Hierarchical Bayesian analysis investigating the causal effect of R&D expenditure on GDP growth
across 7 European countries (2010-2024, N=96 observations). Extended comprehensive analysis uses
Student-t likelihoods for robustness, varying slopes for country-specific effects, and explicit
crisis modeling (2009 financial crisis, 2020 COVID-19). Results show substantial heterogeneity:
Germany shows positive effect (β = +0.15), while aggregate effects remain uncertain. COVID-19
crisis effect (-1.6 SD) dwarfs all R&D effects by 10x.
Original study (2016-2022, N=49): β = -0.023 [-0.297, 0.284]
Additional Bayesian causal inference studies to be added here.
References & Infrastructure
Methodology Framework
"The simplest graphical causal model is a directed acyclic graph, usually called a DAG."
— McElreath (2020), Statistical Rethinking, Chapter 6
mcelreath2020statistical:para-1388
"Hierarchical modeling automatically partially pools estimates of different θj's towards their common mean."
— Gelman et al. (2013), Bayesian Data Analysis, Chapter 5
gelman2013bayesian:para-7725