Wednesday, June 01, 2016

Please make ME stop!

Just noting the narrative, where context is all. has a new poll up for California.  Granted, it's a YouGov poll, and it only gets a 0.12 weighting in the current reckoning at FiveThirtyEight for the California primary, which is lower than the 0.23 that the PPIC poll every news report now cites (the one showing Clinton and Sanders splitting the state by 2%).  The YouGov poll puts Clinton ahead in California by 13%.  The SurveyUSA poll (which FiveThirtyEight gives 1.00 weight), has Clinton ahead by 18%.

So which poll is the one all news reports are still talking about?

This will play into the narrative that the Democratic primaries were rigged because exit polls.  No, seriously, that's the new frontier in Bernie support:  he wuz robbed!

Again, after every election somebody cries "FOUL!" and there may even be some basis for it.  I don't like voting machines, period; but then again, paper trails can be corrupted as easily as digital ones, although either would require a massive conspiracy on the order of what Mulder and Scully fell into.  Which is why I don't accept them based on one analysis, and why the burden of proof is on the person making the allegations.

Of course, if exit polls are so completely accurate, why does the only exit polling company in America do this?

When the polls close in each primary, media organizations release the full exit polling data. Afterwards, Edison Research compares them to actual voting results and makes adjustments. For instance, the exit polls initially showed a 4 point win by Hillary Clinton in New York at 9 PM, but were changed to 12 points at 9:45 once Edison obtained voting results.
One set of data from the source is sound, the argument goes, and the other set isn't.  In the real world we call that "cherry picking."  But then one set of numbers makes your argument, and the other doesn't; so what're ya gonna do?  Which puts the argument back to ignorance:  Bernie supporters prefer the first set of numbers over the second, so this proves there is voter fraud.  But if the first set of numbers is accurate, why does the polling company adjust them?  Because Hillary?*

In the meantime, we're all still talking about a California poll which is probably an outlier because  "Horse Race."  But if Hillary wins California, it will be more proof of corruption and voter suppression, especially if she wins by more than 2 points.

I'll retire to Bedlam.

*There are more general reasons to ignore exit polls, too.

No comments:

Post a Comment