Yes, you read that right. Donald Trump’s rapid fall in the polls so far hasn’t had a negative impact on Republicans running for the U.S. Senate. Far from it actually, as Harry Enten of FiveThirtyEight explains:

Clinton’s pre-Democratic-convention swoon was matched by Democrats doing worse in the polls. But Democrats rebounded as Clinton did after the convention. Then their odds all fell together in the second half of August and through September. Clinton’s chances began to rise again after the first debate, but unlike after the convention, Democrats’ chances of taking back the Senate haven’t followed Clinton’s presidential odds upward. To put this in mathematical terms, the correlation between the Democratic chances in the Senate and Clinton’s chances was a very high +0.87 in the 74 days before the first debate. In the 16 days since, it’s been -.23, indicating that they’re moving in opposite directions.

We can see this by looking at the FiveThirtyEight polls-only forecast on the date of the debate and on Oct. 12 for both the Senate and the presidency in the 26 states where at least one poll was conducted before the first debate.

Clinton has improved her margin over Trump by an average of 5.1 percentage points in these 26 states, while the average Democratic candidate’s margin is down 1.5 percentage points. And while Clinton has improved her margin by at least 3.5 percentage points in every one of these 26 states, Republican Senate candidates have improved their chances in all but four: The Democratic candidates have gained ground in Colorado and Illinois, and the races in Idaho and Missouri have stayed roughly unchanged.

Enten mentions two possible explanations as to what’s going on:

The most hopeful hypothesis for Democrats is that there’s a lag. That is, eventually these Senate races will start reflecting the presidential race. That’s certainly plausible, given voters’ recent pattern of picking the same party in both Senate and presidential races. Down-ballot races can break late, as they did in 2006 when Democrats did considerably better than expected based on predictions from a month before the election.

Another, far less optimistic hypothesis for Democrats is that voters are purposely splitting their tickets. As my colleague Nate Silver pointed out on Tuesday, there’s some evidence that voters split their tickets when they feel confident in predicting who the next president will be.