Posts

Flipping the sign of hazard ratios

r
survival analysis
causal inference
Hazard ratios are problematic summary measures for survival analysis (Dumas and Stensrud 2025; Post, Van Den Heuvel, and Putter 2024; Stensrud et al. 2019; Aalen, Cook, and…
Jul 4, 2025
Wouter van Amsterdam

What is stronger evidence of prediction model robustness?

statistics
prediction
causal inference
We’re evaluating a prediction model \(f\) that takes features \(X\) to predict binary outcome \(Y\). We’d like the model to be robust, meaning that it’s predictive…
Apr 25, 2025
Wouter van Amsterdam

Is the JAMA opening up their language for causal effects?

causal inference
Randomized controlled trials (RCTs) measure the causal effect of interventions, but results from observational studies should be interpreted as mere associations, right? In…
Jun 7, 2024
Wouter van Amsterdam

Partial residual plots with multiply imputed data

r
linear regression
data visualization
In an earlier blog post I show how to plot the dependence of a response variable on covariate, both conditional on other covariates with a partial residual plot In this post…
Mar 22, 2024
Wouter van Amsterdam

The need for speed, performing simulation studies in R, JAX and Julia

r
julia
jax
python
simulation studies
Simulation experiments are important when evaluating methods but also for applied work in for example power analyses (e.g. Amsterdam, Harlianto, et al. 2022) or sensitivity…
Mar 8, 2024
Wouter van Amsterdam
 

The difference between intervention and counterfactuals

causal inference
statistics
Sometimes there is confusion about the difference between counterfactual predictions and interventional predictions. According to the ladder of causation introduced by Pearl…
Jul 25, 2021
Wouter van Amsterdam
 

When good predictions lead to bad decisions

causal inference
predictions
A common premise in prediction research for clinical outcomes is that better predictions lead to better (informed) decisions. This causal statement, that the intervention of…
Jul 20, 2021
Wouter van Amsterdam

Finding the functional form for multiple linear regression

statistics
simulations
linear regression
A frequent question that comes up when modeling continuous outcomes with multiple linear regression is what the correct functional form for the relationship between the…
Aug 16, 2019
Wouter van Amsterdam
No matching items