(CNN) As the campaign lurches into its final three weeks, Donald Trump has turned up the rhetoric, attacking his opponent, Hillary Clinton, the news media covering the campaign and, most controversially, the process used to select the next president.
"Of course there is large scale voter fraud happening on and before election day," Trump tweeted Monday. "Why do Republican leaders deny what is going on? So naive!"
Trump's claim that he is about to be cheated out of a victory at the ballot box is not new. In August he railed against federal court decisions that struck down ballot restrictions, including that voters present specific forms of identification, in several states. "We may have people vote 10 times," he told the Washington Post. "it's inconceivable that you don't have to show identification in order to vote or that the identification doesn't have to be somewhat foolproof."
In recent days Trump's argument has picked up in intensity. When asked for proof of his allegations of potential vote stealing Trump's campaign provided the following data points:
A study by the Pew Center on the States entitled "Inaccurate, Costly and Inefficient: Evidence that American Voter Registration Systems Needs an Upgrade."
The 2012 study presented a number of disquieting findings, including:
The Pew report certainly underscores deficiencies in the voter registration system. But the Trump campaign doesn't say how these deficiencies disadvantage his candidacy. For example, there is no evidence in the Pew report that people who have registrations in two states are voting twice for Democrats or for Republicans. And while Trump surrogate Rudy Giuliani has said, "dead people generally vote for Democrats rather than Republicans," he has not proffered any evidence to back up that statement.
RELATED: Why Trump's talk of a rigged vote is so dangerous
The Trump campaign noted media reports that indicated that, in 2012, more than 700 people might have voted twice in Pennsylvania and that Philadelphia elections officials have indicated they have referred a number of voting irregularity cases for prosecution.
But those reports he cites do not indicate who these individuals voted for -- if they even voted twice at all. (The report only said they "may" have voted twice.) Furthermore, the Pennsylvania Secretary of State told local media that potential voter fraud may have occurred among members of both parties.
In addition, Philadelphia officials say only nine cases have been referred for prosecution. Most of these involve local election boards adding votes to machine tallies to ensure the totals that are recorded on machines match the number of voters on sign-in sheets.
"This indicates that voting fraud does occur," says Al Schmidt, a Republican and one of three City Commissioners who oversee elections in Philadelphia. "But if voter fraud was occurring in a widespread conspiracy to sway a presidential election it would have been the easiest thing to see."
The Trump campaign claims that "illegal immigrants" are voting, citing research conducted by Jesse Richman and David Earnest, two professors at Old Dominion University. Using data from the Cooperative Congressional Election Study, the two men estimate that 6.4% of non-citizens voted in 2008 and 2.2% of them cast ballots in 2010. Studies by the Public Interest Legal Foundation, a conservative voting watchdog group, also estimated more than 1,000 non-citizens registered to vote in eight Virginia counties since 2011 and that 86 non-citizens were registered to vote in Philadelphia in between 2013 and 2015 -- 40 of them voted in at least one election.
Neither the report by the researchers nor the conservative watchdogs say if these non-citizens who voted were, in fact, undocumented immigrants, though, no matter their immigration status, it is against the law for non-citizens to vote in state and national elections. The Old Dominion researchers stressed that the non-citizen vote tends to favor Democrats with President Barack Obama getting more than 80% of the 339 votes they studied in 2008.
But in some cases, there are too few of these voters to make any difference in a presidential election. For example, the watchdog group tallied what they believe were 40 votes cast by non-citizens in Philadelphia elections in 2013, 2014 and 2015. That's out of a total of more than 1.3 million votes who voted in those years -- hardly enough for a rounding error.
Given the evidence Trump has supplied, we rate his claim that large scale voter fraud exists as FALSE.