I am using "synth" package to estimate the impact of donor funding on under-five mortality rates in a set of countries (let's call this Group 1 which contains 8 sub-Saharan African countries). The intervention year is 1999 and pre-intervention period is 1980-1998 and post-intervention period is 2000-2016. I use a number of predictors in the model, and below is the code I run after I set up the data:
synth group1u5mr group1u5mr group1u5mr_lagged group1odapc group1polityivr group1TFR group1loggdp group1urb group1stunting group1HIV group1sanit group1cleanwater, trunit(686) counit( 12 32 72 76 120 140 144 148 170 174 178 188 192 218 226 232 262 266 270 328 364 384 388 418 422 426 458 478 484 496 498 504 516 562 591 600 624 694 704 710 740 748 764 768 788 807 854 862) trperiod(1999) nested fig xperiod(1980(1)1998) allopt keep(group1.dta, replace)
The problem is that when I change the order of the predictor variables, I get different results; results that are wildly different in terms of predictor balance, donor weights and RMPSE. So, I wanted to check if anyone else has run into this issue and if there are steps I can take to correct this error.
FYI, the numbers following counit are id numbers of control countries.
Thanks in advance!
synth group1u5mr group1u5mr group1u5mr_lagged group1odapc group1polityivr group1TFR group1loggdp group1urb group1stunting group1HIV group1sanit group1cleanwater, trunit(686) counit( 12 32 72 76 120 140 144 148 170 174 178 188 192 218 226 232 262 266 270 328 364 384 388 418 422 426 458 478 484 496 498 504 516 562 591 600 624 694 704 710 740 748 764 768 788 807 854 862) trperiod(1999) nested fig xperiod(1980(1)1998) allopt keep(group1.dta, replace)
The problem is that when I change the order of the predictor variables, I get different results; results that are wildly different in terms of predictor balance, donor weights and RMPSE. So, I wanted to check if anyone else has run into this issue and if there are steps I can take to correct this error.
FYI, the numbers following counit are id numbers of control countries.
Thanks in advance!
