2016 Postmortem
Related: About this forumCNN's own numbers dispute their flawed 52-48 exit poll number
http://www.cnn.com/election/primaries/polls/ny/demThe various demographic breakdowns from the CNN exit polls prove that 52-48 final number was a major screw up. All of the key demographic data pointed to a big Clinton win. Here are CNN's own numbers from the link above. How they derived 52-48 with these sub-groups is a mystery. Nonetheless, the 52-48 number should never have been published because it was so far off from the supporting data.
Here's the key demo data from CNN. In all instances below, I have put Clinton's percentage first followed by Sanders' percentage.
Gender:
Men: 50-50
Women: 63-37
Age:
18-29: 35-65
30-44: 53-47
45-64: 63-37
65+: 73-27
Race:
White: 50-50
Black: 75-25
Latino: 64-36
None of this data supports a 52-48 race. There's a lot more data at the link and none of it supports 52-48 either.
People need to stop quoting or using that 52-48 number because it is complete junk. I have no idea where it came from but it was not based on any of the other data collected by CNN.
This demo data does support a double digit win for Clinton and that matches the official results.
geek tragedy
(68,868 posts)KPN
(15,646 posts)refute a total of 52-48. That would be true if every demographic represented an identical share of the total, but that's not the case. Do you have more numbers to support your claim?
you are being argumentative to any one with understanding of the context of those numbers
CrowCityDem
(2,348 posts)LonePirate
(13,426 posts)Look at the gender gap. For a 52-48 final number based on those gender splits, you would need roughly 6.5 times as many men to vote as women. That hasn't happened in this country in a century. 52-48 is simply not possible, realistic or believable.
DemocratSinceBirth
(99,710 posts)Men: 50-50
Women: 63-37
Men made up 41% of the electorate and women made up 59% of the electorate:
http://www.cnn.com/election/primaries/polls/NY/Dem
50 X .41 = 21%
63 X .59 = 37%
58% BOOM!!!
randome
(34,845 posts)So showing Sanders as somehow gaining on Clinton would be...the opposite of that?
[hr][font color="blue"][center]There is nothing you can't do if you put your mind to it.
Nothing.[/center][/font][hr]
bobbobbins01
(1,681 posts)Exit polls are usually the opposite, they help keep elections legitimate. The more you know!
LonePirate
(13,426 posts)bobbobbins01
(1,681 posts)Any idea where the number comes from or if the raw data has been released?
LonePirate
(13,426 posts)The raw data for all of the demos is at link. There are two and a half pages of this data at the link. The rest is simple math.
bobbobbins01
(1,681 posts)I have Bernie with a 47.27%. If you account for rounding, i would say they could match. Are you seeing something different?
LonePirate
(13,426 posts)Use the gender numbers.
Women split 63-37 for Clinton and men split evenly 50-50.
Now if the same number of men and women voted, you could expect a final number of 56.5-43.5 based on the 63/37 and 50/50 splits (63+50)/2 and (37+50)/2 . That 56.5-43.5 result is a lot closer to the official result than it is to 52-48.
How did you come up with 47.27? There is no way you can reach that number based on the gender numbers CNN provided unless men outnumber women in the final tally by 5 to 1 at a minimum.
bobbobbins01
(1,681 posts)So with men, each candidate gets 41%...then you take women, Bernie got 37, hillary got 63...you do the proportions there... 37/59 and 63/59 and you get 6.27 and 10.6...add those to the 41 and you get 47.27% to 51.6%.
LonePirate
(13,426 posts)That is simply an indication of how many men and women were included a n the exit poll. It is not a breakdown of the ratio of women to men who actually voted. The 59-41 exit poll sample split is only applicable to the exit poll. You need to use a much simpler method that approximates an equal number of male and female voters. The 50/50 ratio is much closer to the usual reality that tops out around 47 men to 53 women per every 100 voters.
