How the Georgia runoffs changed the 'polling is broken' narrative

Nate Silver and Galen Druke look back at the results of the Georgia Senate runoffs and discuss what the Democratic wins say about polling.
39:47 | 01/29/21

Coming up in the next {{countdown}} {{countdownlbl}}

Coming up next:

{{nextVideo.title}}

{{nextVideo.description}}

Skip to this video now

Now Playing:

{{currentVideo.title}}

Comments
Related Extras
Related Videos
Video Transcript
Transcript for How the Georgia runoffs changed the 'polling is broken' narrative
The Scottish teens who were making bets maybe they've stopped that helped create. Gamespot. But. It's topped. Kids. Sorry there is and what cycle of gamespot so is extremely confusing. Hello and welcome to the fact dirty politics pod cast I'm dealing drew. Eyeing me smoldered. And this. A lot all Clark. RA this is our long lost. Post a Georgia model clock episode. As listeners well now the real and the twentieth when he election those double senate run ops in Georgia. Were eclipsed by the attack on the capital and we've been covering the fallout of that ever sets so today we're gonna take the opportunity to look back on what happened in those two elections. We'll also try to set some expectations for the electoral environment of the next two years that's a large part due to be dictating how lawmakers. Haiti's. And by the way related to that we launched our approval rating tracker for president I didn't today what you can go check out on the web say. He begins act 54%. Approve 35%. Did support. But hey before we get to any of that I have to ask use. Did you manage to sell off all of your GameStop stock before it crashed and act. You know I'm I should be like in that corner fascination point for the story every kind of not. Thing is what's his position but yet it not I do not get in on the game. Spot in the GameStop. Ole. White is you're were not it has been too it's time on this but what is your liking 92 take odd. What is the take away from date GameStop debacle. I mean. LO well efficient market hypothesis I get sick I don't know crappy. Certainly it's pretty far removed from like Tom. From. You know the fundamentals of this of this company expose a there was some initial idea that maybe they can be kind of you know my takeaways I want it I want that PS five and one is fine on it. Code vaccines. We're waiting. Wearing a light I've forgotten those. But. If you could only one which would be the vaccine and Emma yet. I mean I'm I'm kidding right you know aren't so let's get to the matter and as much fun the GameStop story has been so in the end. Rockwell were knocked. But from a nuts and bolts perspective. Why did Warren pocket costs off wins those two elections. It just feel like it was much longer go a little ones. I think they want and Vick is Georgia is increasingly purple and offer a variety of reasons. Including the conduct. Of trump and Republican. Elected officials. After they hit November election. Democratic voters were more vote motivated turnout so a higher percentage of Democrats especially black Democrats. But turned out again in January after having turned out in November when you head to grace that we're very close before. It's NAFTA based within the doubts were David Perdue actually won the morality of oats can start us off in November and went whizzing by also by. A point or so one point two points to be exact. In January so. You know he had about it I three point swing in this mostly looks like to me. Turn actor and so subways and talked to us think when we talked a little bit after Georgia it is just kind of more like that it. Stacey Abrams thesis of Georgia we can get a voters turn out. More than they can our voters turn out towards purple upper regions are to win some races so I think that's. That's more or less what happens. So how well June the ridge the takeaways. From. The general election in Georgia and the runoff election in Georgia because. During the actual you know November presidential election it seems light. Those upscale suburban areas. Flipping votes towards Joseph Biden from I've inverted for Republicans. In large part contributed Q Biden's win because there was high turn our. Across the board. Whereas in the runoff it really was. You know unequal levels of turnout turnout among Republicans was quite high but turnout Ian you know largely black priest saints was even higher. So what does that tell us about the democratic coalition. Going forward I mean is why not clear model. For success can you combine the two. Where does that leave us. I mean I think in order Democrats would have to mix and match a little bit I mean I think it's scary up I kind of look at from the GOP when it anymore I suppose we are like. The thing that scares the GOP is if you have a certain. Voter low engagement voter whose only gonna turn out if down trump himself. It's on the ballot and yet you know you don't gain back anything reflect a suburban swing voters construct is off about it and one prior that I had was that. OK Joseph Biden won the election over the fact that trump was not an alternate to bite when election is that big. Asterisk but Joseph Biden won the election so we could have a certain suburban swing voters like well I didn't like from but I also like lower taxes so. I'm an emaciated produce stays in office so Democrats can raise my taxes right an attack in order apparently either where avenue that type of voter. They were may be just as many voters who said well I have to give Joseph Biden a chance to enact his agenda I think Joseph Biden's pretty moderate and reasonable maybe so you know the GOP you can blame it. Think that deeply