######STEP 1: Run outcome analysis without propensity scores
sat=read.csv("sat2.csv")
model0=glm(Satisfaction ~ Nurse + MI + CHF + Diabetes +Injuries + Lung.Cancer +Age + Under.staffed + Pain , data = sat)
summary(model)

######STEP 2: Balance analysis prior to propensity score implementation
analysis=glm(Nurse ~ MI + CHF + Diabetes +Injuries + Lung.Cancer +Age + Under.staffed + Pain ,
data = sat)
summary(analysis)

######STEP 3: Estimate Propensity scores
propensity=glm(Nurse ~ MI + CHF + Diabetes +Injuries + Lung.Cancer +Age + Under.staffed + Pain ,data = sat, family=binomial)
summary(propensity) 
pscore= predict(propensity, sat , type="response") 

######STEP 4: Estimate weights using propensity scores
w = ( sat$Nurse / pscore) + (( 1- sat$Nurse ) / ( 1- pscore) )  
sat= cbind(w,sat) 
sat= cbind(sat,pscore)

######STEP 5: Balance analysis after implementing propensity scores
propensity2=glm(Nurse ~ MI + CHF + Diabetes +Injuries + Lung.Cancer +Age + Under.staffed + Pain, data = sat, family=binomial,weights=w)
summary(propensity2)

######STEP 6: Outcome analysis using propensity scores in weighted regression 
outcome= glm(Satisfaction ~ Nurse + MI + CHF + Diabetes +Injuries + Lung.Cancer +Age + Under.staffed + Pain , data = sat, weights = w) 
summary(outcome)