Why I am confident Clinton would have won without the Electoral College
To start with, I don’t think it’s a good idea to make general agnostic statements like, “Election is a multi-factor process, who knows what would have changed, if there were no Electoral College…” The factors that influenced this election are quite well understood (though not for making precise state-by-state polling forecasts), and there is only one serious factor that could have significantly changed in the absence of the Electoral College.
This factor isn’t about the preferences of various demographics. Trump would have remained Trump (he only managed to behave himself somewhat in the last few days of the campaign), and the influence of his rallies would have been diminished by spreading them over the entire country rather than a few swing states (SS).
However, some commentators reasonably claimed that in the absence of the Electoral College, the turnout would have been higher in the current non-swing states (NSS). Since the pro-Democrat NSS are more numerous, if the increase in turnout in them were disproportionally due to the GOP voters, it would be hard to predict the outcome. The turnout estimates prima facie verify part of this theory, with swing states generally having a significantly higher turnout (http://www.electproject.org/2016g).
With this in mind, let us consider a scenario with several assumptions that, taken together, are so heavily improbable and favoring Trump that if he would still clearly have lost in it, the case would be closed, even though the calculations are fairly rough.
1) The turnout in the two giant Democrat NSS (California and NY) rises by 15 p.p. to the current SS level (improbable because it suggests that the only factor driving the very low turnout in them this time was their non-SS status).
2) This increase in turnout is 60% Republican-driven.
3) The current SS vote the same way with the same high turnout (very improbable because they would have ceased to be SS).
4) The net increase in the Democrat turnout in deep red states is exactly balanced by the increase in the Republican turnout in the remaining deep blue states (which is very unprobable, given the giant of Texas).
In this scenario, the net GOP gain is 3 p.p. in California and NY. The population of these states is roughly 20% of the US total. Thus, the net gain would have been around 0.5 p.p. nationally, given that the total would have been more than the current one with which the comparison should be made. Clinton is projected to win the popular vote by 1.5–2 p.p. QED.