Blog

I’ve written about testimony by a pair of lobbyists for the National Popular Vote interstate compact (NPV) in North Dakota at a 2021 hearing. Their testimony concerned a bill aimed at thwarting NPV that would keep North Dakota’s vote totals secret before the Electoral College meets. Instead, the state would report only percentages of the vote for each candidate.

As the two lobbyists correctly explained, NPV states might just use whatever information and method they felt appropriate to “calculate,” “determine,” and “assign” (their words) vote totals for states without publicly available official vote totals by the deadline, and then plug those estimates into their state’s national vote count.

But, presumably understanding that legislators would be very uneasy about using anything other than real, certified and official vote totals to determine the winner, the lobbyists tried to reassure the committee that “calculated” vote totals would be very, very close to the real totals. 

Here’s how longtime NPV lobbyists Saul Anuzis explained it in his written testimony (edited for brevity):

“…simple arithmetic will quickly reveal the lowest possible number of votes that each presidential candidate received in the state, and the highest possible number… this calculation could be done… using the total number of voters who are publicly reported to have voted in the simultaneous non-secret voting for Members of Congress, state legislators, other officials, and ballot propositions.

…President Trump’s percentage in 2020 was 65.11% (to the nearest hundredth of a percent), then President Trump could have received anywhere between 65.105% and 65.115% of the vote. That means President Trump received between 235,562 and 235,598 votes – a difference of 36 votes.

In short, the effect of SB2271 would be – no more or less than – to create 36 votes of uncertainty in North Dakota’s vote for President.”

… If the calculation were performed on the state’s entire population (from the Census Bureau), it would create 68 votes of uncertainty.”

While he didn’t explain his own methodology, his colleague Pat Rosenstiel made a similar claim of minimal error, stating that NPV states would “…get within like 36, a hundred votes of what the actual vote result was.”

These estimates were wrong in a couple of ways, but I’ll just focus on the main problem. Anuzis applied the percentages that would have been reported under the bill to the total number of votes cast for president, a number that would not have been available under the proposed legislation. 

Instead, the percentages reported would have been applied to the total number of votes cast in elections for “Congress, state legislators, other officials, and ballot propositions” or as he described a bit later to “…the reported number of persons actually going to the polls…” The former numbers are going to be smaller than the number of people actually voting on the presidential line, while the latter will be larger.

In 2020 a total of 361,819 people voted on the presidential line in North Dakota, the number Anuzis used to calculate the “36 votes of uncertainty.” But only 355,598 voted in the U.S. House election, while 364,251 were reported as having gone to the polls, according to North Dakota’s official results. Applying Anuzis’ methodology to the Congressional race credits Trump with as few as 233,503 votes while applying it to the total turnout gives him 237,182 votes – a range of 3,679, almost exactly 100 times greater than the claimed “36 votes of uncertainty.” 

North Dakota is of course a lightly-populated state, meaning the range for “votes of uncertainty” might not seem like much in the context of an election with 160 million voters. But that error range gets much bigger in more populous states. In Louisiana, for example, applying the same methodology to the 2024 election results produces a range for Trump of between 1,147,528 and  1,217,246 votes – a spread of 69,718 “votes of uncertainty” (he actually received 1,208, 505, meaning the lower estimate would have taken about 60,000 votes from his national vote total if that was what was “assigned” to him). In larger states the difference between the real vote totals and what would be “calculated” as the vote totals to “assign” candidates could easily be in the hundreds of thousands. 

The compact’s advocates have for nearly two decades been insistent that NPV is easy to implement and each compact state’s task of adding up vote totals from across the country won’t be a problem because “we can all do the math,” as one of its lobbyists put it in a different hearing. Of course it would be easy to add estimated vote totals to real vote totals, but the fact that Anuzis was off by a factor of 100 shows that the math involved in “calculating” a vote total to “assign” a state is much trickier. 

And the lack of any specified method or standards for the “calculations” of vote totals to “assign” opens the door to manipulation to benefit or harm particular candidates, and would likely result in compact states arriving at different versions of the national vote totals, undermining confidence in the election. Which perhaps explains why, since that North Dakota hearing, NPV’s lobbyists have done everything they can to pretend this testimony doesn’t exist.