Described as an "Election data sicko" by Hellgate NY
Polling/methods @hitstrat
tweets my opinion alone. Should tweet fewer jokes
Oth socials at website
Sep 7, 2023 • 11 tweets • 3 min read
The implied criticism is just totally wrong.
Oversampling a group in your data collection doesn't mean they are over represented in the topline or crosstabs.
When you oversample you adjust the weighting so that the group you oversample still makes up the correct % of the total
For example the US is roughly
59% White
19% Latino
13% Black
9% other folks
So if I did a poll with 1000 sample size I would expect about (est MOE for each group in ()):
1000 total (3.1%)
590 White folks (4%)
190 Latinos (7%)
130 Black folks (9%)
90 other folks (10%).
Oct 25, 2022 • 5 tweets • 1 min read
Stat blowing my mind.
Among current reg voters there were ≈64,000,000 3 of 3 general voters going into 2018 (16,14,12)
Among current reg voters there were ≈94,000,000 3 of 3 general voters going into 2022 (20,18,16)
That's a 47% increase. 🤯
Jake you might say, 2014 had terrible turnout, of course it's huge growth!
Among currently registered voters
Going into 2018 2 of 3 voters (16,14,12): ≈102,000,000
Going into 2022 2 of 3 voters (20,18,16): ≈129,000,000
