遊,住宿,民宿,飯宿,旅遊,住宿,民宿,飯宿,旅遊,住宿,民宿,飯宿,旅遊,美容,美髮,整形,造型,美容,美髮,整形,造型,美容,美髮,整形,造型,美容,美髮,整形,造型,美容,美髮,整形,造型,美容,美髮,整形,造型,美容,美髮,整形,造型,設計,室內設計,裝潢,房地產,設計,室內設計,裝潢,房地產,設計,室內設計,裝潢,房地產,設計,室內設計,裝潢,房地產,設計,室內設計,裝潢,房地產,設計,室內設計,裝潢,房地產,設計,室內設計,裝潢,房地產,設計,室內設計,裝潢,房地產,進修,在職進修,MBA,EMBA,進修,在職進修,MBA,EMBA,進修,在職進修,MBA,EMBA,進修,在職進修,MBA,EMBA,進修,在職進修,MBA,EMBA,進修,在職進修,MBA,EMBA,進修,在職進修,MBA,EMBA,住宿,民宿,飯店,旅遊,美容,美髮,整形,造型,設計,室內設計,裝潢,房地產,進修,在職進修,MBA,EMBA,羅志祥,周杰倫,五月天,蔡依林,林志玲,羅志祥,周杰倫,五月天,蔡依林,林志玲,羅志祥,周杰倫,五月天,蔡依林,羅志祥,周杰倫,五月天,蔡依林
No doubt demographics had a lot to do with the results in Nevada, but as an employee with the SEIU working in Northern Nevada I've gotta say that the number of staff and volunteers on the ground was just enormous. When knocking doors we'd almost always run into volunteers from the Obama campaign no matter how rural the area we went into. This probably explains how Carson went Blue for the first time in ages.
Add in McCain's virtual absence from the state and you've a very lopsided swing toward Obama. At UNR, the main university in Reno and Northern Nevada, the McCain camp had a single lightly-manned table while there were occassionally up to a dozen different individuals doing registration or GOTV for Obama.
Like anything else, there were a host of factors. Demographics made it possible, a great ground operation led to a win, and McCain's absence from the state led to a surprise margin.
美女聊天室,18禁,美女自拍,美女自拍,美女聊天室聊天,免費聊天,免費視訊,一對一視訊,一對多視訊,聊天人緣,談情說愛,成人視訊,影音視訊聊天室,相片,靚辣妹群,成人,DVD,成人影城,成人影片,線上成人影片,18成人,18禁,宅男,巨乳,女高中生援交妹,美女妹妹,美女,高貴美少女,美乳偶像護士,酒店,24小時,酒店美人
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,
Post a Comment