4 Comments

I've noticed that it's common in predictive or analytical discussions to simply never state key assumptions (like how one thinks undecideds will split, which are often implicitly assumed to split in a particular way by an author/op-ed). In macroeconomic analysis it is also common (especially in the recent inflation discussion).

Expand full comment
author

Yep. Even in cases where the assumption is wrong, the vibe is "if we assume 50/50, how wrong could it be?" Which is a major premise of this post - and why we shouldn't place assumptions where data should go

Expand full comment
Oct 30Liked by Carl Allen

@Lorenzo Warby

Expand full comment
Oct 29Liked by Carl Allen

Excellent analysis.

I have felt for sometime that the political pundits were spending too much time on the current spread between Harris and Trump in the swing states and not enough on the undecided voters. Most swing state polls “predict” a close race because the spreads are two points or less. But common sense tells me that a poll showing a 46-44 lead with 10 undecided is much different than a 49-47 lead with 4 undecided. The trailing candidate needs to get 6 out every 10 undecideds in the first example and 3 out of every 4 undecideds in the second example. And this assumes that all of the undecideds actually vote!

I may betray my bias here but I sense the undecideds in the swing states are more likely to break for Harris than Trump. If I’m right then Harris will win the swing states closer to a 3-4 point margin than the current “tie.”

Expand full comment