Ok pals. Looking to move all my teaching materials open and online, and am developing new modules that I want to start off this way. Content = reading, quizzes, videos, and code tutorials. Opinions on the best platform? Good examples of best practice I can steal from? Many thanks.
the goat
Though I should note that I, in general, don't care for change scores. Nevertheless, they can be useful in some cases.
In an RCT, pre/post differences *within* groups are not themselves causal effects. You have confounding with history. However, one can use change scores to compute the causal effect of treatment *between* the groups. It’s all very tricky and easy to confuse.
Has anyone tried teaching *only* Bayesian inferential #stats for undergrad psych (or other UG sosci programs)?
Like completely foregoing NHST, confidence whatevers, p values, CLT...
I'd be very interested in seeing a syllabus of such a course/ program!
Looking for insight 💡
What tools are people using these days to make external facing dashboards that are low/no cost and where are you hosting them?
New post!
It may seem ambitious to ask for individual patient data from clinical trials to be shared, anonymized, for use by other researchers.
But the history of medicine shows us that clinical trials have already undergone a series of transformations that once seemed equally bold:
I like your overall layout. I was surprised you didn't suppress the running chain output, though.
My approach to analyzing data vs other approaches
One of the important lessons of my postdoc
Calling @taymalsalti.bsky.social
🎉 It's livestream day 🎉
Join me at 1700 (CET; 1600 UCT) today for two hours of GAM goodness 🤤
📽️ Youtube: youtube.com/live/A9U8e1K...
Hit the Notify me 🔔 to get a reminder when I go live
#RStats #mgcv #statistics #GAMs #DataScience 🧪
gotta have something to talk about
Everyone give Matt papers
alwayse
If I understand the term, which is tenuous given it's not used in my discipline, pragmatic randomized trials seem like a nice design
correct
I agree, but would add that that's applicable to a wide range of research and data-analytic procedures.
Pharma trials are sufficiently out of my domain that it wouldn't be wise for me to comment
Come for the stats, stay for the Platz
Even Dinesh from Silicon Valley pulled off the Fight Club physique
I mean, many dudes could get like 65% there with a good weight cut. Pitt wasn't exactly stacked
I'm not trying to be Tom Platz out here. I'm just looking for that Brad Pitt Fight Club physique.
Back to your lifting example. I'm not particularly interested in a 12-week protocol that requires perfect gym adherence, and 100% effort for every set and every rep. Most lifters don't go that hard, and I'd like the exercise-science research to speak to what what imperfect lifters might expect.
In my area, ITT has also been a way of talking about missing data, such that if you don't drop cases with missingness, you're doing ITT. It's not great that ITT has been so tightly linked with missing data methods, but so it goes. My forebears did what they did.
Non-adherence is central to therapy research, and it's among the primary reasons I like ITT. Removing non-adherent persons from a therapy RCT would result in a causal estimand that has less relevance to real-world therapy than otherwise. Real world therapy clients don't perfectly comply.
I have never seen ITT defined along those lines, and it's certainly not what what I mean by ITT.
I can't speak for other disciplines, but I find it hard to see why I'd want anything other than an ITT effect.
