提醒创始会员：我们与 Strength In Numbers 和 The Downballot 举行的私人大选前简报会将于本周二（10月27日）美国东部时间下午6:30举行。详情请见此处。大选之夜（11月3日周二），我将与 David Nir 联合直播，预计美国东部时间晚上7:00左右开始。
A reminder for Founding Members: our private pre-election briefing with Strength In Numbers and The Downballot is this Tuesday, October 27, at 6:30 p.m. ET. Details here. And I’ll be live on election night, Tuesday, November 3, with David Nir starting around 7:00 p.m. ET.
本周 Strength In Numbers 及相关项目事务繁忙，下周也将如此。因此我写一篇相对非正式的主贴，下方的其他链接列表也比往常短。
It was a very busy week at Strength In Numbers and associated projects this week, and this week will be busy too. So I’m writing a more informal main post here, and then have a shorter-than-usual list of other links below.
顺便提醒一下，周日综述的评论区对所有 SIN 读者开放，欢迎大家在那里分享自己的链接给社区其他成员。
As a reminder, comments on the Sunday roundups are unlocked for all readers of SIN, so feel free to share your own links with the rest of the community there.
1a. 未决定选民 我昨天上午上了一档电视节目，主持人想了解未决定选民的情况。有多少人？我们能从中推断出他们可能怎么投票吗？
1a. Undecided voters I did a TV hit yesterday morning in which the hosts wanted to know about undecided voters. How many are there, and is there anything we can glean about their likely vote?
为回答这个问题，我汇总了最近四次 Strength In Numbers/Verasight 民调的数据，获得了一个包含391名登记选民的小样本，这些选民在被问及倾向后，在通用选票问题上仍表示“不知道”投给谁。
To answer this, I combined data from the last four Strength In Numbers/Verasight polls to get a small sample of 391 registered voters who still say they “don’t know” who they’ll vote for on the generic ballot after we ask which way they lean.
结果主持人最终没问我关于未决定选民的事，所以我把发现分享给大家：首先，未决定选民相对较少。视民调而定，在被追问后，约有5%-10%的选民仍未决定。这个比例小于传统出口民调中表示在最后一周才决定投给谁的选民比例（2022年和2024年约为15%-20%），但由于这些民调在选举日抽样方式的复杂原因，那很可能是高估。
Well, the TV hosts didn’t end up asking me about undecided voters, so I’m going to share the findings here with you all now instead: First, relatively few voters are undecided. Depending on the poll, about 5-10% of voters are undecided after being pushed. This number is smaller than the percent of voters in traditional exit polls that say they decided who to vote for in the last week (about 15-20% in 2022 and 2024), but for complex reasons having to do with how those polls sample people on Election Day, that’s likely an overestimate.
2025年，弗吉尼亚州仅有约5%的选民在 The Voter Poll —— 由媒体机构合作伙伴、SSRS 和民调机构 Verasight 推出的一种新型传统出口民调，严重依赖选前访谈 —— 中表示自己未决定。
In 2025, only about 5% of voters in Virginia told The Voter Poll — a new version of the traditional exit poll from partners at media organizations, SSRS, and the pollster Verasight that relies heavily on pre-election interviews — that they were undecided.
摇摆选民比已定选民更关注物价。根据我们的综合民调数据，45%的摇摆选民将物价和通胀列为国家面临的最重要问题，另有18%选择就业和经济。合计63%的摇摆选民提及物价或经济，而表示已决定投票给谁以及是否投票的选民中，这一比例为48%。
Undecided voters also care more about prices than decided voters. According to our combined polling data, 45% of undecided voters name prices and inflation as the most important problem facing the country, and another 18% pick jobs and the economy. That’s 63% naming prices or the economy, compared with 48% of voters who say they’ve already decided for whom and whether to vote.
在几乎所有议题的相对优先次序上，摇摆选民更像是已决定投票给民主党人的选民，而非表示将投票给共和党人的选民。
On the relative prioritization of almost every issue, undecided voters look more like voters who have already decided to vote for Democrats than those who say they’ll vote Republican.
他们也非常不满特朗普对物价的处理。如果摇摆选民刚开始关注，且将物价列为首要优先事项，那么他们对现任总统在物价方面的表现如何看待，很可能成为决定性因素。总体而言，69%的摇摆选民不赞成特朗普的整体表现，仅14%表示赞成。在物价问题上，不赞成与赞成的比例为77%对8%。若排除表示不知如何评价特朗普的人，不足十分之一的摇摆选民赞成他对通胀的处理。
They also really don’t like how Trump is handling prices. If undecided voters are just tuning in, and if they rank prices as a top priority, then how they feel about the incumbent’s performance on prices is likely to be decisive. Overall, 69% of undecided voters disapprove of the job Trump is doing in general, and just 14% approve. On prices, they are split 77% disapprove to 8% approve. If you are ignoring people who say they don’t know how they feel about Trump, fewer than 1 in 10 undecideds approve of his handling of inflation.
