America's next top chamber, modelled: the Senate battle


In, America's elections just six weeks away. The most important thing to be decided is who will win the presidency. Time before. has there been a clear choice between two parties to visions to philosophies and two agendas for the future. There's never been a vision like this. Sleepy. Joe Jeff from. What pundits had been talking a lot less about is which party will end up with control of the Senate, the legislature's upper house. Attention and donations have focused on Senate races. This week is a fierce battle begins over the replacement of Supreme Court Justice Ruth Bader. GINSBURG. The president plans to use the power the voters gave him to make a nomination. Senators will use the power the vote gave us. Either provide or withhold consent. As we say, fifth. Republican senators are keen to get a nomination through before the elections. Earlier. In the year, the party seemed set to keep its majority but President Donald. Trump's handling the pandemic and racial justice protests have opened the possibility the Democrats could take it. In America's rigid two party system control of the Senate is a big determinant of what any government can actually get done. But forecasting which party will win control is more complex than it might seem. Predicting Races for Congress is a bit more complicated than predicting the race for the presidency. There are thirty five different races fought in different states between different candidates over different issues and often using different rules. Then Rosenheck is the economists data editor. I have spent the past few months building a statistical model that seeks to predict the results of this November's congressional elections for the House of Representatives, and the Senate in the United States and so how. Do you get past those complexities to t to model it then? So our model makes use of a few broad types of information. It first looks at the state of the race nationally using generic ballot polling as well as the president's approval rating, and if you other predictors to say, okay, we think this is going to be a good year for the Democrats a good year for the Republicans overall. Then it drills down to the state level and. says. Okay. What did we know about this Senate race before we saw our first poll of that race? Well, we know how it voted in past elections. We know if there's an incumbent or not. We know the candidates fundraising totals. If there's an incumbent, we know how they voted in Congress. We know how much experience in politics candidates have, and then the final step is to blend those prior expectations with whatever the actual polls of the race. Say, and so it synthesizes all of that information in two and overall prediction by exploring three hundred fifty thousand different scenarios for the Senate and four point thirty, five, million for the house every single day. So it conservative explore every possible universe. We can think of that this election might take and see which scenarios are most probable but the universe that we're having these elections in is one that contains what the historical data don't, which is a pandemic. That is a very good point and it is definitely a weakness. We don't know whether one party or the other will benefit from that. But I think what we can say is that it would needs some pretty large effects that would need to be consistent and systematic across states to drive the results far outside the confidence intervals that we would expect based on decades and decades of data on past elections. With those caveats behind this then what does the model currently predict for? November. The model shows that the Democrats are a clear but narrow favorite to win back the Senate and currently shows that they have a sixty seven percent or two in three chance of winning the upper chamber, and it's kind of an interesting path of how they get there because assuming that they lose Doug Jones seat in Alabama, which is very conservative state they need to flip four Republican held seats and they only lead in four states and they would need to win all four that would be something of a tall order and what are those four states what are the issues there So the Democrats have pretty clear leads in Colorado and in Arizona in Colorado Democrats have nominated John Hickenlooper is a moderate former governor. He is currently leading in the polls against the Republican incumbent Cory Gardner. The picture in Arizona is similar you have a Republican incumbent and is now facing an extremely strong Democratic nominee in Mark Kelly. WHO's an astronaut? He's raised an enormous amount of money are model certainly thinks those two are not short flip the Democrats but eighty ninety percent that range then. There are two more states in North Carolina. You have a rather unloved Republican incumbent clue has been running consistently behind, and then finally there is main where Susan Collins, the most moderate Republican in the Senate. Now faces a very tough challenge from the speaker of the State House of Representatives who is up by maybe five six points on average. So you're got a headline prediction of the Democrat chances. Here is based on those four races being cinched up I mean is that really so certain? So. The reason why the model is so bullish on the Democrats is what we're calling the long tail, the donkey's long tail. There is a surprisingly extensive list of sort of Second Tier Senate races in which Democratic candidates are underdogs, but have a realistic credible shock to come from behind and win. So that's I Alwa- that's Kansas. That's both the seats that are up for election in Georgia, Montana Alaska and South Carolina, and even Texas so in every one. Of these races, Armato thinks it is more likely than not that the Republican will win however the odds of the Republicans win all of them in the Democrats win none are not so great and if Democrats even score one or two upsets from this very wide list of targets states, their position gets a lot

Coming up next