Twitter | Search | |
Ruben C. Arslan
Bayescurious evidence enthusiast. psychology, genetics, open science, R, , . Interests: mutation, ovulation, evolution, IQ, risk
7,145
Tweets
1,361
Following
2,770
Followers
Tweets
Ruben C. Arslan 4h
I've always put complete lists in online supplements, but of course those don't get indexed.. in my defense I use a lot of packages... it also would be nice to give credit to packages under the hood (i.e. dependencies of the ones you call).
Reply Retweet Like
Ruben C. Arslan retweeted
Richard McElreath 19h
Half of experiments in 3 top political science journals introduce bias by conditioning on post-treatment variables?! (If there were a similar survey for biology, I expect it would be at least this bad.) The paper:
Reply Retweet Like
Ruben C. Arslan 5h
Replying to @nickchk @elbersb
I didn't know this package. It is really cool . It doesn't specifically mention duplication in the way I imagine would help my students (will have to see about it of course), but it seems extensible? Link for others who follow along
Reply Retweet Like
Ruben C. Arslan 8h
Replying to @pegleraj
Oh, yeah it's always unanticipated messiness in the data like what you mention. I just feel like that these problems might be easier to discover if people "fail" earlier.
Reply Retweet Like
Ruben C. Arslan 8h
Really? Does anyone ever want to duplicate 2% of the rows on the left-hand side? I mean of course there are legit reasons to do so. But I think my students (and I) would make fewer mistakes if there was a warning for it. But yeah, a lot is specific.
Reply Retweet Like
Ruben C. Arslan 9h
Not to confuse with 's fuzzy joins of course ;-) But I feel like this probably exists somewhere and I just have weak Google-Fu today. do you know anything?
Reply Retweet Like
Ruben C. Arslan 9h
Reply Retweet Like
Ruben C. Arslan 9h
I'm procrastinating so I wrote a little helper that predicts duplication to investigate overlap before joining (in RStudio viewer). Maybe more useful would be *_join functions that warn you more often (training wheels)..
Reply Retweet Like
Ruben C. Arslan 10h
Replying to @ludmila_janda
both news to me (I mean I know base setdiff, but I hadn't realised dplyr supplies a generic)!
Reply Retweet Like
Ruben C. Arslan 10h
Replying to @dccc_phd
Yeah, that's what I do, but when merging on multiple variables etc. and with students I could use some helpers that simplify checking.
Reply Retweet Like
Ruben C. Arslan 10h
Replying to @ianhussey
Yeah, that's kind of what I do (once I stumble), but then I thought maybe someone has already baked this into a utility package for learning or something like that.
Reply Retweet Like
Ruben C. Arslan retweeted
Felix Henninger 11h
The browser is a fantastic tool for behavioral science — thanks to generous support from , we're giving it the ability to interact with external biomedical sensors, too. Couldn't be more psyched to work on this! 👇
Reply Retweet Like
Ruben C. Arslan 11h
Replying to @dccc_phd
It's great, but doesn't really help avoid accidental duplication (which is mostly about wrongly expecting the data on RHS to have a different structure than it does).
Reply Retweet Like
Ruben C. Arslan 11h
Replying to @rubenarslan
Or functions that error when e.g. merging in the right-hand side would lead to duplication of rows, or when a merge would introduce a small number of dupes (likely unintentional)? This happens so much (to me too).
Reply Retweet Like
Ruben C. Arslan 11h
Tweeps, what do you use to 'debug' joins/merges? Common problems my students run into: - unique ID doesn't stay unique (accidental dupes) - unexpected setdiffs (missing from RHS) - finding source of dupes Are there more explicit ways to teach this? Preferably with dplyr.
Reply Retweet Like
Ruben C. Arslan Jul 17
It's nice, but searching turns up mainly the statcheck auto comments in psychology, so it's hard for me to even find the substantive critiques in psychology that I know of.
Reply Retweet Like
Ruben C. Arslan retweeted
Dr. Dan Vahaba 🧠 Jul 16
The mammoth that is my last chapter of my dissertation from is up on ! Estrogens, such as estradiol, are generally thought of as enhancing adult cognition. BUT few studies have asked if estradiol is imp for *developing* animals for other learning 1/n
Reply Retweet Like
Ruben C. Arslan Jul 17
Replying to @siminevazire
Fair
Reply Retweet Like
Ruben C. Arslan Jul 17
Surprising to me, but I'm no expert. ?
Reply Retweet Like
Ruben C. Arslan Jul 17
Replying to @siminevazire
I have no disagreement with the substance of this tweet, but you're getting your wisdom on corrupted systems from Brooklyn 99 instead of The Wire? That's no way to live
Reply Retweet Like