当不可阻挡的力量遭遇不可移动的物体，二者必有一让。唐纳德·特朗普第二任期的大部分时间里，评论员们都认为，那个物体——共和党看似固若金汤的参议院版图——占据优势，压倒了那股力量——民主党在全国民调中的领先。去年，知名预测机构“萨巴托水晶球”称共和党是“强有力的热门”，可守住参议院多数派地位，并将阿拉斯加、爱荷华和堪萨斯三州列为“稳固”红州席位。直到今年6月，该机构仍认为有足够多的竞选“至少倾向”共和党，足以让该党控制参议院。
WHEN AN UNSTOPPABLE force meets an immovable object, one or the other has to give. For most of Donald Trump’s second term, pundits have assumed that the object, Republicans’ seemingly impregnable Senate map, had the edge over the force, Democrats’ lead in national polls. Last year Sabato’s Crystal Ball, a leading forecasting outfit, called Republicans “strong favourites” to hold their Senate majority, and classified Alaska, Iowa and Kansas as “safe” red seats. As late as this June, the group still found enough races at least “leaning” towards Republicans for the party to control the chamber.
如今，距离中期选举不足一周，通行的看法认为那股力量已占据上风。“水晶球”最新更新称，民主党“在参议院争夺战中略占优势”，而博彩市场看好民主党以64%的概率翻盘控制参议院。美国最知名的量化选举预测专家内特·西尔弗给出民主党76%的胜算。但《经济学人》自有的预测模型更为审慎，认为参议院易手的概率仅为53%。
Now, with midterm elections less than a week away, conventional wisdom holds that the force has claimed the upper hand. Crystal Ball’s latest update says Democrats “have a slight advantage in the battle for the Senate”, while betting markets make the party a 64% favourite to flip the chamber. Nate Silver, America’s most prominent quantitative election forecaster, gives Democrats a 76% chance. But The Economist’s own prediction model is more cautious, putting the probability of the Senate changing hands at just 53%.
这一分歧的主要源头，在于我们的模型有意打破了一个常见的预测准则：切勿试图预测民调误差的方向。尽管“通用选票”民调——询问受访者计划支持哪个党派竞选国会——拥有令人印象深刻的准确记录，但它们也有显著的高估倾向：波浪式选举的实际规模通常不如宣称的那样巨大。例如，在2006年中期选举前夕，民主党在此类民调中平均领先13个百分点，而其在全国众议院选票中的最终领先优势仅为7个百分点。而且，当全国民调失准时，这种误差通常也会出现在州级民调中。
The main source of this disagreement is our model’s intentional breach of a common forecasting maxim: thou shalt not attempt to predict the direction of polling errors. Although “generic-ballot” surveys, which ask respondents which party they plan to support for Congress, have an impressive record of accuracy, they also have a marked tendency to overshoot: wave elections usually prove less towering than advertised. On the eve of the 2006 midterms, for example, Democrats led these surveys by 13 percentage points on average, while their final margin in the national House vote was a mere seven. And when national polls misfire, that error usually appears in state-specific surveys as well.
今年民主党在通用选票上的优势约为 8.5 个百分点，不如 2006 年那样极端。尽管如此，根据历史经验，我们的模型预测全国民调将高估该党优势约 1 个百分点。这一差距在历史上倾向民主党的州较小，但在倾向共和党的州较大——而要控制参议院，民主党需要在特朗普 2024 年以两位数优势获胜的州中至少赢下两个。
This year’s Democratic advantage, at around 8.5 points on the generic ballot, is less extreme than that of 2006. Nonetheless, based on history, our model projects that polls across the country will overestimate the party’s margin by around a point. This gap is smaller in historically Democratic states, but larger in Republican ones—and to control the Senate, Democrats need to win at least two states that Mr Trump won by double digits in 2024.
另一个潜在因素是筹款。总体而言，民主党参议院候选人从其所在州的个人捐赠者（捐款上限为 3500 美元）那里获得的捐款远多于其对手。许多预测模型对此权重很高。然而，自特朗普首次就职以来的每一次选举中，民主党参议院提名人作为一个群体在这一类别上的筹款额也都超过了共和党人。这并未转化为超越民调预期的选举表现。
Another potential factor is fundraising. Overall, Democratic Senate candidates have received far more in contributions from individual donors (which are capped at $3,500) in their states than their opponents have. Many forecasts weight this heavily. However, Democratic Senate nominees as a group have also out-raised Republicans in this category in every other election since Mr Trump’s first inauguration. This has not translated into poll-beating electoral performances.
