mkiang/intro-ci-shortcourse

ITAM Workshop on Causal Inference

Recent(ish) projects

• Quantifying exposure to wildfire smoke among California school kids in JAMA Network Open
• Excess physician mortality in JAMA Internal Medicine
• Excess external causes of death during COVID-19 in JAMA Internal Medicine
• Excess drug poisonings in California during COVID-19 in The Lancet Regional Health Americas
• The dynamics of racial/ethnic inequities during COVID-19 in PNAS
• Black/white opioid inequities in Journal of Urban Health
• Urban/rural disparities in distance to opioid treatment in Drug and Alcohol Dependence
• COVID-19 and airline testing strategies in The Lancet Infectious Diseases
• Opioid prescribing patterns in BMJ
• Opioid mortality across states in JAMA Network Open
• Medical school statements after George Floyd's murder in Annals of Epidemiology

Random side projects

• Counting the number of times they say "family" in the Fast and Furious movies
• Using Oura data to look at how COVID-19 impacted me
• Scraping data on SF public schools
• Increasing profanity in Taylor Swift's discography
• Simulating Candyland rules for toddlers
• Shiny app for SEIR / SIR infectious disease models

Projects I participated in (but are not my code repositories)

• Estimating youth experiencing parental deaths due to drugs or firearms by Ben Schlüter in JAMA
• Estimating lead exposure among children in Chicago by Ben Huynh in JAMA Pediatrics

Bio

• My CV
• My Google Scholar
• My lab website
• The Department of Epidemiology and Population Health

intro-ci-shortcourse's Issues

We still need an example of good CI for lecture 1

Zhe improve his personal page

Moving it here so we can clear out the README and keep all TODO/issues in a single place

Sort out who has what for first drafts

@writezhe: Can you confirm what you want to take the first draft on? Obviously, we can continue to edit and modify after, but it'll be useful if we assign the first round just so we can get a skeleton down. Then we'll flesh everything out as a group.

• 1. Introduction to CI (Matt)
• 2. Evaluating causality in this framework (Matt)
• 3. Biases: Confounding / Selection (Zhe)
• 4. Instrumental variables (Zhe)
• 5. Natural experiments (Zhe)
• 6. Regression discontinuity (TBD, pending other sections)
• 7. Matching designs (Matt)
• 8. Study design (Matt)

Table CSS is SUPER UGLY

Fix.

Article for IV section

@writezhe assume you've already got enough of them, but this is the reference article I always use when I need to brush up on IV.

http://economics.mit.edu/files/18

Right now, calling two versions of `custom.css`

In order to keep the moon_reader previews workable, we are calling two versions of custom.css. (See YAML on slide Rmd file.)

Once we have all the slides done, we will want to remove the custom.css files and references that are local to the directory and only call ../custom.css in the YAML so that CSS is consistent and easily modified for all slides.

Convert slides to PDFs once we have finalized

I'll do this since I already have decktape installed.

https://github.com/yihui/xaringan/wiki/Export-Slides-to-PDF

Download remark.js and use a single shared lib

To do when we are finished with all slides.

Are we happy with this title for the class?

Add in discussion of bad controls

https://en.wikipedia.org/wiki/Berkson%27s_paradox#Examples

Also see "bad controls" section of MHE

Not enough memes and images.

Lecture 1 Changes

• Still need example of good CI
• Missing some sources
• > One suggestion I have is to do a review at the transition between our motivation and the causal framework. Something short to summarize and reiterate our points. Also, I'll help add 1-2 more examples as I review literature in the next week. - @writezhe
• > Examples of counterfactuals in various settings (some easy, some hard)
  twins, change to Amazon business model, consumer demand shock, policy change, election politics, medicine, change to the transportation system. - @writezhe
• h tags might need to be more consistent — unclear.
• Add h titles to beginning slides
• Not all ? in example table data are .red[]
• Additional reading probably needs additional readings
• Not all slides have presenter notes — Add them before I forget what I was doing with them

CSS is kinda ugly. Fix if we have time.