alexh.bsky.social
Science-ing, and trying to improve science. Cognitive and perceptual psychologist, and meta-scientist.
Biases include @simine (💍), cats (🚫)
Mastodon: @[email protected]
9,454 posts
5,569 followers
1,544 following
Regular Contributor
Active Commenter
comment in response to
post
ok this means i can still talk to u about bayesianising
comment in response to
post
Example of Wikipedia's influence and the shortfall of expertise contributed substack.com/@statesofexc...
comment in response to
post
It's also a full-fledged academic journal en.wikiversity.org/wiki/WikiJou... By the way, we are also looking for more academic editors, get in touch if you’re interested.
comment in response to
post
We will get your article peer-reviewed and if accepted, we will publish it and (under many circumstances) will replace the existing Wikipedia article with your revised article. The WikiJournal of Science is a Wikipedia Foundation initiative designed, in part, to bring more expertise into Wikipedia.
comment in response to
post
onlinelibrary.wiley.com/doi/10.1111/...
comment in response to
post
True that a lot of the responses reflect generalized genAI hate, but it is a problem with the blog post that they didn't address whether data will be shared with Google and other details of the relationship.
comment in response to
post
I think it's time to switch up the comparison, from the Stasi to ICE... for a more contemporary nightmare.
comment in response to
post
True, we had to wait for Cass Sunstein for that.
comment in response to
post
ok but you forgot the Strava link
comment in response to
post
Can you give me some intuition for why, with fixed power (80%), your graph shows the proportion of fragile p-values increases with effect size up to 40% (for implausibly-large-for-psychology effect sizes).
My naive intuition was that effect size does not matter, only power.
comment in response to
post
Thankfully, there are alternatives to proprietary LLMs. Check out this amazing resource that evaluates dozens of LLMs on transparency, accessibility and more! opening-up-chatgpt.github.io (also note how poorly GPT does on these metrics)
comment in response to
post
Agree that in many countries, many researchers may depend heavily on an impact factor. So it's great to see that Imaging Neuroscience if thriving nonetheless! bsky.app/profile/imag...
comment in response to
post
hey , thanks!
comment in response to
post
Not to mention that many are young, too young for an impact factor, or choose not to apply for one (or are discriminated against by Thomson ISI? And they certainly don't try to game their impact factor as much)
comment in response to
post
Rulers! The crown and throne
Are given to you by the Law — not by nature
;You stand above the people,
But the eternal Law stands above you
You stand above the people,
But the eternal Law stands above you.
comment in response to
post
And to bring this to your original point, I think it helped break up certain self-contained, circular don't-question-each-others-practices communities.
comment in response to
post
About other plausible causal mechanisms, traditional diffusion happens through academic publications and conferences but I think that predicts a slower rate.
comment in response to
post
Twitter was interdisciplinary and played a huge role in spreading open science across communities. Although I have no evidence for that to satisfy a causal inference researcher haha! I'm just going on a prioris, post hoc ergo propter hoc, anecdata, and the lack of other plausible causal mechanisms.
comment in response to
post
Sample sizes have definitely gone up due to online testing, which can mean worse studies validity wise, yes! Although that still makes it harder to p hack.
comment in response to
post
The usethis package makes it pretty easy! But then one can still get confused if you need to modify things until you fully understand what packages need.
comment in response to
post
Hm. Now I want to look up what millenium(), era(), and age() return
comment in response to
post
Earthlings are, in *reality*, mostly harmful.
comment in response to
post
Great, so how about this for a reckless new social media take 😉:
the open science movement, far from creating the framework for the Republicans (simply with naive good intentions), instead the open science movement was imitating the Republicans from the very beginning!
comment in response to
post
To be clear, I still advocate for it and think it's crucial to self correction in psychology and other sciences and to building trust in science (long term; in the short term in the US everything's headed down the toilet).
comment in response to
post
Of course there are deeper roots to everything, but I didn't know the (separate) Republican agenda pre-dated this.
I found Jean-Claude Bradley's vision and practice very inspiring, which is largely what led me to perennially advocate for it on social media from Friendfeed through to twitter.
comment in response to
post
There’s also a rule in the Council that no resolution can be debated on the day it’s first proposed…Otherwise someone’s liable to say the first thing that comes into his head, and then start thinking up arguments to justify what he has said, instead of trying to decide what’s best for the community.