最后，他们中许多人是感到失望的特朗普选民。据Verasight在受访者加入调查社区时记录的数据，约6成摇摆选民在2024年投票过。在投票给特朗普或卡马拉·哈里斯的人中，特朗普选民数量几乎是哈里斯选民的2倍。总体而言，28%的摇摆选民投了特朗普，15%投了哈里斯。其余的人未投票（38%）或支持第三方候选人（18%）。
Finally, many of them are disappointed Trump voters. About 6 in 10 undecideds voted in 2024, according to the data Verasight recorded for its panelists when they joined the survey community. Among those who voted for Trump or Kamala Harris, Trump voters outnumber Harris voters nearly 2 to 1. Altogether, 28% of undecideds voted for Trump and 15% for Harris. The rest didn’t vote (38%) or backed a third-party candidate (18%).
这些摇摆中的特朗普选民已背弃他。他们给他的净支持率为-31，比注册选民整体（-25）略差。在物价问题上，他的净支持率为-59。
Those undecided Trump voters have turned on him. They give him a net approval rating of −31, a bit worse than registered voters overall (−25). On prices, he’s at −59.
很大一部分摇摆选民是怨愤的前2024年特朗普选民。其中许多人是年轻人，他们以惊人的数量转而反对共和党。
A large portion of undecided voters are aggrieved former 2024 Trump voters. Many of these are young people who have turned against Republicans in incredible numbers.
这与我周五的发现吻合，当时我写道，帮助特朗普在 2024 年当选的低信息量选民，现在转而最强烈地反对他。摇摆选民中低信息量选民占比不成比例地高；10 月份，63% 的摇摆选民在关于哪个党派控制国会的两个基本问题中，至少答错了一道，而已选边站的选民中，这一比例不足四分之一。
This fits with what I found on Friday, when I wrote that the low-information voters who helped elect Trump in 2024 have turned hardest against him. Undecideds are disproportionately low-information voters; in October, 63% of them missed at least one of two basic questions about which party controls Congress, compared with fewer than 1 in 4 voters who’d picked a side.
这意味着，随着摇摆选民开始关注大选，他们最可能对两类信息做出反应：物价和“氛围感”。而他们从特朗普和共和党那里听到的收官信息是什么？除了他对参议员乔恩·休斯特臀部的古怪执着外，几乎没有实质内容——尤其是很少谈论物价。特朗普最近的大多数演讲，都集中攻击各位民主党提名人的所谓意识形态极端性；在不攻击的时候，他则对密歇根州民主党参议员提名人阿卜杜勒·埃尔-赛义德发表伊斯兰恐惧症言论，称其为“阿卜杜拉拉拉拉拉巴巴·穆罕默德·埃尔-赛义德”。这或许能讨好基本盘，但不是摇摆选民关心的。
This means undecided voters are most likely reacting to two pieces of information as they tune in: prices, and “vibes.”And what is the closing message they are hearing from Trump and the Republican Party? Aside from his weird fixation on Senator Jon Husted’s rear end, not a whole lot — and especially not a lot about prices. Most of the president’s recent speeches focus on the alleged ideological extremity of various Democratic nominees, when he’s not being Islamophobic against Democratic Senate nominee in Michigan Abdul El-Sayed, whom he calls ”Abdulalalalallahdbaba Mohammed El-Sayed.”That may work with his base, but it’s not what undecided voters care about.
不仅仅是特朗普。《纽约时报》发现，最近共和党的广播和流媒体广告中，只有极小一部分试图为他的总统任期说好话。《华尔街日报》发现，民主党正猛烈抨击特朗普和共和党的关税政策。派凯文·哈塞特这位“笑面虎”出来说明高油价，于事无补。
And it’s not just Trump. The New York Times found only a small fraction of recent GOP broadcast and streaming are even trying to go positive on his presidency. The Wall Street Journal found Democrats are pummeling Trump and Republicans over his tariffs. It’s not like sending Smiley Kevin Hassett out to explain high gas prices helps here.
从某种意义上说，大局已定。若还有变数，只能来自那些尚未决定投给谁的、反复无常的低信息量选民。我们的调查显示，这些选民主要对现任政府表现的总体信息做出反应。这……对共和党今年来说，绝非好兆头：我正在撰写另一篇关于提前投票的博客更新，将纳入更多州的数据。正如我上周一所写，当提前投票参与率真正升高时，距离选举还有一周，我们从提前投票分析中获得的信号会急剧增强。
So, in a sense, the cake is mostly baked. To the extent things can change, it’s because of volatile low-information voters who have yet to decide who to vote for. Our survey shows these voters respond mostly to general information about performance of the incumbent. That’s… not a good bet for Republicans this year: 1b. The early vote I’m working on another update to the early vote blog with data from more states. As I wrote last Monday, the signal we get from EV analysis increases sharply at a week out, when uptake is really high.
今年将有远超半数的美国人，甚至可能接近 70% 选择提前投票（无论是现场还是邮寄）。随着分母（提前投票总人数）增大，分子（各党派已投票人数）变得更有意义。
Well over half of Americans, and maybe verging on 70%, will vote early (either in-person or by mail) this year. As the denominator (number of people voting early) grows, the numerator (number of people in each party that have voted) becomes more meaningful.
过去一个月我收集了相当多的提前投票现时和历史数据。我将通过下方链接公开其中部分数据，随后简要通报目前的情况。
I’ve gathered quite a bit of current and historical data on early voting over the past month. I will make some of this data available via the link below, and then give a quick update on where things stand.