My understanding is that for a binary outcome the ATET is equivalent to the absolute risk difference between i. those treated and ii. those matched to the treated using PSM
So to my thinking the ATET for my primary outcome, depression, is .0604, so the absolute risk difference is 6.0%. The proportion of those treated (received a new form of welfare) who were depressed was 51.7%, therefore the proportion of those untreated ought to be 51.7-6.0 = 45.7%. From this you can calculate the relative risk at 51.7/45.7% = 1.13.
My question are
1. Is this correct? I haven't found anything relating to calculating the proportions of those with the outcome in the untreated or risk ratios.
2. if it is correct how do I calculate confidence intervals?
code
Code:
teffects psmatch (mcs_bin) (switch sex_dv1 i.agecat1 i.hhdi_5_1 health1 i.jbstat2 i.marstat_dv1 smoker1 i.hiqual2 i.region1 i.ethn2 ca1 i.intdaty_dv1), atet nn(3) proportion mcs_bin if switch==1
0 Response to Relative risk ratios and their CIs from teffects psmatch
Post a Comment