At 42%, the model envisions virtually no chance for Democrats to hold the Senate and predicts a loss of four seats as the most likely outcome. At 42%, the Colorado Senate seat could potentially come into play, assuming that Republicans produced a credible candidate (remember that a relatively unheralded candidate held Sen. Michael Bennet to a six-point margin in 2016).

Now, it doesn’t necessarily follow that if things get even worse for the Democrats that they will lose six or seven seats. There’s a realistic floor for political parties in elections where even the hardcore partisans begin to turn on the president but nevertheless vote for their party’s nominee in Senate and House elections. The classic example of this phenomenon is 2008, when President George W. Bush’s job approval fell into the 20s nationally. Despite this, Republicans managed to win Senate elections in places like Kentucky and Georgia even though Bush’s disapproval brushed up against 60% in both states.

Nonetheless, if the president’s job approval doesn’t improve, his party will find itself in dire straits in the midterms. The good news for Democrats is that there is still time for things to turn around. The bad news is that public perceptions of issues like the economy and the pandemic are relatively slow to do so; the clock is starting to tick.