For simplicity, let's use the 50/50 ratio and assume only 200 people actually voted in the primary instead of roughly 2 million. So we have 100 men and 100 women. Using the splits from the exit poll, 50 men each voted for Hillary and Bernie. However, 63 women voted for Hillary and 37 women voted for Bernie. That's 113 votes for Hillary and 87 for Bernie in our pool of 200 or 56.5% to 43.5%. That's closer to the official result than it is to 52-48.
In order to reach a final result that is 52-48, one of these needs to happen:
1. The number of male voters needs to increase significantly (500 men and 100 women)
2. The number of female voters needs to decrease significantly (100 men and 20 women)
3. Some combination of 1 and 2 (300 men and 50 women)
Yet if you change the voting pool to align with 1, 2 or 3, you radically alter the composition of the electorate and it no longer approximates the 50-50 split, let alone a typical 53-47 split in favor of women that is seen in most elections nowadays. Example 1 is a 17-83 female to male split which reality tells us is impossible.
There is no way to crunch the internal numbers to have them reflect reality and also reach the 52-48 prediction. I guess someone could make a case that the 52-48 number is right and all three pages of internal numbers are wrong. That would beg the question of where are the correct internal numbers.
DemocratSinceBirth
(99,710 posts)bobbobbins01
(1,681 posts)Exit polls are used much of the time to determine if an election shows any evidence of fraud. I'm not sure why you'd argue against that.
DemocratSinceBirth
(99,710 posts)Here are the votes by neighborhoods:
http://www.nytimes.com/interactive/2016/04/19/us/elections/new-york-city-democratic-primary-results.html?_r=0#12/40.7125/-73.8547
If you want to get granular you can get precinct voting records but that will take a lot of leg work.
bobbobbins01
(1,681 posts)I don't know how valid or invalid the exit polls were for NY. I don't think your link shows that, it just shows the vote counts(but I could be reading it wrong). I'd love to see the raw data if its available.
DemocratSinceBirth
(99,710 posts)I know they used to have precinct level reporting. I am sure it's available for NY. I used to look at predominately African American precincts that went 115-3 for the Democrat and got a chuckle.
tritsofme
(17,380 posts)These people are truly unhinged with very little connection to reality. Wow...
LonePirate
(13,426 posts)I am not saying there was no disenfranchisement or tampering. I am saying all of the other CNN data (like what I posted above) supports the official tally and they do not support that ridiculous 52-48 number.
randome
(34,845 posts)[hr][font color="blue"][center]There is nothing you can't do if you put your mind to it.
Nothing.[/center][/font][hr]
CobaltBlue
(1,122 posts)It depends on when you look at exit polls.
As the state of New York closed at 09:00 p.m. ET, CNN showed a 52-to-48 percent estimate.
I looked at the exit polls. Hillary was carrying women by +13; Bernie was carrying men by +09. (Size of the vote was something like 57/58 percent of women and 42/43 percent of men.)
But that had to do with timing. That was with what CNN had in within, say, 10 minutes after the polls closed.
What happens is that updated information gets reflected with a tweaking of the numbers in the exit polls.
So, to answer this thread's OP, you have to be mindful of exactly when you are looking at the exit polls information.
LonePirate
(13,426 posts)Then they are over 12 hours late in publishing it. Not only that, but if they had data to support 52-48, they definitely should publish it because none of their other data supports it.
CobaltBlue
(1,122 posts)It seems you're not understanding what I posted.
CNN did updating. (It does that.)
One would have had to have taken screenshots of what the exit polls, from CNN's website, reported at 09:00 p.m. ET. And then at 09:05 p.m. ET, 09:10 p.m. ET, 09:15 p.m. ET, 09:20 p.m. ET, 09:25 p.m. ET, 09:30 p.m. ET, 09:35 p.m. ET, 09:40 p.m. ET, et al.
LonePirate
(13,426 posts)Please feel free to point to new data from CNN that overrules the data from their own web site which I linked above. The data in my post comes directly from CNN. If CNN has data to support their 52-48 result, then that data is not on their website right as of roughly 12:30 PM EDT on April 20, 2016. That is pretty simple to understand.
