efun <- model.matrix(m1) * resid(m1,"response")
infl <- efun %*% m1$naive.cov/mean(weights(design_1))
design_2 <- svydesign(id=~cntry, weights=~dweight, data=ESS)
svytotal(infl, design_2)