I'm confusion. How does an error like this come about? By itself, there is no justification for a 500 vote swing on 4k or 100k votes. Does it accumulate in the running sum?
As you said, a 500 vote rounding error is huge. In small precincts and house races, this would represent a dramatic change. As the other pede said, I can't see a reason why we need to implement rounding at any stage here. The numbers are neither big or complex enough to necessitate rounding.
A rounding error will be noticed when the true vote total shifts from something like 2.05% to 2.04% (or 2.07 to 2.03, or 2.08 to 2.04, etc.)
Since the data is lacking precision, it will be reported as shifting from 2.1% to 2.0%. Then we just estimate the vote count by multiplying the percentage and the vote total.
So if there are 1 million total votes at that time, this .1% shift could represent a rounding error shift of up to 1000 votes.
So as an example, again say there are 1 million votes and you add 20,000. Let's say the true third party totals are 20,600 + 300 added.
The spreadsheet will show this as 21,000 (2.1% * 1,000,000) votes before the new votes are added - a 400 point error in the positive direction - but only 20,400 after (2.0% * 1,020,000), despite the fact you actually added 300 votes.
So while you actually gained 300 votes, using the percentages to approximate makes it look like you lost 300 votes.
That's perfectly understandable in terms of math, but I don't see WHY they'd count this way. Originally the number is just a sum, so instead of getting the sum of votes into a percentage, and then using a rounded percentage to calculate approximate votes, why not just keep using the original number when it is already there?
2nd question: is this rounding just used with regard to News reporting and only relevant there, or did they get this number from some type of output at vote counting areas?
If you're asking why the video creator did it, supposedly that's all the info that was reported in the feed.
If you're asking why the source information did it this way, well I guess either incompetence or they want to minimize available data to make hiding fraud easier.
Regarding your second question, unfortunately I'm not sure exactly who was the originator of the data and the extent it's used. The video says the data was retrieved from the new york times, so I don't know if they also have exact vote counts somewhere else or they used this stupid method for their state result reporting as well.
Presumably the actual vote counts exist somewhere... you might have to search state government/secretary of state webpages for individual states to find it though, and it's probably only updated far more slowly. I'm unaware if there's another place to find it.
I'm confusion. How does an error like this come about? By itself, there is no justification for a 500 vote swing on 4k or 100k votes. Does it accumulate in the running sum?
As you said, a 500 vote rounding error is huge. In small precincts and house races, this would represent a dramatic change. As the other pede said, I can't see a reason why we need to implement rounding at any stage here. The numbers are neither big or complex enough to necessitate rounding.
A rounding error will be noticed when the true vote total shifts from something like 2.05% to 2.04% (or 2.07 to 2.03, or 2.08 to 2.04, etc.)
Since the data is lacking precision, it will be reported as shifting from 2.1% to 2.0%. Then we just estimate the vote count by multiplying the percentage and the vote total.
So if there are 1 million total votes at that time, this .1% shift could represent a rounding error shift of up to 1000 votes.
So as an example, again say there are 1 million votes and you add 20,000. Let's say the true third party totals are 20,600 + 300 added.
The spreadsheet will show this as 21,000 (2.1% * 1,000,000) votes before the new votes are added - a 400 point error in the positive direction - but only 20,400 after (2.0% * 1,020,000), despite the fact you actually added 300 votes.
So while you actually gained 300 votes, using the percentages to approximate makes it look like you lost 300 votes.
That's perfectly understandable in terms of math, but I don't see WHY they'd count this way. Originally the number is just a sum, so instead of getting the sum of votes into a percentage, and then using a rounded percentage to calculate approximate votes, why not just keep using the original number when it is already there?
2nd question: is this rounding just used with regard to News reporting and only relevant there, or did they get this number from some type of output at vote counting areas?
If you're asking why the video creator did it, supposedly that's all the info that was reported in the feed.
If you're asking why the source information did it this way, well I guess either incompetence or they want to minimize available data to make hiding fraud easier.
Regarding your second question, unfortunately I'm not sure exactly who was the originator of the data and the extent it's used. The video says the data was retrieved from the new york times, so I don't know if they also have exact vote counts somewhere else or they used this stupid method for their state result reporting as well.
Presumably the actual vote counts exist somewhere... you might have to search state government/secretary of state webpages for individual states to find it though, and it's probably only updated far more slowly. I'm unaware if there's another place to find it.