about it do most people just say it like hey I'm gonna vote for the Democrat or hey I'm gonna vote for the Republicans. What do voters strategizing law that you're. 90% of voters don't think like this for the 2% make a difference in this race right. There's a fair amount of evidence of tactical. Voting to it kind of preserve divided government traditionally right that's one reason for the mid term. Penalty that part typically suffers maybe there's less of that now. You know bipartisanship is something which you know is functionally very very difficult in the United States and so voters you know but maybe if you want. Gridlock and that's part of the benefit of it you know what I mean. If you like the status quo on fiscal policy that you don't like Donald Trump then. You would be kind irrational today. To you know want Republicans to control of the congress will. Keep in mind I know it's like scenes like an incredibly obvious point but you know keep my Georgia was. Before. The capital. Insurrection it or hate. You know I don't know what would have happened if he held it afterward I think actually my guess is it would have won. By more given that Republicans now have a new wrinkle of being communities somewhat dangerous. Political party. There are remodeling in some states. Small numbers of voters but increasing numbers who have you know who have switched over from Republican registration to democratic or others it's the Capitol Hill. And you know that indicate people there's certain that the voters kind of getting off the GOP train in and was probably more attentive. Higher income. Higher socioeconomic status. May be living in Rick's mixed race communities tepid order that would become inning in Georgia for example. Are so there's a broader conversation here about where the Republican Party goes from here and what kind of coalition Democrats can keep together. When you know they aren't the party in power and no longer are some of the benefits being opposition. We're gonna get in much deeper into all of that in regular podcasts. But I wanna dig into some the numbers here so all hold how did the poles view. In Georgia in the house. So kind of an eight plus plus basically mean pull up or final Georgia polling. Group. So our final polling average war in Iraq would like to point one points and he won I believe 2.2 one what is. Yet. He got it throughout carefully two point one Galen he round if you ground more precisely at nine. It give him. So was every one point eight B one by one point do so between the two races you have it. Three. Poll here on average it's really really good. So. What happened. At a Havel I have layers of questions here. First oral. We talked on this podcasts about the fact that the eight plus rated name brand pollsters. That we think light. Mod rats or New York Times op shock horror. You know ballistic Washington Post ABC news that's utter those pollsters were not in the field Ian. Georgia before the house there were a lot more experimental and lesser known pollsters are there pulling the public. Does this mean that those lesser known more experimental pollsters that we have. -- consider to be equality pollsters are on to something. I mean there's a mix of different types of polls hold Georgia including a couple of traditional polls that where partisan polls. You know you had but yes mostly an eclectic much shall we six. I don't know I think they are to something maybe our the other polls. Broken maybe just an element of. Two wrongs make a right. Maybe ending indeed there are some things I mean yeah I mean I just wish for Medina simply that we. Had in. Mind that there quinnipiac for someone. Told Georgia rate just gonna visit comparison would have come in with. Awesome plus five is an a plus water plus suit right or not I don't know right. And without having back comparison point there's no light. You're still. Light colored nonpartisan poll I don't think in Georgia there's is that my color partisan polls Anderson lots and lots of online in in. Pessimists and whatever else right but it's just it's very hard to answer that question I think right. You know what being a noted subprime is that in elections in the truck era. In which doctor was not a ballot. The polls did pretty darn well right meaning between ET mid terms mr. to run up. The Virginia gubernatorial election in two when he seventeen various special elections look at Doug Jones race right. You know song were a little sort of has been general was a eight. Pretty strong senate races for the polls so maybe there's something about liked. Truck in particular. That was causing problems from lower. Certain types of polls right. I mean I think what what that thing. You know when drugs not on the ballot polls are fine when front is on the ballot there's something ripples can capture. That there are low propensity. Voters. Who do not ordinarily vote and are hard to reach. On the phone and to order have low social trust navy's they're not inclined to like take a poll. And you know. And those who voted to be screened out as a partridge in the first place for the most support problem or maybe they're being screened out. But likely voters screens. By the polls that works just elect a voter screens and so therefore where you have some missing. Truck boaters may be which is separate and shot a truckload Saturday necessarily give it. Deceived you about their intent it's that they will. There's socket you know there are two we took them and they have. Demographic characteristics that make it that are you know go beyond just like education. And our hearts it. Wait your way out of the BI GHT right you know so why can't date. So