Dem
51
GOP
49
Readers of this site are well aware of push polls—polls whose purpose isn't really to collect information, but instead to communicate misleading/false and potentially damaging information to respondents. For example: "Would it change your opinion of Joe Biden if you learned that he was once a member of a Satanic cult?"
Today, courtesy of D.S. in Cleveland Heights, OH, we bring you an example of what you might call a "trap poll," one designed to cause respondents to reveal information that is misleading/false and potentially damaging. This particular poll is courtesy of Heartland/Rasmussen, and its finding is "One-in-Five Mail-In Voters Admit to Committing at Least One Kind of Voter Fraud During 2020."
That would be pretty shocking, if it was true, but for it to be in the same universe as "true," you have to commit gross offenses against data analysis. Just to start, as most readers know, Rasmussen is probably the most partisan pollster in operation right now. Meanwhile, Heartland claims to be a "free-market think tank." We are not sure how much thinking goes on there, but the "free-market" part of that is not terribly on point. As a name like Heartland suggests, they are MAGA populist reactionaries, and very pro-Trump. Anyhow, add it up, and you should take any product of a Rasmussen-Heartland collaboration with several oceans' worth of salt.
So, how did they produce this one-in-five result? Obviously, they did not ask people if they had committed voter fraud, since nobody would admit to that. Instead, the pollster (Rasmussen, in this case), asked about various behaviors that would constitute fraud... in some states. For example, 21% of respondents said that they had helped a friend or family member fill out their ballot. That is indeed "fraud" in some places, but not in all.
To take another example, 17% of mail-in voters reported that they voted "in a state where you were no longer a permanent resident." What this conveniently overlooks is that, in nearly all cases, that is only illegal if you also cast a ballot in another state. One of us, (V), has voted dozens of times in a state (California) where he is no longer a permanent resident, and can assure you it's not a crime, because he doesn't vote in any other state.
So, the numbers here are very obviously being cooked. But the overall implications, of which there are two, are even more problematic. To start, if grandma wants to cast a ballot for Joe Biden, and you help her fill out her mail-in ballot because her sight isn't so good anymore, that's either perfectly legal (most states) or it's only technically fraud, in states that have passed overly restrictive laws for reasons that may have nothing to do with actually combating fraud. In either case, an American citizen who is entitled to vote has registered their preference for the candidate they want. That is what ballots are supposed to do. In other words, Heartland/Rasmussen are equating real fraud (votes that never should have existed) with not fraud/not really fraud (votes that accurately expressed voters' intent).
In addition, Heartland/Rasmussen really, really want you to believe that all of these allegedly fraudulent ballots were cast for Biden. However, roughly 40% of mail-in ballots in 2020 were cast for Donald Trump. If you really wanted to prove that fraud, or even "fraud," somehow changed the election result, you would need to know which candidate the ballots were cast for, and in which state they were cast. Heartland/Rasmussen does not seem to have collected that data, however. We can't tell for sure, because they did not see fit to share their questionnaire or their crosstabs. Such a lack of transparency is yet another great sign.
So, the lessons today: (1) Don't pay attention to presidential polls yet (see above), and (2) don't pay attention to results that seem impossible, particularly if they are from a garbage pollster working for a hyperpartisan client. (Z)