Read on Twitter
Let's play a little game together that I play every morning.
I call it: POLLING IS TRASH.
You ready?
(Thread coming...)
I call it: POLLING IS TRASH.
You ready?
(Thread coming...)
2/ So I usually go to Real Clear Politics and click on the POLLS section on the top nav bar.
This page aggregates most of the public polling made available by news organizations.
A handy resource for politics nerds like me...
This page aggregates most of the public polling made available by news organizations.
A handy resource for politics nerds like me...
3/ Today, I go to the page and what do I see...
Trump down by 3
Trump down by 9
Trump down by 10
Trump down by 10
This narrative lets the media crow "Trump's IN TROUBLE!"
It's designed to mislead the electorate, who prefer to be in the majority. Human nature, etc.
Trump down by 3
Trump down by 9
Trump down by 10
Trump down by 10
This narrative lets the media crow "Trump's IN TROUBLE!"
It's designed to mislead the electorate, who prefer to be in the majority. Human nature, etc.
4/ So let's drill into it a little.
I picked the CNBC Change Research poll first.
You find the crosstabs, and you get to a PDF. It's 25 pages long:
https://changeresearch.com/wp-content/uploads/2020/10/CNBC-CR_National_Toplines_Wave-16_October-17-18-crosstabs.pdf
I picked the CNBC Change Research poll first.
You find the crosstabs, and you get to a PDF. It's 25 pages long:
https://changeresearch.com/wp-content/uploads/2020/10/CNBC-CR_National_Toplines_Wave-16_October-17-18-crosstabs.pdf
5/ So the first thing I do is I scroll to the VERY BOTTOM of the poll to look at who they speak to.
And voila, what do we find?
It skews left, duh! But wait... there's more...
And voila, what do we find?
It skews left, duh! But wait... there's more...
6/ It breaks down even further than just "how did you vote in 2016" which isn't a very reliable metric.
False recall can drastically affect those numbers. Usually in FAVOR of the incumbent. So if 44 said Hillary & 40 said Trump, the real Trump number is likely to be even lower.
False recall can drastically affect those numbers. Usually in FAVOR of the incumbent. So if 44 said Hillary & 40 said Trump, the real Trump number is likely to be even lower.
7/ Now in that graphic above, you see the party breakdown.
In total, 50% of those polled are Dems of some persuasion. Just 40% are Republicans.
In total, 50% of those polled are Dems of some persuasion. Just 40% are Republicans.
8/ The "Party ID" bear this out too. Bear in mind there's a small margin of error here...
8 pts more Dems than Republicans, which is like a 22% distinction...
8 pts more Dems than Republicans, which is like a 22% distinction...
9/ And finally, the allocation reveals the poll to be virtually useless.
They quizzed 10pts more Dems than Republicans.
What's Trump down by in this poll?
10%.
And there you have it. Have a lovely day and don't believe the fake polls.
They quizzed 10pts more Dems than Republicans.
What's Trump down by in this poll?
10%.
And there you have it. Have a lovely day and don't believe the fake polls.
