causal-inference-in-R
causal-inference-in-R copied to clipboard
malcolmbarrett
What to control for
Here is a nice paper on conditioning on instruments: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3254160/ top line result:
The results indicate that effect estimates which are conditional on a perfect IV or near-IV may have larger bias and variance than the unconditional estimate. However, in most scenarios considered, the increases in error due to conditioning were small compared with the total estimation error. In these cases, minimizing unmeasured confounding should be the priority when selecting variables for adjustment, even at the risk of conditioning on IVs.
Also this: https://dash.harvard.edu/bitstream/handle/1/25207409/90937280.pdf?sequence=2&isAllowed=y
Ding, Peng, and Luke W. Miratrix. 2015. “To Adjust or Not to Adjust? Sensitivity Analysis of M-Bias and Butterfly-Bias.” Journal of Causal Inference 3 (1) (January 1). doi:10.1515/jci-2013-0021.
Wooldridge (2009) and Pearl (2010) have shown that when bias due to unmeasured confounding is present, control for an instrument can amplify the existing confounding bias.
Updated the title on this. We now have a short section addressing this topic but should make sure we've covered everything