Predisent Donald J. Trump: James Comey and the Giant Peach Part II: The Ripening

[Byrthnoth]

I like a lot of what WaPo does, but think some of their analysts need to learn advanced division techniques so that they can normalize their results and not look like idiots.

https://www.washingtonpost.com/news/...=.7ddc940d4e66

[archibaldcrane]

I like a lot of what WaPo does, but think some of their analysts need to learn advanced division techniques so that they can normalize their results and not look like idiots.

https://www.washingtonpost.com/news/...=.7ddc940d4e66

While some people are confused into thinking that most Trump voters were working class, the actual issue is more the swing of non-college educated whites to Trump (and college-educated whites to Hillary) vs. 2012. Those two things and the black vote decline/hispanic vote increase were the major differences between 2012 and 2016.

[Byrthnoth]

idk about that because I don't think the census report has come out yet, but I do know that the article I linked is total bullshit that should have died on the editor's table.

The basic mistake they make is failing to normalize to the average subject, which is an analysis 101-level error that makes their analysis useless. For instance, if you want to say that Trump's base has normal levels of education, you have to show that they're the same as the average person polled. Instead, the authors show that the average Trump supporter is the same as the Census results. However, a sample average is not a population average and no poll is perfectly unbiased.

For instance, their "During the Primaries" section is based on analysis of online polling (Survey Monkey) from March of 2016. In it, they say that 1/3 of Trump's supporters are <50k (median income), 1/3 are 50~100k, and 1/3 are >100k, therefore Trump's support isn't biased towards bluecollar workers and most of his supporters make more than the median income. However, that doesn't take the poll's (likely substantial) bias into account. I can't actually find this poll and they don't link it in either of their articles, but I'd bet that the sample population skews higher income than normal because it's an online, opt-in poll.

I'm glad they figured out a way to get paid I guess, but those authors should leave poll analysis to 538 because they're pretty bad at it.

[Byrthnoth]

The CPS supplement for Nov 2016 did actually come out already.
https://www.census.gov/topics/public-sector/voting.html

Methods overview:
* The methods state that it's based on a survey of ~152k people over 18 years old from ~56k households with anyone in the household over 15 years old being able to answer the questions for their household.
* Those numbers don't add up to the total from the tables they provided (245.5k,) which I think this is due to weighting that they do to try and unskew their data.
* If someone refuses to answer a question, say they don't know the answer (is your young adult daughter registered to vote?), or isn't asked, it's counted as a nonanswer and no information is given about that breakdown.
* When they calculate voting/registration rates, for some reason they include noncitizens that were polled in the denominator even though said people cannot register.
* When calculating voting rate, they're dividing by all people polled (including noncitizens, unregistered citizens, and nonresponders.)

Random table 1 stats:
* The reported registration rate is 64% and voting rate is 56% including noncitizens and nonanswers as unregistered nonvoters.
* If you calculate registration rates as a percent of citizens (whether they answered or not), we have a 70% registration rate.
* If you calculate registration rates as a percent of registered citizens who replied to the question/survey, we have an 83% registration rate.
* If you calculate voting rates as a percent of registered citizens who replied, we have an 87% voting rate.

In practice, there were ~129mil votes cast out of ~330mil people, about 75% of which are old enough to vote, which would put us at a 52% voting rate, which approximately matches the CPS estimation with the same limitations.

It is not clear how the ~6mil people with felony convictions would have answered the above survey (or had the survey answered for them.)

[Infini]

There's a part of me that's seriously hoping that Day is actually logging all of these posts.

BGGate is going to be fucking lit as hell when it drops.

To be honest wait and see what Comeygate brings first. Could be thr biggest laugh of the year yet! :D

[Fiye]

I really hope comey testifies on a tuesday.

[archibaldcrane]

Unfortunately, no, he's set to testify this Thursday.

[archibaldcrane]

Dunno if you guys are keeping up with the Intercept's reporting on Russian GRU hackers working to breach US voting software companies/databases, but a leaked Top Secret document details what NSA knows so far about the attempt.

https://t.co/3Tlddkz7x6 (it's a good thorough read)

And the DOJ just arrested the suspected leaker, a 25 year old named (seriously) "Reality Winner" who printed out the document and apparently mailed it to the Intercept. http://www.washingtonexaminer.com/fe...rticle/2625014

[Fiye]

He was one of 6 people with physical access and the only one with contact to news outlets. 2ez

[archibaldcrane]

And yes, there is substantial irony in The Intercept breaking this Russia story given that Glenn Greenwald at The Intercept has been one of the biggest proponents of the "Russia story is hyped up nonsense!" thread.

[archibaldcrane]

She was one of 6 people with physical access and the only one with contact to news outlets. 2ez

.

[archibaldcrane]

Of course, the new info in this leak, which has never been confirmed previously, is going to be nearly completely overshadowed by the arrest of the leaker.

[Salodin]

Uh, so the gru attempted to actually hack and influence voting software/hardware but there's no evidence they've suceeded. At the same time, there's nothing to indicate they didn't actually suceed either...

Really hate that this is a thing =/

Sent from my SAMSUNG-SM-G935A using Tapatalk

[archibaldcrane]

I think it would be extremely difficult to hack actual votes without eventually being detected (this wasn't an effort to actually do that, it seems)

[Salodin]

What I'm getting at is, if the largest obstacle between access to some incredibly sensitive voter data is a government employee from any of the 50 states (plus PR?) not clicking on a particular email, then we're fucked homeys.

I mean, imagine if they located a zero day on said software. If it isn't already the plot of a movie, it will be in our lifetimes. I was concerned before, but now I'm shaking in my booties. =(

Sent from my SAMSUNG-SM-G935A using Tapatalk

[archibaldcrane]

What I'm getting at is, if the largest obstacle between access to some incredibly sensitive voter data is a government employee from any of the 50 states (plus PR?) not clicking on a particular email, then we're fucked homeys.

The pro (to the many many cons) of locally controlled elections is that there isn't a centralized weak point either. Tons of different data sets, voting methods, paper vs. electronic ballots, different counting machines, various vendors, etc. etc.

It makes the possibility of one group getting hacked much larger (simply due to the variety of targets) but influencing things nationally much harder. The vendor in question here had their software in use in 8 different states.

[Niiro]

The caveat being, due to the electoral process, the number of targets that you would need to infiltrate is much smaller as well.

ie: 70,000 votes in the right places vs 3,000,000 nationally

[archibaldcrane]

In the right places being the most difficult part, of course.

[archibaldcrane]

She followed both @snowden and @wikileaks on twitter lol. You'd think she would have picked up at least a few OpSec tips.

man

i mean this is almost too perfect

https://twitter.com/Reezlie/following

[Quicklet]

All the major broadcast networks will be airing Comey's testimony live on Thursday. This is being hyped up so much, it's guaranteed to fail to deliver.

Comey will say Trump made inappropriate comments to him about the Flynn investigation, Republicans will grandstand and try to make this about leaks, and we'll end where we started, with an unstable Trump livetweeting while getting angry at his TV instead of doing the actual work he's supposed to be doing.