# Welcome to the bizarre world of conditional probability

### We have a good idea what would happen if Harris wins Pennsylvania. But we're less sure about what things would look like if Trump wins California.

*This is very wonky, so we’re going to run it as a bonus Model Talk column for paid subscribers. There should be a free post coming tomorrow or this weekend on a more mainstream topic: whether Kamala Harris can meme her way to victory. -NS*

Kamala Harris probably didn’t maximize her chances of winning the Electoral College when she chose Tim Walz as her running mate. We thought Josh Shapiro would be a better choice given his potential to help Harris win Pennsylvania — which has a 34 percent chance of deciding the election. But enough, for now, about the veepstakes. Let’s ask a different question: how much does a single battleground state like Pennsylvania tip us off to the outcome in the rest of the swing states?

Let’s figure it out using our new-ish chart that shows the probability of Donald Trump or Harris winning the election conditional on winning each state. Our forecast thinks Trump has a 46 percent chance of winning the Electoral College, but that jumps to 96 percent when he wins Pennsylvania. Phrased differently: **Harris only has a 4 percent chance of winning the election if she loses Pennsylvania**.

It’s the same story in the other Blue Wall states: Harris has a less than 10 percent chance of winning the election if she loses any of them. Trump — who is still ahead in most of the Sunbelt swing states though Harris is closing in on him quickly1 — has a few more paths to victory so these states are less important for him. If Harris wins Michigan for example, our model still gives Trump a 17 percent chance of winning the Electoral College.

So these numbers make it pretty clear that Pennsylvania and the Blue Wall states are *really *important to both candidates’ chances of winning. But where do these probabilities come from?