根据538的最新选举预测副总统卡玛拉·哈里斯赢得11月总统大选的几率为57/100，而前总统川普的几率为42/100。这意味着这场比赛本质上是一场胜负难分的比赛;基于投票的接近性和非投票因素的不确定性，我们称之为“基本原则，“两位候选人都可能在11月轻松获胜，如果民调明显偏离目标，可能会以相当大的优势获胜(又).

　　与几天前相比，哈里斯对特朗普的优势略有下降。周日，她有62/100的机会赢得选举，我们现在预测她将赢得全国普选3.2个百分点，而不是4个百分点。下降的主要原因是纽约时报/锡耶纳学院周一早上公布的几项对特朗普有利的民调。这些民调显示，特朗普在北卡罗来纳州领先哈里斯2个百分点(我们的投票平均数是川普+0.4)，在佐治亚州被3(川普+1.3平均)和亚利桑那州的5(特朗普+1在538的平均值中).最后一次调查对特朗普来说尤其有利，因为民意调查机构在亚利桑那州的最后一次调查中，哈里斯领先5个百分点，领先特朗普10个百分点。

　　然而，我们的模型对特朗普取得如此大的收益表示怀疑。这是因为，尽管《纽约时报》/锡耶纳学院为特朗普带来了好消息，但其他民调机构也一直在发布哈里斯的正面数据——比如周日发布的NBC News的哈里斯+5全国调查，以及MassINC Polling Group周一发布的威斯康星州哈里斯+6民调。因为我们的模型认为哈里斯和特朗普的支持率在各州之间是相关的，所以当一位民意调查者显示出其他地方没有出现的大幅度变动时，它就会打折扣。但是不要仅仅相信我们的话;《纽约时报》的内特·科恩暗示他也持类似的怀疑态度，星期一写作“在这种情况下，我们的经验法则是将结果转化为投票平均值。”

　　现在的底线是选举仍然非常接近。我们的模型目前预测，特朗普和哈里斯之间的差距将在四个州(北卡罗来纳州、乔治亚州、亚利桑那州和内华达州)不到1个百分点，在另外四个州(宾夕法尼亚州、威斯康星州、密歇根州和佛罗里达州)不到4个百分点(正常民调误差的大小)。

　　但是选举不仅仅是在单个州的层面上。在选举团中，这也可能是历史上最接近的。如上表所示，目前哈里斯目前领先的北方战场州和特朗普目前领先的大多数阳光地带州之间存在分歧。这种分裂使得选举团获胜的可能性特别大。

　　例如，如果你假设预测完全正确，并将所有州和国会选区分配给目前领先的候选人，然后你将内华达州向特朗普移动0.8个百分点，哈里斯将赢得270张选举人票，特朗普将赢得268张选举人票。这将是自1876年以来总统选举中最小的选举人票差额，当时共和党人卢瑟福·b·海斯以一张选举人票的优势击败了民主党人塞缪尔·J·蒂尔顿。即使在2000年，共和党人乔治·w·布什也以271票对267票的相对优势击败了民主党人阿尔·戈尔。

　　如果过去的选举有任何指导意义的话，那可能是选举日之后几周混乱的原因。2000年，选举结果在几周内都不明朗，因为佛罗里达州是当年的决定性州多次重新计票。1876年选举结果有争议直到就职日的前几天，一些州向国会提交了相互矛盾的申报表。在2000年，有抗议;在1876年，有暴力威胁。在2024年，在如此激烈的竞争中失败的候选人可能会在一个或多个州参加竞选，这反过来可能会引起选民的不安。例如，特朗普已经对邮寄选票的真实性提出质疑有疑问，共和党人已经开始起诉州政府他们声称将允许非公民在选举中投票。特朗普此前曾拒绝排除政治暴力如果他不相信选举是公平的，他的支持者。

　　尽管民调显示今天的竞选势均力敌，但选举也是“势均力敌”的，因为微小的投票失误或国家环境的微小变化都可能导致一名候选人以相当大的优势获胜。换句话说，结果的不确定性让事情看起来“很接近”概率上(从某种意义上来说，每个候选人都有平等的机会获胜)但这并不意味着大获全胜是不可能的。例如，根据我们的预测，有六分之一的机会其中一名候选人以超过10个百分点的优势赢得普选。

　　我们将关注哈里斯和特朗普在未来几周的预测中的差距变化，以衡量大选之夜的竞选结果。回来看看我们的投票平均数和预报随着比赛进入最后六周!

