You can't judge the outcome of elections by standards under which they were not held. Popular-vote-wins would result in very different campaign efforts, and different voting patterns.

It's like holding a baseball game and then, after the fact, declaring that the winner isn't the team with the most runs, but the one that got the most hits. Sure there's some correlation there, but both teams would have played differently if they'd known that's how they'd be judged, so you can't be sure who would have won if they'd known that's how they were supposed to be playing.