the more eclectic polls. Seem to have fewer problems I mean I don't write. You'd have to go to case by case basis I think there are some of them that are doing different things that are Smart I think there's some of them that. He had gotten a bit lucky right if you credit for example if you call people. Landmines are automated polls those tempted pretty samples that are you know. Kind of older and in more rural right. Usually got people you have to whites and I'm so that could be case of like. Two wrongs making it right potentially. You know some of the more online oriented firms me I don't know maybe they had better ways to reach people actually response rate setters are Haider. In online polls although it's not a random sample the target rate example but like. Actually people that you do get in your panel. Point participating and so on so I don't know again these pastoral places are to answer without that. Without a point of comparison is not having the traditional. You know mom and author of shutter remembers Sheehan and having pull Georgia. 1 other point on this that's worth bringing up is that the polls were actually quite good Ian Georgia Ian November. And so is another element here at that Georgia is. He's eager to poll partner attorney hypotheses for why. I don't actually don't know I think we had a really totally an act why. Why the polls were quite batting in. Some states in quite good and others and furthermore why. These patterns seem to be similar persisted because ordinarily you think that pollsters would in historically. Incorrect like that we missed in this state that state. A couple times at arousal it's really kind of look into the that your I don't think we have great answers today. To that. You know some of states where the planes missed off in November states it head. Higher incidences of coated spread. At that time of the election. The theory being that Democrats. Being you know. I'm more inclined to do listen to public health officials were more candidate in the new to stay home stay home your court. And make you answer phone calls were often right. Georgia had a worse covered situation during the write offs than it did in the run up choose the presidential election if. Also Georgia actually did George yet at times and he runs with Covert Georgia actually did not. Have much movement relatively speaking in likely October. It was doing fairly well. It had more or in in. Yeah in. December. January so you think the polls like it works there that's the reason why but. They did if anything little better obviously graph and so you know that period as a kind of neatly explain things as much. You know one thing. I have one question I have here is what we've talked in the past about why it does this year stepping quite a bit back. When we talked in the past about why it's easier to poll for example the United States than maybe the UK is that there is just. A law. Clear demarcation. Between demographic block Heinz demographic groups and so you you can basically figure out that. If your XYZ. Type of person who lives here who has this racial or ethnic background whereas this educational profile was this gender. You have a much better sense of how they will end up voting. And so because Georgia is you know you can look and say okay it's 30% black it's that this percent white educated this percent white non college educators. Where is a state like Wisconsin and it's more homogeneous and you can't really like debt at large chunks of the population by just breaking down right demographic groups does not. It actually make it easier to Poland Georgia. Yes for sure right he could almost just kind of infer what nine accident or do just based on you know their information voter file and George also has information voters. Race. And. Because it's covered by the voting rights act is the reason I collect that data and other states. So we have more precise data and it's you know it's not like these stately you have like. He had as many states but you know Democrats and Georgia are. Dimple plot Democrats and George are very democratic race hysteria like something. You know sum up swell like conservative. Republican. African American voters right so again that that helps a lot the more Chris the Democrat experts are in the more competitive. Almost triples like a model where your Pryor's basically can can make things easier for sure. So we've gotten into the nitty gritty. The data in Georgia. How much does this change our understanding of the 22 party election. After all side and I write the narrative after the November election was like. Loud you know Biden barely eked it out it wasn't exactly repudiation of trop just because Republicans beat expectations and the house. And held on to you know a lot of seats in the senate. And so it was kind of a mixed message does this change. Much about how Alou. It is rigged. Well looks a marketing net literally put some of them well and that's a problem for the GOP. What is that the thing that's driving the lower turned out or is it simply that trumps not on the ballot. Because trump was tallied eight voters that the election was raged before the November election to. By when he was on the ballot does lower propensity voters seem to return. I don't know I mean the problem is we have light one data point. And he's drunken really determine if you don't really know where it's what twelve request with a twenty point so we can't do anything but and in corporate and agency Wednesday. I don't know right but like I think people like if you see. This election was stolen and rigged you're gonna catch your ballots right. It seems obvious that some people will