　　This could be the closest presidential election since 1876

　　According to 538'slatest election forecast, Vice President Kamala Harris has a 57-in-100 chance of winning the presidential election in November, while former President Donald Trump has a 42-in-100 chance. That means the race is essentially a toss-up; based on the closeness of the polls and the equivocacy of non-polling factors we call the "fundamentals," either candidate could easily win in November, and perhaps by a somewhat sizable margin if the polls significantly miss the mark (again).

　　Harris's advantage over Trump is slightly reduced from a couple of days ago. On Sunday, she had a 62-in-100 chance of winning the election, and we're now forecasting that she will win the national popular vote by 3.2 percentage points rather than by 4 points. The main reason for the decline is several good polls for Trump from The New York Times/Siena College published Monday morning. Those polls had Trump leading Harris by 2 points in North Carolina (ourpolling averageis Trump+0.4), by 3 in Georgia (Trump+1.3on average) and by 5 in Arizona (Trump+1in 538's average). That last survey was a particular boon for Trump, given the pollster's last survey in Arizona had Harris up by 5 — a 10-point swing toward Trump.

　　However, our model is skeptical that Trump has made gains this large. That's because, while The New York Times/Siena College had good news for Trump, other pollsters have been releasing positive numbers for Harris — such as a Harris+5 national survey from NBC News released Sunday and a Harris+6 poll of Wisconsin released by MassINC Polling Group on Monday. Because our model sees support for Harris and Trump as correlated across states, it discounts when one pollster shows large movements that don't appear elsewhere. But don't just take our word for it; The New York Times's Nate Cohn suggested he is similarly skeptical,writing on Monday, "In circumstances like this, our rule of thumb is to toss the results into a polling average."

　　The bottom line right now is that the election remains incredibly close. Our model is currently forecasting that the margin between Trump and Harris will be less than 1 point in four states (North Carolina, Georgia, Arizona and Nevada), and within 4 points (the size of a normal polling error) in four more (Pennsylvania, Wisconsin, Michigan and Florida).

　　But the election isn't just close on the individual state level. It could also be historically close in the Electoral College. As the table above shows, there is currently a split between the northern battleground states, where Harris is currently leading, and most Sun Belt states, where Trump is currently leading. And this split makes especially tight Electoral College victories more likely.

　　For example, if you assume the forecast will be exactly correct and assign all states and congressional districts to the candidate who currently leads in them, and then you move Nevada just 0.8 percentage points toward Trump, Harris would win 270 electoral votes and Trump would win 268. That would be the smallest electoral-vote margin in a presidential election since 1876, when Republican Rutherford B. Hayes beat Democrat Samuel J. Tilden by one electoral vote. Even in 2000, Republican George W. Bush defeated Democrat Al Gore by the comparably comfortable margin of 271 to 267.

　　If those past elections are any guide, that could be a recipe for chaos in the weeks after Election Day. In 2000, the outcome of the election was murky for weeks as Florida, the decisive state that year, conductedmultiple recounts. In 1876, theelection result was disputeduntil just days before Inauguration Day, as some states sent conflicting sets of returns to Congress. In 2000, there wereprotests; in 1876, there werethreats of violence. Here in 2024, it's possible that the losing candidate in such a tight race could contest the election in one or more states, which could in turn cause unrest among voters. Trump, for instance, has already called the veracity of mail-in ballotsinto question, and Republicans have begunsuing state governmentsover registration processes they allege will let noncitizens vote in the election. And Trump has previouslydeclined to rule out political violencefrom his supporters if he doesn't believe the election is fair.

　　But while the polls point to a close race today, the election is also "close" in the sense that small polling misses or slight changes in the national environment could lead to one candidate winning by a comfortable margin. The uncertainty about the outcome, in other words, makes things look "close"probabilistically(in the sense that each candidate has about an equal chance to win) but does not mean that big wins aren't also possible. According to our forecast, for example,there is a 1-in-6 chancethat one of the candidates wins the popular vote by more than 10 points.

　　We'll be watching how Harris's and Trump's margins change in our forecast over the next few weeks in order to gauge the closeness of the race going into election night. Check back in on ourpolling averagesandforecastas the race heads into its final six weeks!