因此，我们的模型将每位候选人的筹款额与其所在党派的平均水平进行比较。这有助于改善共和党在爱荷华等州的前景，因为我们奖励其提名人取得了微弱的财务优势。
As a result, our model compares each candidate’s fundraising to their own party’s average. This helps Republicans’ prospects in states like Iowa, where we reward their nominee for eking out a narrow financial advantage.
相较于共识预期，我们的模型对民主党在阿拉斯加的前景最为悲观。博彩市场认为玛丽·佩尔托拉以 73% 的概率击败共和党现任议员丹·沙利文，而我们的模型显示沙利文以 51 比 49 领先。此外，佩尔托拉还可能因其团队被指控在选票上安插搅局候选人而受损，这一争议极有可能主导竞选最后一周的舆论。
Relative to the consensus, our model is most pessimistic about Democratic hopes in Alaska. Betting markets make Mary Peltola a 73% favourite over Dan Sullivan, the Republican incumbent, whereas we show a 51-49 race favouring Mr Sullivan. And Ms Peltola may yet suffer from allegations that her team planted spoiler candidates on the ballot, a controversy that is likely to dominate the final week of campaigning.
尽管如此，我们的模型仍然明确显示民主党在足够多的州——包括缅因州、密歇根州、北卡罗来纳州和俄亥俄州——处于领先地位，足以赢得50个参议院席位。模型显示，民主党在另外两个州（阿拉斯加和得克萨斯州）与共和党势均力敌，这两个州可能产生决定多数党地位的第51个席位。重要的是，我们的模型给民主党控制众议院的概率为94%，这将使他们能够否决特朗普的议程，并对其政府发起调查。赢得参议院将使他们也能阻挠他的司法提名人。
Nonetheless, our model still clearly favours Democrats in enough states—including Maine, Michigan, North Carolina and Ohio—to win 50 Senate seats. It gives them even odds in two more (Alaska and Texas) that could be the majority-making 51st seat. Importantly, our model gives Democrats a 94% chance of gaining control of the House of Representatives, which would allow them to reject Mr Trump’s agenda and to launch investigations of his administration. Winning the Senate would enable them to block his judicial nominees as well.
参议院的组成（任期六年）也将强烈影响特朗普的继任者（将在两年后选出）能否实施党派议程。展望2028年以后，民主党的第52个席位可能与第51个席位同样重要。而且他们目前有一条合理的路径，可以赢得远多于52个席位。在堪萨斯州、爱荷华州甚至南卡罗来纳州的民调中，他们的候选人与共和党候选人不相上下。
The makeup of the Senate, where terms are six years, will also strongly influence whether Mr Trump’s successor, who will be elected in two, can implement a partisan agenda. Looking beyond 2028, Democrats’ 52nd seat could matter just as much as their 51st. And they currently have a plausible path to winning considerably more than 52 seats. Their candidates are neck-and-neck in polls of Kansas, Iowa and even South Carolina, where Lindsey Graham, a longtime Republican incumbent, died in July and was replaced by his sister Darline, a political novice.
在南卡罗来纳州，长期担任共和党参议员的林赛·格雷厄姆于7月去世，由其妹妹达琳接任，后者是政治新手。10月28日，美国最知名的民调团队《纽约时报》和锡耶纳学院报告称，南卡罗来纳州的竞选呈现平局，而两年前特朗普在该州以18个百分点的优势获胜。
On October 28th the New York Times and Siena College, America’s most prominent team of pollsters, reported a tied race in South Carolina, where Mr Trump won by 18 points two years ago.
民主党在所有这些州中略处劣势。然而，如果这些多米诺骨牌中哪怕倒下一张，他们很可能受益于全国性的民调误差，从而增加其他州也倒向他们的可能性。假设内布拉斯加州的独立候选人丹·奥斯本如果当选将加入民主党党团会议，我们给民主党至少占据55个席位的概率为15%。即使在2006年，当时通用选票民调大幅高估了民主党的优势，该党仍设法翻转了六个参议院席位并确保了多数党地位。■
Democrats are modest underdogs in all these states. However, if even one of these dominoes falls, they may well have benefited from a nationwide polling error, increasing the chances that the others fall too. Assuming that Dan Osborn, an independent in Nebraska, would caucus with them if elected, we give Democrats a 15% chance of holding at least 55 seats. Even in 2006, when the generic ballot vastly overstated the Democratic advantage, the party still managed to flip six Senate seats and secure a majority. ■