maybe turn out as a result write it so that's how Pryor's. And I don't know right tipping direction only matches what happened it's not the only explanation but like. People seem hesitant to like handed tickets involved either. Yet but obviously now that light the story that makes sense pressure. Are. You know I don't your time it was Big Apple stores in Georgia with the daily podcast with listening to view you know this may head when he went on to rally is in trouble in finding people alleys he would say yeah well. I'm gonna turn out that if you rallied in the self selected sample rate Condace and ultimately well. And it would return of course but I feel like maybe my freedom of friends and turnout or something else I don't know I mean they're more detail wasted yet that like. And it'll be nervous if I were Republicans. So. To that point more specifically and that's assortment will cover for the coming two years we don't afternoon deacon to read. That much right here right now but considering this may be art math last model clock for awhile let's talk a little nervous leaves the two parties. 42020. And again because it will shape lawmaker behavior between now and that it's already shaping lawmakers' ears so. The current number of seats in the house is a little unclear because of one lawmaker staff and that also kind of endless election in an upstate New York district has still not. How to resolve determine. Republicans probably need to win all need six seats in the house to take control of that chamber and once he. Ian the senate. Should our starting expectation do you that Republicans will control congress and 20/20 two. You know. Hum. No I don't. Thanks so I think this is a man I think. The expectation should be that it's. Highly competitive. And it's UP as you know. Maybe it's 5050 looks cute girl friend's chicken girlfriends. Scottish teens and see what they think. Scottish team discourage teens. What I guess you're actually. Went of those artists. Discourage teens who were making bets maybe they've stopped that helped create. Gamespot. But. Its top. It's sort there is a replica of gamespot so is extremely confusing 50% chance Democrats retain control of senate. For troops assistant encrypting vote house so basically toss ups. You know. Seems reasonable. Enough to me indecent pattern to the Democrats are. Have better chance in the senate and house but actually having somewhat decent man captain. The senate where of course only 13 of seats are played worse. House probably lose and we'll seat because it. Redistricting. Wiped out the small edge they have I think George it ought to be a little bit of it wake up call city. Our Pryor's assume it's typical. Mid term backlash their girl could be public I think. You know the events of 16. Parker. The only area people's minds like in certainly so for one thing I don't think you really have. The drop off in democratic turnout. That you'd ordinarily expect when your president is in charge right because I think Democrats will remain very worried that like. To be blind right that if Republicans control the senate in 20/20 four house dental tries to election I think that's a ticket right. I think Democrats you know he did it would regret their right to worry about that are not like I think that's it. Different type of issue that democracy itself is kind of on the ballot here. And it I mean you can kind of compare a little bit too after after September 11. Or after Watergate and personalities it was a case we're kind of the usual religions were affected by by. Events in the previous. You know a couple of years right so yeah. Saddam out of about but the fights of the last three months well kind of continued for hour. The next two well. There and there are some other stuff to you here right on the GOP is gonna have a lot of conflict between. Truck Republicans and anti. From Republicans. Bryant equipment and maybe it's authoritarian Saddam authoritarian right you know that could lead to them dominating so. Sub optimal candidate threat like it became moderate for example. Came in Ohio where proper retired would win going away. If Jim Jordan gets the nominees and nation. I don't know I think it's more competitive race essentially you know I'll. So the probable nominee is an optimal candidates and in. Some races also like. If you look at the timing. Kind of economic recovery reported to have. Well I mean it's not yet you're probably have to talk about Covert again and a upcoming regular podcasts you know if you if things go woo. Some what. To the textbook. Then. Buying. Summer. Certainly by fall. There's a much greater degree of normalcy people's lives. People who want the vaccine can get things start to come back to economy will also. Rebound right. By the time tweet tweet tweet to maybe things. You'll actually. Close to it. Normal normal or not partially normal all right Tom and the economy is rebounding game for misconduct. Difficult period so things could lineup after a magical mystery decent. Approval ratings of of their action to hype I sort you know he's. McLaughlin trucker was there's this area here we're. We are right as that wind is back for the first two years his term. Are well we've covered that amount of Gartner so let's move on to you some listener questions before we wrap things up just ask. Now that we have another election due to under our belts while the model be much more bearish on democratic chances in future presidential and war. Statewide races. And perhaps what there's was getting act is. Ease the kind of like fundamentals model of the overall partisan of the partisanship of this speech. Is acts and so we should just expect it's