You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
JobMarketPaper/logs.org

3.3 KiB

2025

2025-W03

2025-01-18 Saturday

[2025-01-18 Sat 11:54] [[/home/will/research/phd_deliverables/JobMarketPaper/Paper/sections/11_intro_and_lit.tex::45]]

Need to decide whether or not to include this set of sentences.

[2025-01-18 Sat 11:58] [[/home/will/research/phd_deliverables/JobMarketPaper/Paper/sections/11_intro_and_lit.tex::45]]

decide whether to include these details here

2025-W17

2025-04-21 Monday

[2025-04-21 Mon 11:17] Plan based on last weeks thinking things through

get list of things that Tom says I'm Missing

  • Needs more citations
  • Standard econometric concerns: Endogenetiy, Simultineatiy, etc.
  • Needs to justify why I am doing what I am doing. What do I add? Marketwide attempt to measure the impact of enrollment, an operational concern.

Integrate additional literature I've worked with.

  • How big of a concern is operational results (about 22% of failures)
  • Topics of how to address issues and what issues arise are common (give a couple of examples)
  • Efforts to reduce failures include better pharmokinetics, attempts at improving enrollment, better enrollment prediction (huge lit).

Then look at my outline:

  • How can I adjust it to address those missing bits?
  • How can I simplify the structure?

Maybe a discussion of concerns about simultineity/endogeneity/other confounds/etc is where I bring up the confounding parameters and then build a list of how things interact. I then use this to flesh out the DAG, and introduce the backdoor criterion.

I think I'll put this together as a bullet point draft, using the * and - notation for paragraphs and sentences respectively. Try to get the main points of each sentence/paragraph out.

List of issues identified by Tom:

Reference style (Author year) Reference better and more often. Introduction needs to motivate the problem & what I am trying to do. (could use the sources I have on reasons for failures) Various issues with tense etc. Use Claude.ai as editor for those. Reorder sections or outline better

Causal inference vs DAG approach

  • standard concens in causal inference
  • DAG isn't causal inference in Toms view. He is right, DAG isn't but backdoor criterion is.
  • Will need to discuss standard concerns and how they may be related and then incorporate that into the DAG
  • Then will need to discuss backdoor criterion, the backdoor paths that exist, and choosing adjustment sets
  • Replace bullet points with paragraphs (page 12) maybe use claude to convert that?
  • Page 18 comment: Refer to Robins What IF book to get citation

Thoughts: Chapter 10: Lists 3 sources of bias in preceeding chapters (7,8,9)

  • Selection
  • Measurement
  • Confounders

As I understand it, setting up the graph allows you to note where you might have issues with all 3. Do-calc gives you the adjustment set to handle confounding and selection, while measurement is handled either through modelling uncertanty or improving you measurement approach.

Reading to complete before rewriting:

I think I should start by rereading (and taking notes on) What If and the Causal Mixtape.