The raw effects show GPA normalization after sharp decreases in GPA in the semesters leading up to the effect, not sharp increases in GPA. The apparent sharp increase in Figure 7 is the effect relative to the sharp decrease, not inflation relative to a static baseline.

A student's grade will reflect their ability, resources, and disabilities. If grading is meant to reflect the first, then being hurt in the 3rd column should be made up by adding to the 2nd column. Restricting resources is *itself* the artificialization of the academic world v. workplace.

Admissions aside, regarding disability accommodations: I think grading in general cannot be treated unconditionally. Arguments against accommodations under-appreciate that accommodations—especially access to self-provided resources—are common in the workplace, and test-like constraints are not.

bsky.app/profile/mtsw...

^wokest*, I guess

This requires using grading as honest feedback and ensuring students have pre-requisite knowledge. But school is where you go to learn, and even non-excellent students can learn.

You want my woke-ist opinion, it’s that admissions to higher education should consider only whether or not a student will be able to succeed in *completing* the degree with the resources available to them from the school, and no further ascertainment of relative skill is needed.

See also admissions based on excellent performance on standardized test scores (who gives a shit?)

It’s also just incoherent. These are very young teens that could pass as younger but also seem legal? Huh? Like what does it mean to be, say, a 15 year old that could pass as *younger* while simultaneously *seeming older* to others? Passing as younger means *seeming younger*.

Nonce defenders, ffs

Ohhhh welp, yeah that’s not good.

Also yikes! At best there are processing steps that were not adequately described; I would never present a very strangely rounded dataset just in Excel, that’s so sloppy. At worst it’s manipulation, yeah.

bsky.app/profile/rhac...

Averaging 1.2 and 0.4 (say) would get you 0.8. If you were using whole integer rounding in the original columns and half-integer rounding in the averaged columns, that would look like averaging 1 and 0 and getting 1. But again that’s very strange if that’s what they did!

The second row has averaging 0 and 1 to get 0.5. The other example at row 21 is odd—is it possible the value in the two left columns, unrounded, could be above 1?

(It’s possible the above image is from a CSV that was merely opened in Excel.)

I haven’t read any of it so cannot comment. Do they have R scripts or something like that? Or is it all Excel?

It is possible the chosen resolution for the averaged columns was higher than the resolution of the former to account for averaging. It’s weird, and I wouldn’t do it, but the reason could be that anodyne.

Also some averaging of 0’s to get 0.5.

To blackmail Trump, too.

One of the emails explicitly says they would plan on blackmailing Trump. It’s also not just rewarding the behavior as if Epstein was his friend, but being in debt or trying to cover up.