...and Yes, Napoleon is right.
You'll need a multivariate regression to get closer.
You can't use multivariate regression because no election is similar to the previous one. If you find that, for example, California tends to swing more strongly than the nation (ie, when republicans win, it is more republican and conversely), it might just be that California has trended democrat by the time democrats won bigger victories. Presidential elections are too rare and subject to bold swings for complicated statistical analyses to make any sense.
Conclusion : we have no way to know how an EV map would have turned out with a different PV margin. When you have no way to know something,
the best assumption is the simplest one. In your own words, Napoleon...