Nikola Grubor, M.D.
LightNews LightNews
banner
ngrubor.bsky.social
Nikola Grubor, M.D.
@ngrubor.bsky.social
26 followers 140 following 16 posts
Rads Resident, UCCS | Ph.D. Student @ Belgrade University, Faculty of Medicine
Posts Replies Media Videos
Reposted by Nikola Grubor, M.D.
soashworth.bsky.social
Tapping the sign:
INTERPRETING A REGRESSION A common mistake is failing to notice when the best linear predictor cannot be credibly interpreted as the causal relationship A regression of height on weight answers questions like: Bob weighs 200 pounds. What's your best guess for his height? It is useless for questions like: If Bob went on a diet and lost 20 pounds, how much less would his height be?
October 28, 2025 at 3:13 AM
Reposted by Nikola Grubor, M.D.
rmcelreath.bsky.social
Was asked about collinearity again, so here's Vahove's 2019 post on why it isn't a problem that needs a solution. Design the model(s) to answer a formal question and free your mind janhove.github.io/posts/2019-0...
tl;dr Collinearity is a form of lack of information that is appropriately reflected in the output of your statistical model. When collinearity is associated with interpretational difficulties, these difficulties aren’t caused by the collinearity itself. Rather, they reveal that the model was poorly specified (in that it answers a question different to the one of interest), that the analyst overly focuses on significance rather than estimates and the uncertainty about them or that the analyst took a mental shortcut in interpreting the model that could’ve also led them astray in the absence of collinearity. If you do decide to “deal with” collinearity, make sure you can still answer the question of interest.
October 1, 2025 at 5:29 AM
Reposted by Nikola Grubor, M.D.
dingdingpeng.the100.ci
Just finished reading this *excellent* article by Gabriel et al. which discusses which effects can be identified in randomized controlled trials. With DAGs!>

link.springer.com/article/10.1...
Elucidating some common biases in randomized controlled trials using directed acyclic graphs Although the ideal randomized clinical trial is the gold standard for causal inference, real randomized trials often suffer from imperfections that may hamper causal effect estimation. Stating the estimand of interest can help reduce confusion about what is being estimated, but it is often difficult to determine what is and is not identifiable given a trial’s specific imperfections. We demonstrate how directed acyclic graphs can be used to elucidate the consequences of common imperfections, such as noncompliance, unblinding, and drop-out, for the identification of the intention-to-treat effect, the total treatment effect and the physiological treatment effect. We assert that the physiological treatment effect is not identifiable outside a trial with perfect compliance and no dropout, where blinding is perfectly maintained Table 1 showing the Identifiability of target estimands depending on whether there is blinding, full compliance, and no drop-out An example DAG from the paper. Fig. 4: A blinded trial with noncompliance. U are unobserved confounders, Z is treatment assignment, C is compliance, X is the realized treatment, S is the subject's physical and mental health status, Xself and Xcln are the treatment that the participant and the clinician believed the participant received, Y is the outcome.
October 2, 2025 at 8:09 AM
Reposted by Nikola Grubor, M.D.
mzefferman.bsky.social
Does anyone have an accessible guide for best practices in presenting DAGs and associated regression estimates in a paper?

