Week 6 - PS Methods with Multiple Treatments Flashcards
What is weak unconfoundness?
The requirement that assignment to each treatment is independent of the potential outcomes of the respective treatment.
What is the difference between weak unconfoundness and strong ignorability of treatment assignment?
Weak unconfoundness is equivalent to weak ignorability of treatment assignment. The difference between strong ignorability and weak ignorability is that the later does not require that the assignment to one treatment is independent of all potential outcomes.
What are the assumptions necessary for propensity score methods for multiple treatments?
- weak ignorability of treatment assignment
- overlap
- stable unit treatment value assumption (SUTVA).
What is the generalized propensity score?
GPS is the conditional probability of receiving a particular level of treatment given covariates.
What are two methods that can be used to estimate generalized propensity scores?
- multinomial logistic regression model
* data mining
How can data mining methods be used to estimate generalized propensity scores?
You create a dummy indicator for each treatment version, and predict each dummy indicator.
The GPS for each individual is the predicted probability of the treatment version that the individual was exposed to.
How can common support be assessed with multiple treatments?
- Visual inspection of distributions is useful as a preliminary check.
- Examination of the minimum and maximum of the distribution of GPS for each group.
- Common support should hold for all pairs of treatment versions with respect to each of the J vectors of GPS.
- Common support is problematic if trimming the distribution of weights at the 99% percentile results in substantial improvement in covariate balance.
How can weights based on the generalized propensity score be calculated?
Taking the inverse of the GPS will give the weight.
How are marginal mean weights through stratification calculated with multiple treatments?
- Define treatment levels.
- Obtain generalized propensity score
- Check for overlap.
- Divide propensity scores into strata.
- Obtain marginal mean weights through stratification.
- Estimate the treatment effects across strata using the weights.
How can covariate balance be assessed with multiple treatments?
The covariate balance can be evaluated for each treatment version by comparison with all other versions, or by pairwise comparison of treatment versions.
What are two methods to estimate the ATE with multiple treatments?
- weighted mean differences
* generalized linear models (regression model for continuous outcomes or logistic regression model for binary outcomes)
How can control for covariates be performed in the estimation of effects of multiple treatments?
By…
- Regression Model Continuous Outcomes
- Logistic Regression Model Binary Outcomes
