Drawing causal inferences from nonexperimental data is difficult due to the presence of confounders, variables that affect both the selection into treatment groups and the outcome. Post-hoc matching and stratification can be used to group individuals who are comparable with respect to important variables, but commonly used methods often fail to balance confounders between groups. We introduce model-based matching, a nonparametric method which groups observations that would be alike aside from the treatment. We use model-based matching to conduct stratified permutation tests of association between the treatment and outcome, controlling for other variables. Under standard assumptions from the causal inference literature, model-based matching can be used to estimate average treatment effects. We give examples of model-based matching to test the effect of packstock use on endangered toads and of salt consumption on mortality at the level of nations.
- Start date: 2016-03-15 11:00:00
- End date: 2016-03-15 12:30:00
- Venue: 639 Evans Hall at UC Berkeley
- Address: 639 Evans Hall, Berkeley, CA, 94720