This is for a student thesis project (and, frankly, for myself, as I have done it admittedly ad hoc, although no reviewers have so far complained.)
June 3, 2025 at 5:22 PM
Reposted by Nikola Grubor, M.D.
rmcelreath.bsky.social
This is a good joke about meta-science. I expect to see it in many talks in future years
June 13, 2025 at 8:11 AM
ngrubor.bsky.social
@pwgtennant.bsky.social Did I imagine it, or did you give a talk about using DAGs with deterministic variables? If yes, I would appreciate your input.
March 21, 2025 at 4:31 PM
Reposted by Nikola Grubor, M.D.
vickersbiostats.bsky.social
My first post on BlueSky has to be about decision curve analysis, right? We created a website to make it very easy for everyone. Code, tutorials, references can all be found at www.decisioncurveanalysis.org
DCA: Homepage
If you can’t find an answer to a question in one of the other tabs, click on the “Discussions” tab, which will navigate you to the GitHub discussions forum for all Q&A. If you post a question, someone from our team will get back to you.
Www.decisioncurveanalysis.org
November 14, 2024 at 4:23 PM
ngrubor.bsky.social
It's normal approximations - all the way down.
October 31, 2024 at 7:03 PM
Reposted by Nikola Grubor, M.D.
f2harrell.bsky.social
#statistics thought of the day - my fundamental statistical principles: www.fharrell.com/post/princip...
Fundamental Principles of Statistics – Statistical Thinking
This brief note catalogs what I feel are some of the most important principles to guide statistical practice.
www.fharrell.com
October 27, 2024 at 3:59 PM
Reposted by Nikola Grubor, M.D.
pwgtennant.bsky.social
👀
October 27, 2024 at 11:09 AM
ngrubor.bsky.social
Though physicians probably don't realize it, they're doing Baysian updating every time they construct a differential diagnosis. Others might just be blackboxing it through experience.
October 27, 2024 at 9:22 AM
Reposted by Nikola Grubor, M.D.
xangregg.bsky.social
Some Friday #dataviz fun: my remake of xkcd.com/2048/ with actual curve fitting (but without the witty commentary).
comic showing 12 curve types fitting the same data set. source: https://xkcd.com/2048/ remake of xkcd comic showing 12 curve types fitting the same data set.
October 18, 2024 at 3:16 PM
Reposted by Nikola Grubor, M.D.
jdwilko.bsky.social
Here’s how it works. A supervisor sets a student a project, but they don’t know how to do it, so they send them to a statistician for advice. The stato tells them something they can do, explaining the inevitable limitations. Student follows the instructions, and then the supervisor says it’s wrong.
October 6, 2024 at 9:42 AM
Reposted by Nikola Grubor, M.D.
dingdingpeng.the100.ci
I have updated the meme to celebrate the occasion.
October 5, 2024 at 12:10 PM
Reposted by Nikola Grubor, M.D.
avehtari.bsky.social
The most recent brms CRAN version added support for loo_epred() and moment matching for LOO-CV predictions, which makes it easy to make, for example, predictive probability calibration plots using the LOO-CV predictions

#Bayesian #rstats
rd<-reliabilitydiag(EMOS = loo_epred(fit), y = df$y) autoplot(rd) + labs(x = "Predicted (LOO)", y = "Conditional event probabilities")
October 1, 2024 at 12:46 PM
Reposted by Nikola Grubor, M.D.
avehtari.bsky.social
All three books I've co-authored are freely available online for non-commercial use:

- #Bayesian Data Analysis, 3rd ed (aka BDA3) at stat.columbia.edu/~gelman/book/

- #Regression and Other Stories at avehtari.github.io/ROS-Examples/

- Active Statistics at avehtari.github.io/ActiveStatis...
The cover of Bayesian Data Analysis book The cover of Regression and Other Stories book The cover of Active Statistics book
August 2, 2024 at 1:35 PM
Reposted by Nikola Grubor, M.D.
jofrhwld.bsky.social
i think i did a good job illustrating and applying pivot_longer() and pivot_wider()

lin611-2024.github.io/notes/meetin...
Pivoting Data – Lin611-2024
lin611-2024.github.io
September 18, 2024 at 6:50 PM
ngrubor.bsky.social
Late to the party, as always😅
September 16, 2024 at 10:41 PM
ngrubor.bsky.social
Adding "NA" color mapping for fills to the legend in ggplot should NOT be this hard. How unexpectedly absurd.
September 10, 2024 at 4:47 PM
Reposted by Nikola Grubor, M.D.
statsepi.bsky.social
How many tantalizing narrative reviews will be written about the gut microbiome and it's relationship to, apparently, every disease before we get another well-evidenced therapeutic success like C.diff?

Or is the goal of these reviews actually more about creating probiotic consumers?
September 10, 2024 at 1:35 PM
ngrubor.bsky.social
Got asked a clever question about metabolic rate and longevity. There is a lot of research about optimal BMI, but I just don't find it insightful. Are there any good causal paprs on the rate-of-living theory?
September 8, 2024 at 10:23 PM
Reposted by Nikola Grubor, M.D.
lmu-osc.bsky.social
⚙️ Learn about enhancing the professionalism of science through transferring aspects of software development. Register for the lecture “Science as Amateur Software Development” by @rmcelreath.bsky.social on 10.10 from 11:00-12:00 on s.gwdg.de/rr94Yi as part of the Open Science Summer School!
Promo image for the lecture “Science as Amateur Software Development” by Richard McElreath on September 10th from 11:00-12:00 as part of the LMU and MPG Open Science Summer School.
September 4, 2024 at 6:05 PM
ngrubor.bsky.social
I'm thinking about just making a Hugo awards winners reading list. 📚 might be worth a try.
September 4, 2024 at 10:49 PM