在唐纳德·特朗普于林肯纪念堂倒影池举行的奇怪的大选前活动中，这位总统说了各种奇怪的话，尽管他确实分享了一个关于近期历史的相当准确的描述。
During Donald Trump’s bizarre pre-election event at the Lincoln Memorial Reflecting Pool, the president said all sorts of strange things, though he did share a reasonably accurate description about recent history.
“通常，无论现任总统是谁，共和党人还是民主党人，好总统还是坏总统，伟大的总统还是糟糕的总统，出于某种原因，现任总统在中期选举中不会获胜，”这位在任者说道。他迅速补充道：“但我们已经全力以赴了。”这八个字的补充暗示特朗普对今年选举周期的结果并不抱太大希望，尽管他关于历史趋势的观点在很大程度上是正确的：尽管有显著的例外，现任总统所属政党在中期选举中往往会遭受损失。这让我开始思考：平均损失是多少？
“Usually, whoever the sitting president is, Republican or Democrat, good or bad, good president, great president, bad, for some reason the sitting president doesn’t win the midterms,” the incumbent said. He quickly added, “But we gave this a very strong shot.”That eight-word addendum suggested that Trump did not have high hopes about the outcome of this year’s cycle, though his point about historical trends was largely true: Though there are notable exceptions, the incumbent president’s party tends to suffer losses in midterm elections. And that got me thinking: What’s the average?
当我开始整理选举数据时，这似乎是一个相当简单的练习。我的计划只是统计结果并进行算术运算，得出现任总统所属政党在众议院和参议院平均失去的席位数。
When I started putting the election data together, it seemed like a fairly straightforward exercise. My plan was to just count up the results and do the arithmetic, coming up with an average number of losses in the House and Senate for the incumbent president’s party.
但我越深入挖掘数据，情况就变得越复杂，因为涉及一些可能并不显而易见的问题。应该以哪届中期选举作为起点？是最好从1914年周期开始，因为那是美国众议院增至435名议员的时候，还是从1958年周期开始，因为那是美国参议院增至100名议员的时候？
But the more I dug into the data, the more complicated it became because of questions that might not seem immediately obvious. Which midterms should be used as a starting point? Is it best to start with the 1914 cycle, because that’s when the U.S. House grew to 435 members, or with the 1958 cycle, because that’s when the U.S. Senate grew to 100 members?
将重点局限于现代政治时代很诱人，但对于现代政治时代何时开始并没有明确的共识。第二次世界大战？水门事件？21世纪初？
It’s tempting to limit the focus to the modern political era, but there’s no definitive consensus on when that began. World War II? Watergate? The start of the 21st century?
总统的第一个中期选举周期与第二个周期之间的差异又该如何看待？
What about the difference between a president’s first midterm cycle versus his second?
换句话说，审视同一组统计数据有多种方式，但以下是我发现的概况。
There are, in other words, a variety of ways to look at the same statistics, but here’s an overview of what I found.
如果从20世纪初（1902年周期）开始计算，现任总统所属政党平均失去约31个众议院席位和4个参议院席位。
If we start with the turn of the 20th century (the 1902 cycle), the incumbent president’s party loses an average of roughly 31 House seats and four Senate seats.
如果从 1914 年的周期开始计算，数据基本相同：平均损失 30 个众议院席位和 3.5 个参议院席位。
If we start with the 1914 cycle, the numbers are largely the same: a loss of 30 House seats and 3.5 Senate seats.
如果从 1946 年的周期（二战后首个周期）开始计算，在任总统所属政党平均损失约 26 个众议院席位和近 4 个参议院席位。
If we start with the 1946 cycle (the first following World War II), the incumbent president’s party loses an average of roughly 26 House seats and nearly four Senate seats.
如果从 1958 年的周期开始计算，数字略有缩减：损失 24 个众议院席位和 3 个参议院席位。
If we start with the 1958 cycle, the numbers shrink a little: a loss of 24 House seats and three Senate seats.
如果从 1974 年的周期（水门事件后尼克松辞职后的首个周期）开始计算，在任总统所属政党平均损失 23 个众议院席位和约 3 个参议院席位。
If we start with the 1974 cycle (the first following Richard Nixon’s resignation after the Watergate scandal), the incumbent president’s party loses an average of 23 House seats and roughly three Senate seats.
如果仅关注 21 世纪的中期选举，数据非常相似：损失近 25 个众议院席位和约 3 个参议院席位。
If we limit the focus to 21st-century midterms, the numbers are very similar: a loss of nearly 25 House seats and roughly three Senate seats.
至于总统首次和第二次中期选举的差异，自 20 世纪以来，在任总统所属政党在首次中期选举的众议院竞选中表现往往更差（首次损失近 33 个席位，第二次损失近 26 个），但在参议院竞选中则相反（首次损失近 3 个席位，第二次损失近 5 个）。2026 年的周期将如何与这些平均值对应？敬请关注。
As for the differences between presidents’ first and second midterms, since the start of the 20th century, the incumbent president’s party tends to fare worse in House races in the first midterms (nearly 33 losses in the first vs. nearly 26 in the second), but in Senate races, it’s the opposite (nearly three losses in the first midterms versus nearly five in the second). How will the 2026 cycle match up to these averages? Watch this space.