the right elections. Much more clearly aligned with acts regardless of what the polls say. I mean there's a couple of questions humans rightly wonders like. If you believe the model as it is and yes Ohio and Iowa now with more years of data it. You know there still are going to be for example. Some 20/20 after my muscle and 120. You're looking it was seen and when he twelve. Indeed including 2012 where Obama actually won. Ohio and I seems crazy now repeated central between forward. Then you no longer have that last gasp of Obama's strength in the midwest and it's purely truck there dated basically and so yes that would meet. The Pryor and Ohio and Iowa. Less eager. It to call them purple states anymore that a question of like after 20/20. If you rely more on fundamentals as opposed to polls in general right. That's something were. A year away from Abington process I think. What's what are your thoughts but a crew the boss fires that people generally are climbed to over fight. The last war and now a look I mean part of it is we are relying on polls. Adjusts. Right. If polls are wrong seem direction she lots of financial and reputation whole and other pressures. For them to not be wrong and right. Yeah I mean the question is is there like a long term decline equality polling is I think a pretty freaking. Tricky one. And for job to answer about before you answer whether. You're gonna dock and fun and well it art. Army terms model actually. Puts quite a bit of shock and fundamentals already so it might not need. As much changing is like and some Israelis were like the polls are all for example head Susan Collins. I'm even though she's bigger underdog in the polls the fundamentals weight pretty heavily on that race for example. Word tomorrow never got to carried away with like you meters in South Carolina so. The picture model is. I think probably. Eating. It's fairly pro fundamental as it is right is probably pretty decent shape. I mean less work for an era in twenty targets him. Yes I mean he moved to some basketball or baseball stuff but like. I don't think amateur models at a perform. We think really pretty well in this senate races this year in because it was like hedging affair the presidential model is kind of like almost. Philosophically designed to be. To kind of dial the fundamentals down. 20. Have yet to Election Day. So opera like this I am. Happy. That we have a mid term. Biffle were the presidential year right. That will help us. To diagnose. Is there something. Fundamentally wrong with polling there was something specific to trump. With some specific circumstances of 20/20 and it pandemic rate and maybe different certificates and when he sixteenth as the reasons people here mr. Hussein. I'm happy to listen what. Lower stakes and it as far as pollsters go reputation Italy eateries high six of the country. Victory coming at between 22. We have model that we think is already. Pretty well equipped to it to you know take polling was something of a grain of salt and then face frankly. Really difficult decisions in 20/20 four urban it'll be an important data point to look at. Are any of us thuggery during an S and 124. We really have another prospects also oath to protect record. Well and will also have the question of like. How much do we think we're actually having a democratic -- hasty election and when it went forward I don't mean it's an alarmist but. But you know maybe things accelerate. Instead of decelerating so. Georgia to a game. Okay next question that what pat back upstairs from it he asks what are the worst model nieces. In other words the differences between predicted margins an actual more. He's asking us to call ourselves out. I imagine there on the house side now the senate side because the senate they weren't like any huge upsets but. But I haven't yet we will we will look. That actually gets another question here which is. The president has the presidential and senate models seem to have done pretty well but house modeled wasn't as accurate as any ideas what was up with done. Out immediately I mean our house passed it was actually a little bit more bearish on Democrats and some of the competition and I don't mean what is district polls this work. Just weren't very good you know you can kind of up wait. Typewriters. Which we need to an analysis of race is what like I think like it probably isn't likely to get a case like. He's person prior. Fundamentals looked in better. Then blended version of what fundamentals post polls strikes will look at I mean. If the polling. Is. Not good. Then you can pick. And our models could have trouble races where it's putting a lot of weight on the polls priest priest triple street vote. Watch our business thrush and it is the empirical arguments that can be made that Georgia will keep moving laughed. Or shift back to the right. I think we've taken a lot of time trying to predict flight is predictable in which direction statesman moved in the answer is. Not really. Although any particular look at what migration patterns. You know more and more georgians are born outside of the state so. So immediately shift but you know I think Perry wrote about this recent states we are like you know. There's a view that hey North Carolina it's going to be a purple state now permit period and trending blue right and it hasn't quite. Done right. And we have tried to model these things cystic leader would get very far supporters and creative ideas and like. Predicting which direction states code and then then let me know. I mean once. Question I have suit that point. Or suggestion or idea is basically what are states have booming cities. Right if