I am finding:
If you order by [state] & [timestamp]:
- You will have rows that are greater in sequence that have less [votes]
If you order by [state] & [votes]:
- You will have a huge swing in the [eevp], [trumpd], [bidenj] columns that distort the numbers by a lot
If you order by [state] & [rowid] (original sequence):
- You will have some rows that have a lesser timestamp
- You will have some rows that have a greater vote count ([votes]) with a sequence lower than the current and in many of these instances the [trumpd], [bidenj] may also be distorted
select v.RowId
, v.seq
, v.[state]
, v.[timestamp]
, ERROR = case when v.votes < lv.votes then 'ERROR: Last Vote Count was Greater than the Current vote'
when v.votes <
( select max(i.votes)
from dbo.Votes i
where i.[state] = v.[state]
and i.seq < v.seq
)
then 'ERROR: Current Vote Count is not greater than a previous Sequence'
when v.votes <
( select max(i.votes)
from dbo.Votes i
where i.[state] = v.[state]
and i.[Timestamp] < v.[Timestamp]
)
then 'ERROR: Current Vote Count is not greater than a previous Vote count where seq is less'
when v.[timestamp] < lv.[timestamp] then 'ERROR: Current Timestamp is not greater than the Last Timestamp'
when v.[timestamp] <
( select max(i.[timestamp])
from dbo.Votes i
where i.[state] = v.[state]
and i.seq < v.seq
)
then 'ERROR: Current Timestamp is not greater than a previous Sequence'
when v.[timestamp] <
( select max(i.[timestamp])
from dbo.Votes i
where i.[state] = v.[state]
and i.[Timestamp] < v.[Timestamp]
)
then 'ERROR: Current Timestamp is not greater than a previous Timestamp where seq is less'
else
NULL
end
, v.votes
, v.eevp
, v.trumpd
, v.bidenj
, v.other
, TrumpVotes = v.votes * v.trumpd
, TrumpVoteChange = v.votes * v.trumpd - isnull(lv.votes * lv.trumpd, 0)
, BidenVotes = v.votes * v.bidenj
, BidenVoteChange = v.votes * v.bidenj - isnull(lv.votes * lv.bidenj, 0)
, otherVotes = v.votes * v.other
, OtherVotesChange = v.votes * v.other - isnull(lv.votes * lv.other, 0)
from dbo.Votes v
left join dbo.Votes lv on lv.[state] = v.[state] and lv.seq + 1 = v.seq
order by v.[state], v.seq
/* choose your preference to order by */
-- update seq by timestamp update v set seq = wrk.seq , other = case when v.[trumpd] + v.[bidenj] = 0 then 0 else (1-(v.[trumpd] + v.[bidenj])) end from dbo.Votes v inner join ( select rowid , seq = ROW_NUMBER() over (partition by [state] order by timestamp asc) from dbo.Votes ) wrk on wrk.rowid = v.rowid
-- update seq by vote count update v set seq = wrk.seq , other = case when v.[trumpd] + v.[bidenj] = 0 then 0 else (1-(v.[trumpd] + v.[bidenj])) end from dbo.Votes v inner join ( select rowid , seq = ROW_NUMBER() over (partition by [state] order by votes asc) from dbo.Votes ) wrk on wrk.rowid = v.rowid
-- update seq by original order update v set seq = wrk.seq , other = case when v.[trumpd] + v.[bidenj] = 0 then 0 else (1-(v.[trumpd] + v.[bidenj])) end from dbo.Votes v inner join ( select rowid , seq = ROW_NUMBER() over (partition by [state] order by rowid asc) from dbo.Votes ) wrk on wrk.rowid = v.rowid
what i mean by not actual data is that it cant possibly be actual election data not edison data. Does anyone here have any sort of access to the schema or data for voting systems? thats the ONLY true way to see real discrepancies that arent caused by rounding issues
ill explain it like i would to a child since it doesnt seem to get through some of you
the json file for nyt scraping IS NOT the actual data
0.12345678901234567 is more accurate than .123
and there lies the discrepancies you are seeing
you cant fix the datasource. its incomplete. the precision on the decimals must be a rounded value from its actual source.
ive said as much to no avail and so has others as ive scrolled through but the author of this content doesn't seem to care about it and now repubs will look stupid as shit soon now that trump and his sons have tweeted about this
my account is new because i saw someone tweet the article and it got me excited to find out where the data came from so i can run my own analysis to find more discrepancies to forward to the lawyers in hopes to overturn the steal
i instead found that the data unusable.
But really, I truely insist that you dont take my word for it. Instead:
- download the csv for yourself
- import it into excel or sql db (ive done both)
- add your formulas or queries to replicate his results
- apply same formula\queries to biden and then remainder and you'll see for yourself
The article links this thread as its source so I created an account to inform y'all to retract this because instead of helping its going to make all republicans look like effin retards that dont know how to do basic analysis of data and instead we are experts at forwarding unvalidated results. I WANTED TO GET THIS FIXED\RETRACTED BEFORE IT GOT TO IMPORTANT PEOPLE.
But now its too late and I'm sure sometime really soon DJT and repubs that linked this or referenced this will be laughed at. So much for trying to help
so many possibilities but likely they have a direct interface that has a much more precise decimal value setup but via their web app they truncate the value
*Again, to make my case that you cant use this Edison dataset as any kind of evidence, this screenshot should be sufficient. *
Light Green = Edison Data
You can see that Neither candidate gained any percentage gain but the only difference was the vote count going up. Yet, all 3 parties "lost" votes.... This can only mean 1 thing, the 3 decimal is a rounded number. I would imagine the true decimal precision is significantly more (actuaries i know use 17 decimals in their calculations)
its vague at best. If i had to take an educated guess, the true decimal precision is much more than 3 decimals
shit... Eric Trump just posted about this now we are going to look like clowns... Bro, delete this thread already its a bad data source as I've already shown!
no, edison data or at least the edison data in the source is either incomplete (ie missing several more decimal points for any accuracy) or just plain an estimated source. I'm for trump all the way since 2015 but we cant stoop down to democrat level and say this dataset is legit when its absolutely clear its not. Programmer\DBA\DB Dev for 20 yrs+
I hate to break the bad news but I downloaded the csv, opened excel, expanded a some additional calculations grabbed the michigan only data and replicated the results but I also noticed that biden had more loss due to same calculations but the real loser was the "others" (jorg and the bunch).
I like the idea but i believe the edison dataset is too loose and vague at best. Need real #s that arent calculated.... or perhaps I'm missing something??
while looking into PA, i found something i cant look up....
anyone have any idea what absentee_max_ballots would mean?
I'm asking because in Allegheny for example looks like this:
absentee_max_ballots = 44826 absentee_votes = 335573
https://i.maga.host/gold70c.png