Causal-Inference

Vignette of an R package for instrumental variable regression.

We propose a Bayesian model averaging method to account for the uncertainty about instrument validity in Mendelian randomization. This model is extended to allow for a large fraction of SNPs violating the InSIDE assumption.

Sparsity is often used to improve the interpretability of a statistical analysis and/or reduce the variance of a statistical estimator. This talk will explore another aspect—the utility of sparsity in model identifiability through two problems …

Many partial identification problems can be characterized by the optimal value of a function over a set where both the function and set need to be estimated by empirical data. Despite some progress for convex problems, statistical inference in this …

We apply the MR-RAPS method we developed in previous articles to infer the potential causal role of lipoprotein subfractions in CAD. This is motivated by the finding in our earlier IJE paper that the association between genetically-determined HDL-C and CAD is heterogeneous according to instrument strength. In this study, We find that HDL subfraction traits, unlike LDL and VLDL subfractions, appear to have heterogeneous effects on coronary artery disease according to particle size. The concentration of medium HDL particles may have a protective effect on CAD that is independent of traditional lipid factors.

Optimal individualized treatment rules try to assign the best treatment to every individual, but it may be very sensitive to unmeasured confounding bias for groups of people exhibiting small treatment effect in the observational study. We give a …

This is an invited commentary for Statistical Science on the causal inference data competition in ACIC 2016.

In a special workshop in ACIC 2018, we were invited to analyze a simulated dataset to detect treatment effect heterogeneity. This article reports our results presented in the workshop. We also tried out more recent selective inference methods based …

We propose a falsification test for the IV assumptions using sub-populations of the data with overwhelming proportion of treated or untreated units. If the IV assumptions hold, we should find the intention-to-treat effect is zero within these …

This research letter proposes a new diagnostic plot for IV analysis, so large bias ratios (compared to OLS estimator) are not over-interpreted when the covariate is unrelated to the outcome.