Appendix

[cdgrady21]

[ ] add appendix files

[cdgrady21]

• Explaining randomization inference and bootstrapping
• Survey questions
• Robustness checks for analysis
  • Additive indices instead of ICW indices
  • Results with differencing method vs results with controlling-for method
  • raw vs ranked variables to create the index
• Placebo analysis + components of placebo
• Balance (bal_tests.rmd in appendix folder):
  • Communities/site: TR vs CO within state & overall for all baseline outcomes (Separate for svy and obs)
  • Individuals: Direct/indirect/control == unneeded because exploratory & no assumption of balance.
• Panel analysis: big table of results (lower priority)
  • Additive indices instead of ICW indices
  • Results with differencing method vs results with controlling-for method
  • raw vs ranked variables to create the index
• State-level differences & farmer-pastoralist differences (ready for reviewers)
• Extra placebos: outcomes we listed in initial PAP as things we measure but do not expect ECPN to change: "changing perceptions of the acceptability of some types of violence, about inter-religious trust, or perceptions of the actors involved in dispute resolution". Must do for individual-level analysis
• From PAP, about individual-level analysis: "To lend credibility to the claim that participants and nonparticipants in the ECPN communities are not “different types” than control group respondents (i.e. that respondents in control communities function as a counterfactual for respondents in treatment communities), we will (1) compare demographic balance of the participants, non-participants, and controls, and (2) provide evidence for parallel trends by conducting a placebo test on outcomes we do not expect ECPN to change." Did not do 2, and I don't think 1 is in the appendix. Also for 1 I should check baseline outcome balance (we decided this is not so important because ind-level analysis isn't important, so this is low priority).

Still need to do bolded items.