Discussion about this post

User's avatar
Douglas Rose's avatar

If we take the presidential year congressional vote as a proxy or estimate for what we expect the next midterm vote to be, we might want to correct for forces peculiar to the presidential year, usually those associated with the presidential election. Removing those would provide a better estimate of what the midterm vote would be if nothing unusual happened. There are a number of possible ways of removing the presidential influence, e.g., estimating how much of a net drag JFK was on the Democratic ticket in 1960 or Goldwater was on the Republican ticket in 1964. The best ones would presumably improve the fit of the regression lines of midterm on prior congressional vote split.

Expand full comment
Douglas Rose's avatar

Very helpful. For the chart on the shift from presidential to midterm vote, what happens if you factor in presidential vote? At first glance, it appears that when Democrat presidents lead the ticket, their absence is sorely missed two years later, while when they drag the ticket while being elected, Dems bounce back well.

Expand full comment
3 more comments...

No posts