MS NOW 正在自行预测哪个政党将在 2026 年美国中期选举中赢得众议院控制权。其做法是估算民主党和共和党在所有 435 个众议院选区结果全部揭晓后，各自将至少获得多少个席位，即“底线席位”。
MS NOW is making its own projection of which party will win control of the U.S. House in the 2026 midterm elections. It will do so by estimating a minimum number of House seats, or “floor,” Democrats and Republicans will reach after all 435 House races have been called.
一旦某党达到至少 218 个席位——即过半数门槛——MS NOW 将预测该党赢得众议院多数派地位。随着更多数据陆续汇总，双方的数字将持续上升。
MS NOW will project a winner of the House majority after one party reaches at least 218 seats — the majority threshold. The numbers for either party will continue ticking up as more data trickles in.
我们为何这样做？我们旨在缩小 MS NOW 所知与您——我们的受众——所听、所见、所读之间的差距。在幕后，我们的选举专家将在美联社——我们的选举数据提供方——预测每个众议院选区结果之前，就已确定各党的“底线席位”。
Why are we doing this? We are seeking to close the gap between what MS NOW knows and what you — our audience — hear, see and read. Behind the scenes, our election experts will determine each party’s “floor” well before The Associated Press — our election data provider — projects the outcome of every individual House contest.
为此，我们的团队将在选举夜分析趋势，以估算全国不同类型的众议院选区呈现何种走势。我们的目标是透明，无论是我们报告的数字，还是生成这些数字的方法。我们希望您能实时掌握与我们相同的信息。因此，MS NOW 众议院预估将按以下方式运作。
To do so, our team will analyze trends on election night to estimate how different types of House districts across the country are breaking. Our goal is transparency, both in the numbers we report and the methodology we use to produce them. We want you to have the same information as we do in real time. So here’s how the MS NOW House estimate will work.
我们的预估将依赖投票结束、实际得票开始陆续汇总后出现的两大趋势。
Our estimate will rely on two major trends that emerge once polls start closing and we begin seeing the actual vote count trickle in.
第一个趋势是相较上一次选举的摆动幅度：即两党得票率差距相较 2024 年总统大选结果的百分点变化。这是一个良好的基准，不仅因为我们将当前结果与最近一次重大选举进行比较，还因为这一差值可在每一个众议院选区计算得出，包括那些近期重新划分过的选区。
The first trend is the swing from the last election: that is, the percentage-point shift in the two-party vote margin from the 2024 presidential results. This is a good baseline, not only because we’re comparing the current results to the most recent major election, but also because that difference is something we can calculate in every single House district, including those that were recently redrawn.
以北卡罗来纳州第1国会选区为例，该选区重新划分后，唐纳德·特朗普总统在2024年大选中本将以12个百分点的优势赢得该选区。如果我们假设共和党人劳里·巴克豪特今年仅以2个百分点的优势击败现任民主党众议员唐·戴维斯，那么我们估算的选情变化将为10个百分点。
Take North Carolina’s 1st Congressional District, for instance, which was redrawn such that President Donald Trump would have carried it by 12 percentage points in 2024. If we hypothetically saw Republican Laurie Buckhout beating incumbent Democratic Rep. Don Davis by only 2 points this year, our estimated shift would be 10 points.
多个选举周期的研究表明，选区倾向于呈现一致性变化。以2018年为例，与2016年总统大选相比，平均得票率变化约为向民主党倾斜5个百分点。绝大多数众议院选区都朝该方向变化，民主党最终赢得了众议院多数席位。这并不意味着所有选区的变化幅度都相同，甚至有些选区可能朝相反方向变化。但党派得票率的平均变化是一个很好的指标，能够概括中期选举的整体表现，并与众议院席位获取情况高度相关。
Research over several election cycles shows that districts tend to shift uniformly. In 2018, for example, the average shift in vote margin from the 2016 presidential contest was about five points toward Democrats. The vast majority of House districts saw shifts in that direction, and Democrats ended up winning the House majority. That does not mean all districts shift the same amount, and some may even shift in the other direction. But the average shift in partisan vote margin is a good metric that summarizes overall midterm performance and correlates well with House seat pickups.
我们将考察的第二大趋势是，实际投票结束后，不同类型的众议院竞选呈现何种走向。我们在选举日前认定为“势均力敌”的竞选，是否不成比例地向某一方倾斜？这通常是“浪潮式”选举年的典型特征。民主党是否赢得了一些我们在选举日前认为倾向共和党的席位？还是说共和党现任议员在这些选区守住了阵地？
The second major trend we will examine is how different types of House races are breaking once the actual votes are cast. Are the races we identified as “toss-ups” before Election Day breaking disproportionately for one party? That’s typically what happens in “wave” years. Are Democrats picking up some seats we viewed as leaning toward Republicans before Election Day, or are GOP incumbents holding on in these districts?
即使我们没有所有选区的最终数据，现有数据也将为我们提供一个窗口，让我们得以窥见如何预期类似选区在当晚晚些时候的走向。逐类别分析并估算各党的胜率，将使我们能够对尚未宣布结果的竞选做出明智的预估。
Even if we do not have final data from every district, the data we do have will give us a window into how we might expect similar districts to trend later in the night. Going category by category and estimating each party’s success rate will allow us to make informed estimates about races that have not yet been called.
最后，我们将综合所有信息。统计建模将使我们能够整合这些信号，将我们在早期结果中看到的情况转化为各党在众议院的总席位数。
Finally, we will bring it all together. Statistical modeling will enable us to combine these signals to translate what we’re seeing in the earlier results into aggregate seat totals for each party in the House.
虽然如果个别选情胶着，宣布结果可能需要数天，但这种建模将使我们能够在尚不知晓每一场选举结果之前，估算出各党的底线席位。对于单个席位的胶着竞争，其不确定性大于对总席位数的预测。重要的是，我们仅在相关州的投票结束、投票站关闭后，才会使用这些众议院选区的数据。
While individual races may take several days to get called if they are close, this modeling will allow us to estimate each party’s floor before we know the outcome of every single race. There is more uncertainty about a close race for an individual seat than there is for aggregate seat numbers. And importantly, we only will use data for House districts once polls in those states are closed and voting has ended in them.
我们预计，在选举之夜，随着模型对各方表现的把握增强，各党最低众议院席位总数的估算值将逐步上升。
We expect the estimate for each party’s minimum House seat total to tick up incrementally on election night, as our model becomes more confident about each side’s performance.