PRESIDENT TRUMP: And you're gonna find ...
DAVID MUIR: ... those people who are on the rolls voted, that there are millions of illegal votes?
PRESIDENT TRUMP: I didn't say there are millions. But I think there could very well be millions of people. That's right.
DAVID MUIR: You tweeted though ...
PRESIDENT TRUMP: And I also say this ...
DAVID MUIR: ... you tweeted, "If you deduct the millions of people who voted illegally, I won the popular vote."
PRESIDENT TRUMP: David, and I also say this, if I was going for the popular vote I would've won easily. But I would've been in California and New York. I wouldn't have been in Maine. I wouldn't have been in Iowa. I wouldn't have been in Nebraska and all of those states that I had to win in order to win this. I would've been in New York, I would've been in California. I never even went there.
DAVID MUIR: Let me just ask you, you did win. You're the president. You're sitting ...
PRESIDENT TRUMP: That’s true.
DAVID MUIR: ... across from me right now.
PRESIDENT TRUMP: That's true.
DAVID MUIR: Do you think that your words matter more now?
PRESIDENT TRUMP: Yes, very much.
DAVID MUIR: Do you think that that talking about millions of illegal votes is dangerous to this country without presenting the evidence?
PRESIDENT TRUMP: No, not at all.
PRESIDENT TRUMP: Not at all because many people feel the same way that I do. And ...
DAVID MUIR: You don't think it undermines your credibility if there’s no evidence?
(OVERTALK)PRESIDENT TRUMP: No, not at all because they didn't come to me. Believe me. Those were Hillary votes. And if you look at it they all voted for Hillary. They all voted for Hillary. They didn't vote for me. I don't believe I got one. Okay, these are people that voted for Hillary Clinton. And if they didn't vote, it would've been different in the popular.
At Dartmouth, David Cottrell, Michael C. Herron and Sean J. Westwood have a study titled "Evaluating Donald Trump’s Allegations of Voter Fraud in the 2016 Presidential Election." The abstract:
During the 2016 presidential campaign, Donald Trump repeatedly claimed that the presidential election would be tainted by massive voter fraud. Despite winning the presidency, Trump has maintained that position through November. We suspected prior to the election that fraud allegations might dominate the post-election political landscape, and to this end we initiated a research project with the objective of evaluating the relationship between election returns and potential sources of fraud in the vein of Trump’s claims. Our research focuses on non-citizen populations, deceased individuals, the timing of results, and voting technology, and we do not uncover any evidence consistent with Trump’s assertions about widespread voter fraud. Moreover, we do not observe any striking abnormalities in two sets of states recently highlighted as potentially problematic: Michigan, Pennsylvania, and Wisconsin (the subject of ongoing recount efforts) and California, New Hampshire, and Virginia (three states cited personally by Trump). Our results do not imply that there was no fraud at all in the 2016 presidential contest, nor do they imply that this contest was error-free. They do strongly suggest, however, that the voter fraud concerns fomented and espoused by the Trump campaign are not grounded in any observable features of the 2016 presidential election.