Difference_in_difference model(DID) with matched pair

[fama]

Hi All,

I am running the DID regression model to estimate the treatment effect.

Prior to this, I have created the dataset using propensity score matching. So, I have matched pairs (each treated subject is matched to two nearest neighbours control subject).

I am using below code. However, this code does not account for the matched data.

Indeed, the matched pair should be the unit of analysis and I need to modify the below code to accout for the  matched pairs

To make it clear, the below code considers all treated subjects versus all control subjects while  I am looking for the differences between treated and control subject in each matched pair.

Thanks in advanced

/*Repeated measure linear regression*/

PROC MIXED DATA = Long; /* long format*/

CLASS A B;

MODEL Y=A|B covariatesX1 X2..Xn / SOLUTION;
LSMEANS      A|B / DIFF;
ESTIMATE     'D-I-D' A*B    1 -1 -1 1;
RANDOM     Int/SUBJECT=     TYPE=UN ;
RUN;

[fama]

Hi All,

I am running the DID regression model to estimate the treatment effect.

Prior to this, I have created the dataset using propensity score matching. So, I have matched pairs (each treated subject is matched to two nearest neighbors control subject).

I am using below code. However, this code does not account for the matched data.

Indeed, the matched pair should be the unit of analysis and I need to modify the below code to account for the  matched pairs

To make it clear, the below code considers all treated subjects versus all control subjects while  I am looking for the differences between treated and control subject in each matched pair.

Thanks in advanced

/*Repeated measure linear regression*/

PROC MIXED DATA = Long; /* long format*/

CLASS A B;

MODEL Y=A|B covariatesX1 X2..Xn / SOLUTION;
LSMEANS      A|B / DIFF;
ESTIMATE     'D-I-D' A*B    1 -1 -1 1;
RANDOM     Int/SUBJECT=     TYPE=UN ;
RUN;

[fama]

Hi All,

I am running the DID regression model to estimate the treatment effect.

Prior to this, I have created the dataset using propensity score matching. So, I have matched pairs (each treated subject is matched to two nearest neighbors control subject).

I am using below code. However, this code does not account for the matched data.

Indeed, the matched pair should be the unit of analysis and I need to modify the below code to account for the  matched pairs

To make it clear, the below code considers all treated subjects versus all control subjects while  I am looking for the differences between treated and control subject in each matched pair.

Thanks in advanced

/*Repeated measure linear regression*/

PROC MIXED DATA = Long; /* long format*/

CLASS A B;

MODEL Y=A|B covariatesX1 X2..Xn / SOLUTION;
LSMEANS      A|B / DIFF;
ESTIMATE     'D-I-D' A*B    1 -1 -1 1;
RANDOM     Int/SUBJECT=     TYPE=UN ;
RUN;