John Flournoy @flourn0@fediscience.org

Posting from the persona of my day job as a research scientist studying #adolescence, early life #adversity, and #cognition using intensive #longitudinal designs and #fMRI.

I'm mostly a methods and scientific #computing nut with a strong streak of scientific #critique and desire to improve scientific practice including #OpenScience and #philosophyofscience.

I think a more just society is possible. I think improving adolescent #wellbeing depends on expanding opportunities.

#introduction

How can we improve inference and replicablility in #DevCogNeuro? Let's 1) strongly demarcate confirmatory NHST from exploration; 2) reward explicitly exploratory work. We offer a set of tools and strategies we think are well suited to how many in #DCN do their work https://psyarxiv.com/ez5sf

Measurement error reduction over time, from https://www.dw.com/en/james-webb-space-telescope-full-color-images-will-wow/a-62236601

Unfortunately this assumes they were acting in good faith.
---
RT @AOC
We should hear from Senators Murkowski, Collins, and Manchin if they believe there should be any consequences at all for misleading members of the US Senate in order to secure a lifetime appointment to the highest court in the land.

Otherwise Roe’s undoing is their legacy too.
https://twitter.com/AOC/status/1540462266256171009

"The most impactful thing you can do at the moment is to donate money. I know, I know. If that isn't an option or if you want to do more, organize other people to donate money."
---
RT @RHAVote
We've revised our "What You Can Do" doc, which you can share with friends and family: https://bit.ly/RoeWYCD #RoeVsWade
https://twitter.com/RHAVote/status/1540364558304428041

Read the thread. just because the votes might not be there right now on the court doesn't mean there isn't a strategy that will f us up.
---
RT @BrynnTannehill
Here's what I'm hearing regarding Obergefell:

* It seems that at least one of Kavanaugh or Gorsuch really does believe that Obergefell is different because it doesn't involve "life". Votes to overturn it may not be there.

* Same with Griswold, Eisenstadt, & Lawrence. 1/n
https://twitter.com/BrynnTannehill/status/1540670958213767168

---
RT @GhawinRiver
See, I get hating that those are your only two options and noping out of electoral processes because of misplaced nihilism and defeatism. But what I can't get is looking at those two and going, "how do I choose it's too hard."
https://twitter.com/GhawinRiver/status/1540397449344335885

⁣

https://www.npr.org/2022/05/05/1096636861/democrats-votes-abortion-rights

This is so good.
---
RT @GhawinRiver
Y'know, I understand non-voters. But what I'll never understand is both (a) "swing" voters and (b) why people don't hate them as much as if not more than non-voters. Apparently, the fate of the world rests upon non-committal goobers from Ohio, Michigan, and Florida.
https://twitter.com/GhawinRiver/status/1540397438724440067

Cc @dccc
---
RT @NeilLewisJr
Thinking about this passage again this morning. From How Democracies Die.
https://twitter.com/NeilLewisJr/status/1540349684946030594

Love this opening line from Patrick Mair's Modern Psychometrics with R book:

"It is always good to start a book with a formula right away:"

Amazing thread. It's got it all: gender pay equity, causal inference fiasco, big tech.
---
RT @PHuenermund
This is my favorite teaching example for showing the importance of #CausalInference: @Google conducts an annual pay equity analysis in which they use fairly advanced statistical techniques. In 2019 they found that they were actually underpaying MEN?! https://www.npr.org/2019/03/05/700288695/google-pay-study-finds-its-underpaying-m…
https://twitter.com/PHuenermund/status/1540262891890278402

Reminder: Don't accept the null hypothesis (without evidence) and don't reason based on unsupported acceptance of the null.

It are a praxis
---
RT @lastpositivist
The forces of capital and their various imperialist enforces are going to feel pretty silly when they read my next paper, really skewered them. I expect they shall just surrender, especially once I start tweeting some zingers to go along with it.
https://twitter.com/lastpositivist/status/1539965119798468609

The chance of making a career in academia is so low, you might as well do it on your own terms.
---
RT @lakens
@NoahHaber I have some trinkets on my desk that belonged to former colleagues who for that reason no longer work in science to remind me the sacrifice is real. If it helps, when I talk to them now (for some a decade after they left), no one regretted not abandoning their principles.
https://twitter.com/lakens/status/1539692299185266691

Ahhh, some super racist get off my lawn bs
---
RT @JSOPIO
Lower your radio volume! Starting July 1, 2022, Florida State Statute 316.3045 will once again become enforceable. What does this mean? It means that you will have to listen to your radio at a volume that is NOT plainly audible at a distance of 25 feet or more.
https://twitter.com/JSOPIO/status/1539256220875464704

Intellectual property is theft.
---
RT @AP
Ohio State University has won its fight to trademark the word “The," allowing it to control use of the word on branded products associated with and sold through athletics and collegiate channels. http://apne.ws/jYXg20E
https://twitter.com/AP/status/1539750867611664395

Has anybody seen a good primer on using brms (or maybe just Stan) in a typical machine-learning setting? That is, selecting models based on out-of-sample prediction performance. Maybe this is just {loo}, but that feels too easy.

woops, should be
---
RT @flourn0
@tervoclemmensb oh yes, I was silly and left out the effects of the factors. Should have been:
gam(y ~ A_levels_12 + s(x, by = A_levels_12)) and
gam(y ~ A_levels_1234 + s(x, by = A_levels_1234)).
https://twitter.com/flourn0/status/1539711699409829894

Go work for someone that is making work better both through the work and how she organizes, motivates, and does the work. It's work all the way down really.

Seriously, if you have half an inclination, apply.
---
RT @grimalkina
Cannot even tell you how much I've been looking forward to writing this thread: I AM HIRING!!

Please share widely: come join our innovative, caring, and impact-driven research team, the Developer Insights Lab.…
https://twitter.com/grimalkina/status/1535301738978193413

In other words, gam(y ~ s(x, by = A_levels_12)) gives very different expectations for levels 1 and 2 than gam(y ~ s(x, by = A_levels_1234)).