Test your logic against nearly any other year. Let's use 2004 as it was a relatively calm election, and assume that the number of people who would not answer whether they vote is around 20% of the number that claimed they voted. Yes, I am making up that number because that question was never asked, but from above you can see that 20% is reasonable. As usual numbers are in thousands.
From above:
2004 125,736 122,295 3,441
and the theoretical unanswered population: 25,147
If your method works then we have a range between 125,736 and 150,883. This is obviously wrong because the actually number of votes according to the FEC was 122,295. In fact for any year except 2008 (barely) and 2020, your method would have the actual number of votes outside of your calculated range. This is entirely because the census number of people voting is ALWAYS greater than the actual number of votes except for 2020. It means there was not just 4 million votes made up but more likely 6 to 12 million votes made up, especially if we compare 2020 to the Bush and Clinton elections.
But you again have to assume that every year will have the same distribution, which will not always take place, and it's illogical to assume it will always follow a predicted trend.
Test your logic against nearly any other year. Let's use 2004 as it was a relatively calm election, and assume that the number of people who would not answer whether they vote is around 20% of the number that claimed they voted. Yes, I am making up that number because that question was never asked, but from above you can see that 20% is reasonable. As usual numbers are in thousands.
From above:
2004 125,736 122,295 3,441
and the theoretical unanswered population: 25,147
If your method works then we have a range between 125,736 and 150,883. This is obviously wrong because the actually number of votes according to the FEC was 122,295. In fact for any year except 2008 (barely) and 2020, your method would have the actual number of votes outside of your calculated range. This is entirely because the census number of people voting is ALWAYS greater than the actual number of votes except for 2020. It means there was not just 4 million votes made up but more likely 6 to 12 million votes made up, especially if we compare 2020 to the Bush and Clinton elections.
But you again have to assume that every year will have the same distribution, which will not always take place, and it's illogical to assume it will always follow a predicted trend.