Veli-Matti Karhulahti

Sometimes it feels science isn't that hard after all, all one needs to do is think hard for 5 seconds

I'm aware of pilots/plans where applicants have an option to submit their plan as stage 1 draft, this makes sense imo (raise awareness etc) but nuance is so important in such changes

This is such a key hermeneutic for older texts, there's always stuff going on behind the scenes, the thing that's being built on

Do you know if there's an English version available or coming out?

meanwhile, hoping the funders' own publication portals and diamond venues solve the ACP issue (as some already do to some degree)

I don't know anyone who has worked deeply with RRs (authors/editors/reviewers) who'd support something like this for so many reasons-- a friendly reminder that RRs are a tool for certain scenarios, definitely underused, and like any other tool, to be used wisely

(0.1%>2.3%>12.8%>22.7%) failed first control, corresponding to having problems "never", "sometimes", "often", "almost always"

We likewise found a steep increase in failed controls (0.1>2.3>12.8>22.7) the more severely ppl replied to a single-item tech problems question-- this could be intentional bad responding too but nonetheless critical to be aware of

Many EU/US ppl say they must keep submitting to these publishers bc it'd be unfair for co-author students not to-- fair enough sometimes, but if one is a student in a major uni and/or working for a known lab, they've already got a top 1% global advantage without prestige publishing, they'll be ok

i understand it's difficult to run large data collections like this (especially in HBSC) but that's exactly what's wrong today: brute force large datasets with whatever measures & then think later if any of the investment was worth anything

IGDS has been considered one of the better-functioning measurements, with satisfactory reliability and validity when compared to other measures (Anthony, Mills, & Nower, 2023; Karhulahti, Martoncik, & Adamkovic, 2023

7 symptoms partial, 1 symptom missing, 1 symptom fully content valid

funnily enough, the paper says the scale (IGDS) is one of the "better functioning measurements" and cites our paper-- in our paper we actually found only 1 of 9 symptoms measured in a content valid way 🫠 (also curious how translation across 12 languages took place)

it can be interesting to add multiple items even when they don't signal problems (especially if we take networks seriously) but the 2013 symptom list, sketched in dsm-5 appendix, is way outdated and never worked tbh

Average number of symptoms is 3 in boys both (almost) every day and (almost) never groups. In girls the (almost) never group has 1 symptom on average.

New WHO-collaborated (HBSC) study, n=44k, finds that boys who never play games have the same amount of gaming disorder symptoms vs those who play daily-- it's 2025, what are we doing? measurement?

a) avoid Finnish food
b) many nice museums, Villa Gyllenberg & Didrichsen worth a visit for the island space alone
c) best coffee: Päiväkahvibaari 1 (vallila)
d) library Oodi
e) saunas, Sompasauna 24/7 is classic (recently moved tho, not sure how good the new location is)

Can they argue the ad is for single-player as long as multiplayer not mentioned? (assuming some purchases will be offered in single mode too tho)

📢 Register for the 5th Helsinki Initiative webinar (8 December) on Multilingualism in Scholarly Communication with presentations by @tatsuya-amano.bsky.social, @karenstroobants.bsky.social and Andre Brasil!

More information and registration: www.helsinki-initiative.org/en/events/5t...

One of my all-time fav rants on this topic-- especially love this figure demonstrating how expert clinicians fail to agree on major depression diagnosis most of the time (57%) based on DSM5 field trials

currently editors (handling tons of papers) must heavily trust reviewers as they cannot be experts in everything-- a move toward more distributed editorial labor (in exchange for less reviewing) expects more human scrutiny from topic-fit editors who'd then also manage less papers on average

The reason why this is interesting & maybe even promising is: it isn't simply "less human scrutiny" but but a shift from reviewer trust to editor trust--

attention to alternatives is good as it contributes to gradual, slow changes that over years (decades) can lead to system level changes too-- but it's those institutions that offer alternatives which need to become more sustainable, visible, and "prestige" for progress to keep happening

Having a reward like this is ok to increase visibility of alternatives but it also struggles to address the real problem: the system remains broken bc contracts (=lives) of many ppl globally depend on publishing in metrics journals, they don't have a choice

the right solution would be ofc to go back to the drawing board and figure out what's the state of art in theory & practice, but the structures we have don't allow it: authors need to get their paper out to satisfy the funder who gave money to do the flawed test 🫠

As RR editor/reviewer I've linked this paper to authors many times when they start with a 12-hypotheses testing plan-- alas, it isn't merely a H-testing issue but usually reflects how research programs are broken deep down, trying to ask RQs that simply cannot be answered by any effect size

Thought about this plenty over the last years; imo it's already a huge step to actively reflect on it-- grey areas will always be massive & impossible to justify clearcut lines, but openly disclosing humility for effect meaning immediately increases my trust for authors/results

I've met some of the folks running it & they seemed professionals with shared values-- planning to submit my own next ms there ( they don't have much marketing power so it's a clear tradeoff for less reach but i can afford it at this point)

Tampere University Press | Tampere universities Tampere University Press offers Tampere University researchers, as well as researchers working in other research organisations, the opportunity to publish in Open Access without Book Processing Cha...

www.tuni.fi/en/library/t...

Many ppl think of registered reports as a tool for bias control (for good historic reasons) but ime this is the truly useful part of RRs: get feedback on the *design* before it's too late

--not specific to h-testing but applies to any kind of data, method, or study in general

Cartoon: simple black and white line drawing. A couple in a living room. The man is sitting in an armchair reading a newspaper, head turned slightly towards his wife who is standing at the window, tugging at the curtain and saying "I see the Renegade Master's back once again with the ill behaviour".

My most popular cartoon by a long chalk is also oddly niche, in that it mostly appeals to a generation who remember the song it refers to and enjoy the nostalgia for their youth. I can draw one for you if you like.

www.worldofmoose.com/collections/...

yeah, how they highlight working with original authors may also imply resistance toward work that doesn't-- in theory, the eLife model could be optimal for replication publishing as it does offer a platform for voicing multiple different interpretations simultaneously

I agree with the post but my guess is the mechanism will be the opposite: selection bias will favour failed replications, as they represent the more interesting results in this case

The #1 psychiatry journal, world psychiatry, also doesn't allow preprints