hello ,
i m new to Stata , i use the Stata 12 IC version , in my project i started with a multiple imputation to impute missing data , followed by a propensity score on each imputed data set by the (IPTW approach), followed by a survival analysis to have the ATE
in order to assess the balance of my score after creating the weights i m using the pbalchk command , the graph i get doesnt allow me to have the absolute std differences option, i wonder if there is a way to asses the balance other than this command and that allows me too to get the absolute std differences ,
i also want to know how to get the number of observations per covariable , per group of treatment after implementing the weights (in the pseudopopulation) so that i can make a descriptive table of my population ( before weighting and after)
finally i have a probleme with my survival analysis , when i declare my data as mi stset data , and implement the weights i ve created , it excludes 533 weights , and when i verified those weights it was the patients in m0 that had missing values (as the imputed variables where included in the score) , my question is why does it include the m0 in the pooling step and thus remove the weights and how to solve that
mi xeq : logit TTaccesgraves1 i.pretraitement i.annee_rec i.ethnie_rec i.gender i.cl_age i.pays_endemie1_recod2 i.chimioprophylaxie_dec i.immunodep_connue_dec i.critereg_clin1_trouble i.critereg_clin_2_coma i.gb_rec i.parasitemia_dec i.plaquettes_rec i.delai_rec i.activity_rec
mi xeq: predict preprob, pr
list TTaccesgraves1 preprob in 1/5
mi xeq: gen psweight =.
mi xeq : replace psweight = (1/preprob) if TTaccesgraves1==1
mi xeq : replace psweight = (1/(1-preprob)) if TTaccesgraves1==0
mi xeq : xi: pbalchk TTaccesgraves1 i.pretraitement i.annee_rec i.gender i.cl_age i.chimioprophylaxie_dec i.immunodep_connue_dec i.critereg_clin1_trouble i.critereg_clin_2_coma i.gb_rec i.parasitemia_dec i.plaquettes_rec i.delai_rec i.activity_rec , wt(psweight) f graph and this command gives me the std diff with + and -ve values
mi stset time [pweight=psweight], failure (outcome_var) id(palu_id_dec)
and here i get that 533 weights are not valid
mi estimate, eform: stcox i.TTaccesgraves1
thanks a lot in advance
i m new to Stata , i use the Stata 12 IC version , in my project i started with a multiple imputation to impute missing data , followed by a propensity score on each imputed data set by the (IPTW approach), followed by a survival analysis to have the ATE
in order to assess the balance of my score after creating the weights i m using the pbalchk command , the graph i get doesnt allow me to have the absolute std differences option, i wonder if there is a way to asses the balance other than this command and that allows me too to get the absolute std differences ,
i also want to know how to get the number of observations per covariable , per group of treatment after implementing the weights (in the pseudopopulation) so that i can make a descriptive table of my population ( before weighting and after)
finally i have a probleme with my survival analysis , when i declare my data as mi stset data , and implement the weights i ve created , it excludes 533 weights , and when i verified those weights it was the patients in m0 that had missing values (as the imputed variables where included in the score) , my question is why does it include the m0 in the pooling step and thus remove the weights and how to solve that
mi xeq : logit TTaccesgraves1 i.pretraitement i.annee_rec i.ethnie_rec i.gender i.cl_age i.pays_endemie1_recod2 i.chimioprophylaxie_dec i.immunodep_connue_dec i.critereg_clin1_trouble i.critereg_clin_2_coma i.gb_rec i.parasitemia_dec i.plaquettes_rec i.delai_rec i.activity_rec
mi xeq: predict preprob, pr
list TTaccesgraves1 preprob in 1/5
mi xeq: gen psweight =.
mi xeq : replace psweight = (1/preprob) if TTaccesgraves1==1
mi xeq : replace psweight = (1/(1-preprob)) if TTaccesgraves1==0
mi xeq : xi: pbalchk TTaccesgraves1 i.pretraitement i.annee_rec i.gender i.cl_age i.chimioprophylaxie_dec i.immunodep_connue_dec i.critereg_clin1_trouble i.critereg_clin_2_coma i.gb_rec i.parasitemia_dec i.plaquettes_rec i.delai_rec i.activity_rec , wt(psweight) f graph and this command gives me the std diff with + and -ve values
mi stset time [pweight=psweight], failure (outcome_var) id(palu_id_dec)
and here i get that 533 weights are not valid
mi estimate, eform: stcox i.TTaccesgraves1
thanks a lot in advance