Response to LonePirate (Reply #27)
CobaltBlue This message was self-deleted by its author.
LonePirate
(13,426 posts)I only looked at the internals this morning so I have no idea if they changed or by how much compared to when that 52-48 announcement was made. Your hypothesis is one I cannot weigh in on because I never saw the demo data grow from its initial sample size to its final one. That being said, it is odd or interesting that the overall final exit poll number, be it 52-48 or something else is not included with the other exit poll data. Regardless of that, there is no way 52-48 can be obtained from the CNN data I linked above which is the point of my OP.
CNN got burned big time if they did release a "final" number based on remarkably incomplete sample data. Or they got burned by botching the final number based on the final sample data. Either way, CNN screwed up and now we have people screaming that the election is a sham based on this wildly inaccurate exit poll.
Response to LonePirate (Reply #45)
CobaltBlue This message was self-deleted by its author.
glowing
(12,233 posts)The vote was most likely F'd with at the central tabulator. She needed a big win. I think they made it bigger than it
Really was.
It should be more than clear since 2000, that we need election monitors! We also need to have money out of politics. I almost think it should be illegal to buy tv political ads because it's costly and manipulates the media coverage.
DemocratSinceBirth
(99,710 posts)Some people think the Moon landing was faked but they can't prove it either.
The vote was most likely F'd with at the central tabulator.
Central tabulator???
Here is a map of election results by neighborhood:
http://www.wnyc.org/story/map-ny-primary-vote-nyc-2016
There are also precinct level results available.
LonePirate
(13,426 posts)IamMab
(1,359 posts)I usually have to pay for comedy like that.
oberliner
(58,724 posts)So they made it appear that the race was closer than it was so that people would keep watching (since the Republican race was called immediately).
MadBadger
(24,089 posts)LonePirate
(13,426 posts)MadBadger
(24,089 posts)The exit polls did indeed show 52-48, but you can't expect by nights end for the demographic data to add up to that because they change the numbers. I believe they had Hillary winning 57% of Women for example as the polls closed. That number climbed to over 60% by nights end.
MoonRiver
(36,926 posts)muriel_volestrangler
(101,322 posts)After all, they have an actual overall result to quote, no an exit poll. The exit poll is now only useful for teh sub-categories.
auntpurl
(4,311 posts)that 4-point difference in the exit poll cited as evidence that the vote was rigged.
muriel_volestrangler
(101,322 posts)It's the curse of DU. Possibly the biggest collective delusion DU has.
Agnosticsherbet
(11,619 posts)They were very much an outlier last night.
DemocratSinceBirth
(99,710 posts)Benchmark Politics @benchmarkpol 16h16 hours ago
Why are the exit polls wrong? Well, Bernie is killing it in Buffalo. We know that the exit polling was done heavily there.
Agnosticsherbet
(11,619 posts)in the state.
Sounds shifty to me, but I don't work at CNN.
politicaljunkie41910
(3,335 posts)in the hour leading up to the closing of the polls, Brian Williams and Rachal Maddow were giddy over the numbers that the exit poll were reporting and they were alluding to the fact that it wasn't even close. Then once the polls closed, they immediately said that the race was too close to call. I immediately said that's bullcrap. They just spent the last hour telling us that both parties were going to be happy with the results and that they were not even close, though careful not to mention who it was who was going to be pleased.
I'm not in general a conspiracy theorist but, I think both CNN and MSNBC misled us for ratings because they both had a vested interest in keeping us all glued to the TV for the next several hours with their News Teams all ready to go, and they knew that was not going to happen if they had told us that Hillary had a 58-42 lead at the outset. A lead that didn't seem to change at all from beginning to end once it was revealed. Sure it went back and forth between a point or less throughout the evening, but the final numbers were essentially the same as the beginning numbers. But no where near that Exit Poll number they threw out at the beginning of the 9:00 hour, after having just spent the 8:00 hour telling us both parties were going to be happy about the results.
MoonRiver
(36,926 posts)Of course they want the most easily defeated Democratic candidate to win. That's the crap they are selling to we consumers.