Posts

Unit-average lift or cluster-average lift in cluster-randomized experiments

network experiment

In a cluster-randomized experiment, we can choose between computing a unit average or cluster average. The two averages certainly bare different units, but which one should we use if we are reporting on relative terms?

Post-selection unbiased estimator

selective inference

Does a post-selection unbiased estimator for the mean of a normal distribution exist?

Parametric mean estimation

applied statistics

data science

A common task in data science is mean estimation. The go-to estimator is sample mean, which is non-parametric in nature. But what if we have good reasons to believe in a parametric model for the population?

Philosophy of statistics, or data science

data science

This is an ambitious title for a post, but perhaps, and I hope, every data scientist or statistician eventually has their own philosophy on what we are doing.

Meta-analysis on nudging experiments

multiple testing

A recent meta-analysis on choice architecture, or “nudging”, but people do not seem to agree on if “nudging” works.

Moment conditions heuristics

applied statistics

Moment conditions are common in literature, but how do we assess if they reasonably apply?

Playing with Bayes and RStan

statistical computing

First time pretending to be a Bayesian, and trying RStan.

Normal distribution in Python

statistical computing

Getting the hang of statistical computing in Python

Binomial ranking with SARS data

multiple testing

If deaths in each country follows a binomial distribution, how do we rank them by the probability parameter?

p-value screening

multiple testing

selective inference

Better multiple testing by screening p-values first, to reduce the penalty in common multiple testing procedure

L1-penalized likelihood asymptotics

selective inference

Generalizing Lasso penalty to non-linear model, e.g. a binomial model.