Will Lowe

Amy A @lolennui.bsky.social [ View ] Jul 03, 2024 at 15:22 UTC

I think it was Lenin who said, “there are decades where nothing happens, and then there are weeks where holy shit what the fuck”

8 replies 69 reposts 460 likes

Will Lowe @conjugateprior.org [ View ] Jul 03, 2024 at 12:06 UTC

I never do. And if variables arrive in a data set with labels I take them off.

0 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 03, 2024 at 08:29 UTC

0 replies 0 reposts 2 likes

Will Lowe @conjugateprior.org [ View ] Jul 02, 2024 at 14:09 UTC

Just used the personal one, eh?

0 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jul 02, 2024 at 13:57 UTC

Indeed, the correct term is "skype". Because it's not like anyone is using it for anything else at the moment.

0 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 02, 2024 at 13:01 UTC

Institutional, yeah. As does CBS, no?

1 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jul 02, 2024 at 11:31 UTC

Interesting. I'm surprised. For calibration, does this also seem compelling?

www.ft.com/content/e577...

1 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jul 02, 2024 at 10:12 UTC

Measurement models do seem quite rarely taught. isbnsearch.org/isbn/0367612... is... ok. But I think the psychometricians have good books.

imho polarization is hard because, when not 'affective', it mixes an ideal point measurement problem (harder than usual) with a cross-level causation question.

1 replies 0 reposts 4 likes

Will Lowe @conjugateprior.org [ View ] Jul 02, 2024 at 09:38 UTC

That paper is... wow. A compositionally changing subgroup of a compositionally changing social media population gets measured with tendentiously named and GPT-constructed indexes of who knows what. Story time with Noah is pretty clearly where it belongs.

1 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jul 02, 2024 at 08:57 UTC

. o O (the main effect or the interaction effect?)

0 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 17:43 UTC

well that was a goal best witnessed in a Paris bar

0 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 14:06 UTC

<insert creeping dread about here>

1 replies 0 reposts 2 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 14:05 UTC

Umm, they're not for making indexes, are they?

1 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 12:20 UTC

Goodness, it's all overdispersed Poisson models with offset as needed where I come from.

0 replies 0 reposts 2 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 12:15 UTC

And also when there isn't 🕺

0 replies 0 reposts 4 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 12:11 UTC

A quick question though: Where do the "Model B as a count with log link and offset(log(M)) on the RHS" people land in the marginality bestiary. Are they modeling M or not?

1 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 12:05 UTC

I have to not understand that? That's unfortunate for a budding extremist, as I do :-)

1 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 12:00 UTC

I take this to be the point of Case 2 in the Notorious Marginality Article (and not much to do with marginality)

0 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 11:55 UTC

Yeah, I guess I'm a marginality extremist. Except that 'higher' and 'lower order' terms is a syntactic distinction that doesn't really matter. Not losing degrees of freedom down the back of the couch is the main principle, parameterization be damned.

1 replies 0 reposts 3 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 11:49 UTC

1 replies 4 reposts 8 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 11:41 UTC

index goes brrrrrrrr

1 replies 0 reposts 5 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 10:39 UTC

Maybe not quite what you're looking for but I did attempt to reconstruct an index measure analysis in a way that made more statistical sense to me in the ERROR report I just handed in. Much of the thinking is straight Kronmal 1993 though.

0 replies 0 reposts 2 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 10:19 UTC

the institution weighted fractional citation count is a wild nearly count-ish thing

1 replies 0 reposts 2 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 10:18 UTC

all indices afaict

2 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 10:08 UTC

Rampant index construction is also a result a genuinely valuable insight: that anything can be a statistic and have a sampling distribution, so you really should define the quantity you want, not borrow a thing you saw in a textbook ...which then slips rather too easily into "Hi, this is my new DV".

0 replies 0 reposts 3 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 10:03 UTC

I don't think it's malign at all. I think the intrusive thoughts are sort of diffuse, nagging worries that what they're doing sounds quite close to the part of that thing they read a while back, didn't quite get, and hoped wouldn't be relevant to their work.

1 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 10:00 UTC

Off the top of my head: fractional citation counts in innovation research, most of the literature on political polarization, and more or less all the conflict event data literature.

2 replies 0 reposts 1 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 09:55 UTC

It's probably different for you guys, but whenever I see an index being introduced I think to myself "here's some folk who would like to measure a thing but would rather not derive a measurement model for it because that would lead to intrusive thoughts about measurement validity, DIF, and all that"

2 replies 0 reposts 8 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 09:42 UTC

Oh no... this is going to wake up the propensity scores people.

0 replies 0 reposts 2 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 08:02 UTC

0 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jul 01, 2024 at 07:53 UTC

My current project on "the void, and how to scream into it" has attracted the interest of a research support company that wants to help me transcribe what it screams back.

0 replies 0 reposts 12 likes

Will Lowe @conjugateprior.org [ View ] Jun 30, 2024 at 09:54 UTC

Random good examples: Sisters of Mercy cover Hot Chocolate's 'Emma' (it is overwrought tragedy, so what if we sing it like that?) or Laibach's cover of Queen's 'One Vision' (see how these lyrics tell you what they are when we switch them into German?). And bad ones: the dance mix of almost anything.

0 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jun 30, 2024 at 09:50 UTC

Cover versions are perhaps the the more legitimate form of this as imho they are at their best "here's a something buried in the original that I am well placed to put in the foreground" rather than "form the power set of these elements & I'll pick a combination I like". Something has to be brought.

1 replies 0 reposts 0 likes

Will Lowe @conjugateprior.org [ View ] Jun 30, 2024 at 08:49 UTC

I never knew it had a name, but once you see the low effort recombination gesture "style transfer" you see it everywhere: from "Uber but for dogs", to AI Art slop, to real contemporary art galleries full of 'painting of a bird, but embroidered' and Thing for Doing X Doing Y. Just dull, dull stuff 😞

0 replies 1 reposts 2 likes

Will Lowe @conjugateprior.org [ View ] Jun 29, 2024 at 15:18 UTC

So you're a taxidermist, that's someone who - eh, never mind. So every bee has to be mounted on a little stick & you want to use as short sticks as possible and - what? yes, of course they're dead. It's an analogy, ok. So, we've got dead bees on little sticks in a box and - wait, why are you crying?

0 replies 1 reposts 2 likes