We all know which side are the (insurrectionist and foreign agent) traitors. I think he might have finally admitted it!
9 Likes
that’s gonna take more than 10 or 15 flushes! 
14 Likes
In his own words: “People are flushing toilets 10 times, 15 times, as opposed to once.”
14 Likes
Having lived in Galena, Illinois where the Grant hagiography was strong when I was a kid, I have a very, very strong feeling that Ulysses Simpson Grant would consider his former party now a hotbed of Confederates and other traitors, and unleash the wrath of Almighty God on them if he could.
14 Likes
I guess they wiped the Biden out of their eyes.
14 Likes
As my colleague Tim Alberta has said, the theme was “strength.” Strength was expressed by exaggerated, absurd, comic-book figures: Hulk Hogan, Kid Rock.
13 Likes
… he’s beyond mockery
Even jokes about him aren’t funny
11 Likes
Why do news organizations who ostensibly know their own polls’ margin for error call a difference within the margin of error a “lead?”
Harris, whose campaign says she has secured the Democratic nomination, led Trump 44% to 42% in the national poll, a difference within the 3-percentage-point margin of error.
Nonetheless, with the inherent poll bias, I’ll take it as an encouraging sign.
11 Likes
17 Likes
… yeah, it’s sort of an anecdote at this point
I mean, all the polls are bullshit, right 
7 Likes
It certainly seems like they have built-in bias when you compare them to actual results, yes.
6 Likes
… when the official projection is, “it’s a toss-up,” well, gee thanks guys, we could have come up with that one on our own
6 Likes
As a number nerd, this makes sense to me.
The really important part: Two distributions with different means and overlapping margins of error still have different means, and we still expect one to usually give higher results.
The margin of error is usually defined as the half-width of the confidence interval. Most polls use the 95% confidence interval (look for the words “nineteen times out of twenty”).
In this case, the results are Harris: 44%, TFG: 42%, with a 3% margin of error. The 95% confidence interval is around +/- 1.95 x SE, where SE is the Standard Error.
One helpful way to look at it is to model (H - T), where H is the Harris support, and T is the support for TFG. We don’t actually know, or try to measure the actual support, person by person, so we poll and model.
Using their numbers and some wrangling, we get a graph like this of the probability of various values of H - T. The 0 line is where both have the same support. Positive numbers are where H leads, and negative numbers are where T leads. This shows a roughly 17% area under the curve where the preference is negative (T leads), i.e. 83% likely H truly leads.
Of course, all this ignores the elephant in the room - your sample will be mostly older people who still answer calls from numbers they don’t know. It’s been really hard to get a sample that represents voters for a decade or so now.
15 Likes
Ah. I’m used to clinical trials. If you don’t reach statistical significance, you haven’t shown the treatment is beneficial over the control.
11 Likes
8 Likes
For the Clicks! They gotta have a horse race.
9 Likes
Pointless performance art. She’s already on the Biden-Harris ticket that the funds were raised for.
Now, if the FEC were to take a hairy eyeball to Trump’s election finances…
17 Likes
