Dear all,
I have a data set, where 80% of total respondents are coming under treated group and rest of them are in control group. In this situation, using PSM technique can be appropriate for estimating unbiased Average Treatment Effect (ATE)? and please suggest any other methods and research work regarding this to solve this problem.
Many thanks in advance
Usually the problem is the other way. The treatment is expensive (or disease rare) so you match several "control" with each treatment. You can just do the opposite, match multiple treatments to each control. See Rosenbaum's textbooks.
Yes, it is appropriate. I see two ways to handle it:
(1) As Daniel suggested, you can determine a ratio for the sample sizes of both groups (how many cases of the treatment group should be matched to one case of the control group, e.g., using nearest neighbor matching).
(2) You can also use full matching calculating a weight for the cases in one of the groups. Then you analyze all of your cases, but cases of one group are weighted. Most functions for analysis in R and also in SPSS offer the option to include weights.
I would go for the second alternative. My experience is that full matching results in less differences between the two groups.
Thanks a lot to Daniel Wright & Brian Schwartz.
Brain, I will try both points and let you know any queries if there is.
In PS matching with a ratio of 1:2 (Treatment:Control) all my treatment cases got two matched controls. None of the treated cases were thrown out. What could be the explanation?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size