if the trends that are happening in Atlanta continues that I think you can. You would expect to see the political realignment in that state. Continue North Carolina. Wallace has like a growing. Research Triangle and Charlotte and so I want to covered have a booming metropolis. And again with the Virginia right like DC has been booming. And and same with Siemens like for example Phoenix Arizona so like if those areas. You're basically how the majority of the population in the state and the trends of continuing to grow as a city and become more diverse and have more of an educated work force continue. Like as I don't know is asked if it is not a good proxy for where his state has added. I mean. I must thinking about that. Pandemic and how that might affect. Migration patterns right kind of it first there's this notion that a lot of people gonna move it. Out of big city isn't it turns out actually well they're just beneath her case it rural areas that won't stop Soviet right but he didn't going to be things that are on. Shaken up by eight increased ability to work from home or people who regal looking can rarely decide. My eyes will do permanently in self worth right. And that could maybe effect mid size metros more than big metros as one theory awesome I think some dislocation from. California in particular. Also in Arizona and a Miami. The billionaires and Miami as astronaut billionaires so much as companies that. No longer require their pleas to be in California were moved their headquarters that could affect you know tens of thousands are aware. Hundreds of thousands of jobs potentially mean I could be. Material potentially some of these Sunbelt states if their current lower tax jurisdictions. That are. Attracting kind of knowledge sector workers in that seems like it could possibly be affecting us a little bit. All right let's answer one more question before we go. And it is another Georgia based question it is why didn't want Iraq to better than offs. I'm. David Perdue is an incumbent. Or they're both look at what's he was elected incumbent they can to have better English for their constituents. Left litter is I think rang. Very hard her right in. In the general election she said I'm more conservative tilt harm. You know I think that might have hurt her she's kind of saying I'm not a moderate anymore she cut it was seen as may be able to order. Trappist. You know if you had an election where. Where Democrats really benefited from high turnout among black voters and more Iraq. His plaque in the area few voters who were really excited about him and felt more lukewarm on Assad wasn't a big gap like. I don't mean those needles were not cats were more. Effective. This case is more historic and somewhat low you know a legend also has also kind of news pretty neat story to Diana I think I think. It's circular rice technological and slightly stronger candidate and docile for me be. Below left there was a slightly weaker candidate and party but to talk policy right. Yet all right well let's leave it there Germany parting thoughts might be our last model for. I eat. Yeah its. It is a let down. That we look like I think that. We each. They were not have a mob we cannot see how. May we can chat whatever you want because it really have a model to work office. Expert a lot of times I didn't either either have the model or an election. Hanging over my head it. It's like dating back to like 2019 year working and raptor coach at an NBA metric you're working and the primary model which took him panic time. We round up kind of doing a lot of work that your election model UN doing some work accident that much. On the commercial model. Between the election itself right suffer like. For two years we had I had something kind of hanging over my head right now whole. Now you don't they'd instantly and I'll. When you don't have a forecast model to work. And the independent XU lots of sciences department and sit home and contemplate what is the meaning of life. Concluded. Haven't figured getting it right. Madeira Pfizer PS five you know and thing and then. He'll feel a bit more to Justin Johnson I think it target groups in its that some of that you know. Yeah I know it's weird it's weird it's a weird. I'm very privileged spoiled so weird adjustment to make it in some ways. The fact that he has crazy election. Made me. Notice the pandemic last as I don't get to much for life collection at. Election years February write downs that you kind of noticed. Destruction from normal low. Our debt well as I've mentioned before us and looking forward to getting back to office it's been too long. But let's leave it there so think CUNY. Thank you. My name is Gail and returning each I was in the virtual control room Claire prettier Curtis is on audio editing. And a touch by emailing us at pod cast at 538 dot com you can also of course tweeted us. But it questions or comments. If you're a fan of the show rivas of Reading a review act fast or actually do it you know. There was a five star rating. Tell it tell us that. Tell me that you appreciated his three year story of forecast modeling. Or just house and us and thanks for listening and weevils. And.

This transcript has been automatically generated and may not be 100% accurate.

{"duration":"39:47","description":"Nate Silver and Galen Druke look back at the results of the Georgia Senate runoffs and discuss what the Democratic wins say about polling.","mediaType":"default","section":"ABCNews/Politics","id":"75548598","title":"How the Georgia runoffs changed the 'polling is broken' narrative ","url":"/Politics/video/georgia-runoffs-changed-polling-broken-narrative-75548598"}