Wednesday, June 22, 2016

Loony Charnin Is At It Again.

I exposed Richard Charnin's nonsense numbers in an earlier blog post, but that hasn't caused Charnin to change his tune. He's still playing the same old song, dissonant, off-tempo, shrieking notes and all, and nothing seems to be able to snap him out of his dream-world where he sees himself as some kind of rock star.

In yet another of my blog posts, I exposed the so-called "Stanford" study (read, paper) that claimed Hillary did well in states without a paper trail, but did not win in states which did have a paper trail. When I corrected that paper's math, it was shown that Hillary still had a large margin in non-paper-trail states, but that she won in paper-trail states as well, thus nullifying the argument of the poorly-written paper.

Well, wouldn't you know it, stupidity loves company. Not long after the fake Stanford paper came out, Richard Charnin posted to his blog on June 15th. And what did he claim? Why, he made exactly the same sort of argument that the fake Stanford paper made, but did an even worse job!

He opens with:
"This [link to the original paper]  is an excellent analysis of the Democratic primaries from Axel Geijsel of  Tilburg University -(The Netherlands) and  Rodolfo Cortes Barragan of  Stanford University  (U.S.A.) ."

Oh it is, is it? Just wait until you see what happens next.

First he points out that Hillary did quite better in the non-paper-trail-states, implying that the paper-trail states showed a truer percentage of the actual vote. Never mind demographic differences between the states. Never mind the differences in urban concentrations in non paper-trail states. And never mind that many of the non-paper-trail states are also significantly bigger. Having a paper trial in gargantuan states like Texas or Pennsylvania is a logistical nightmare! Yet Charnin would have us believe that it's all part of a Clinton conspiracy.

Now, Hillary did do better in states without paper trails, that much is true. Taking a look at which states went her way, it is easy to see why. Hillary's measured strengths are well documented. She does well with older Democrats and minorities. And the states without paper trails where she got these votes big-time were Arkansas, Florida, Georgia, Indiana, Mississippi, Pennsylvania, South Carolina, Tennessee, Texas and Virginia. A mix of big and small states, but most of them with big percentages of older Democrats and minority voters. Florida is a very geriatric state. Mississippi's Democrats are largely African-American. Georgia follows Atlanta, a very black city. Arkansas, the state where Bill Clinton was once governor? Naturally home turf! The only state on this list which Hillary lost was Indiana, and she didn't even campaign there.

But as I showed in my previous blog post, she also did quite well in states that did have a paper trail. In both cases, she won quite handily. The biggest states, New York, Illinois, Ohio and North Carolina, which are the 4th, 5th, 7th and 9th largest states by population, were all won by Clinton. When you tally the percentages based on actual number of votes, Hillary beats Sanders by a healthy 8 percentage points. Geijsel and Barrigan, in their paper, made the mistake of merely tallying the percentages of each state together and then simply dividing by the number of states, giving them an obviously wrong number.

But Charnin, when he did his calculations, took a slightly different tactic. He correctly weighted the totals based on the percentage of the actual number of votes, but he did so using the exit poll respondents to weight his averages instead of the actual vote totals. This gives a percentage of 47.4% for Hillary and a 52.6% percentage for Bernie.

Meaning he saw the mistake, and refused to comment or correct upon it. Why? Because it agreed with his own conclusion, of course.

Charnin's results still show an apparent win for Bernie. But unexpectedly, he blew the lid off any purported reliability of unadjusted exit polls, because by putting the two "studies" close together, we see a problem. The pre-adjusted exit poll data does not match in each "study." For example, in Alabama, Geijsel and Barrigan showed 73.16% support for Hillary, whereas Charnin showed only 70.6%. In Maryland, the G&B numbers showed 65.64% in pre-adjusted poll numbers for Hillary, while the Charnin numbers showed only 63.8%. In Wisconsin, the G&B show early exit polls were 37% in favor of Hillary, whereas Charnin showed that they were 43.5%.

Just whose unadjusted exit poll data is right, here?

Conspicuously, 37% was also the exit poll number given for Arizona.

Wait a minute, Arizona? I thought Arizona didn't have any exit poll data.

That's right! Arizona didn't have any exit poll data! Arizona's figure of only 37% support for Hillary, a state she won with 57.6% of the vote, is a number which Geijsel and Barrigan completely pulled out of their ass! That absurdly low number then was used to pull Hillary's percentage down to one that seemed to show a clearer win for Bernie.

The 37% assigned to Wisconsin was also probably pulled out of their asses. You know, if you're going to cheat while accusing someone else of cheating, at least pick a different number! Or use one with a fucking decimal!

Charnin must have noticed this. But he didn't care. In his spreadsheet, which he foolishly allowed people to see here, you can clearly tell he used the Geijsel and Barrigan numbers faithfully, knowing damned well they were wrong!

Conspiracy theorists are apparently of a unique mindset. They will play along with other people's bad information if they feel that it will gain their cause additional exposure, if not credibility. So Charnin went along with the antics of Geijsel and Barragan, noticing their mistakes and yet ignoring them, using their bullshit numbers and ignoring that too.

If he's that shoddy with the numbers he got from Geijsel and Barragan, just imagine how shoddy he is with his conclusions.




Snug the Head said...
This comment has been removed by the author.
Snug the Head said...

According to this Reddit post
The AZ data comes from a local newspaper called the Daily Courier but was only done in one county. It has made it's way into Charnin's data although his graph shows the margin of error at ~27%.