i am working on comparative study that discuss the effect of applying a directive, i plan to use DID statistical method that operates with 1 and zero. but my supervisor asks to not use dummies and convert them to ratios (that represent the extent to which this directive is applied).
is it right to use DID statistical method to test ratios as a treatment group and control group knowing that 99% of control group contains zero? if not what is the appropriate statistical method?
i am thinking of using panel regression to test ratios variable for just the treatment group and then using DID for treatment and control groups and use dummies.
You can use DID with continuous treatment if and only if 1) there is variation in the ratios both before and after treatment, and 2) the ratio is exogenous (that is, the process generating that ratio must be exogenous to the outcome you're studying). But I'd just do binary dummy or even better as you say, panel regression with a fixed effect for treatment
Standard DiD won't work well with ratios. Your treatment is now a ratio (not just 0/1) and most controls are zero, use a panel regression with fixed effects to test the ratio's impact on the treatment group, then do a separate DiD analysis comparing treated vs. untreated groups if you still need a binary comparison.
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators