Sercan Kahveci

One of the adaptive designs we are seeing more frequently is the Sequential Multiple Assignment Randomised Trial (or SMART trial) 1/7
#MethodologyMonday #126

it is astonishing how many people think that a vegetarian diet is unaffordably expensive

think I just found a way to linearly model which person-level variables increase or decrease sample-wide reliability... very cool

Qualpal: Qualitative Color Palettes for Everyone Larsson, J., (2025). Qualpal: Qualitative Color Palettes for Everyone. Journal of Open Source Software, 10(114), 8936, https://doi.org/10.21105/joss.08936

💡 Came across this nice tool today:

🎨 qualpal for algorithmically choosing maximally distinct colors under certain restrictions #dataviz

JOSS paper, online tool, R package #rstats

joss.theoj.org/papers/10.21...

"Among articles stating that data was available upon request, only 17% shared data upon request."

Now he has doxed me too. 2 days since the conference has ended and I am still shaking. First stop tomorrow: the police station.

I still have pain in my arms and I can't relax. I keep waking up nervous and checking twitter to see what slander he has posted about us now. Sometimes I need to remind myself it's over and I don't need to hyperventilate. This isn't what academia should be like. It's an utter disgrace.

he recorded all of us and posted it to youtube. And worst: the police did not take him. It was relentless. I could not speak to many guests or attend most lectures I wanted to see. The times he wasn't around, I mostly spent trying to calm myself down and drink water to recover from confronting him.

Our conference this week has been my most stressful week in academia ever (13 years). There were constant intrusions into the event space by a serial harrasser who was banned from the building. I had to watch the doors and drag him out of the space all the time.

That senior phd student should be named and shamed, no way this can be allowed to pass unchallenged.

sound ridiculous? shit like this is the reality for most people falling outside a hegemonic identity category

For Viagra we can imitate transgender "care" as it's done in many countries
1) 10+ mandatory counselling sessions
2) Mandatory trial period in which the patient must try to function with the ED
3) Waitlists a decade long because only 1 or 2 clinics do the above
4) and then you miiight get the pill

Not completely implausible, epigenetics is acknowledged to be a thing now. In fact, Darwin's finches have been shown to do their rapid evolution thru epigenetic de/activation of genes rather than thru slow natural selection of gene mutations.

In one of my recent papers I went a little overboard in the opposite direction. There are 4 or 5 paragraphs of future directions in there. It's nicer than delivering that useless oneliner but the paper also became wayyy too long.

Will things feel better in the morning? A time-of-day analysis of mental health and wellbeing from nearly 1 million observations Background Mood is known to change over seasons of the year, days of the week, and even over the course of the day (diurnally). But although broader mental health and well-being also vary over months ...

Cool study tracking mood across the day - confirms that mood is best in the morning and worst in the evening; and also better in summer and autumn than in winter and spring. Would have loved to see analysis of subgroups (age, chronotype) and of dispersion

mentalhealth.bmj.com/content/28/1...

What the fuck

We've known for 3 decades now that the premotor cortex also sends axons down the corticospinal tract, and I don't see that represented here.

I look up data quite often, one time I even did additional analyses on the data of a study because I was curious about an effect that the authors did not look into but which you could see in the graphs.

Conference of the Association for Cognitive Bias Modification (ACBM)

Present Perfect Continuous The Present Perfect Continuous tense, also known as the Present Perfect Progressive tense, denotes an action that began at an unspecified time in the past and has continued up until the present moment. This tense is formed by combining the present perfect tense of the verb “έχω” (to have) with the present tense of “να” along with the verb in its present participle form. Example: Έχω διαβάζοντας το βιβλίο για μία εβδομάδα. (I have been reading the book for a week.)

AI literally has made us unable to communicate/search for stuff, e.g. "Greek present perfect continuous":

1) Greek does not have this tense, and yet it's the top result on Google.

2) να is not a verb, "Έχω διαβάζοντας το βιβλίο για μία εβδομάδα." is nonsense.

Check out my R package for computing permutation-based split-half reliability here :)

Woah! This is a very different style than before :o

Just out! Our peer-reviewed critique of the Cass Review has been published by BMC Medical Research Methodology. Please read and share. We show that the Cass Review is fatally flawed and should not be the basis for policy or practice in transgender healthcare.

link.springer.com/article/10.1...

This is just one way in which AI poisons the well for everyone, including non AI users.

That said, I see a lot of accusations on social media of people (especially artists) using AI. In reality often the person just... isn't so skilled, and it is just badly drawn art or written poetry. A student who knows R & sees the slide might feel they must use readr to evade that accusation

I have this issue at my workplace as well. I'm often called over to troubleshoot AI code that the person has no understanding of to begin with. IMO in science one should never trust any code that one cannot explain from start to finish, otherwise the result is more often than not wrong.

And I hope I don't need to say why one would put na.rm=T inside mean()

Likewise, I always use .groups="drop" inside summarize() because I don't want to trigger the message that dplyr grouped the results by one of the variables - sucks when this occurs hundreds of times over a script. I want it to output only what is relevant to me - errors, progress reports, and output

I do most of the things that you say are signs of AI-based coding. There is good reason to use %>% over |> since the . thing that comes with %>% pipes can be more flexibly applied than the _ thing that comes with |> pipes. I also usually comment on every line to make sure future-me can understand it

Thats a very cool idea. You would be able to identify equivalent of a known variable in another dataset without it being labelled. Would surely require huge samples though given how unstable network structures tend to be