Hello,
In a repeated cross-sectional designed study I would like to calculate the relative and absolute indexes of inequalities for education and test their associations with the prevalence of hypertension across survey periods. To do so I am using riigen to obtaing relative the index of inequality and then poisson and linear regressions for calculating relative (RII) and absolute (SII) inequalities followed by margins.
I am wondering if it is the right approach to include the interaction term between the index of inequality and survey period to then plot the results in a graph and showing the trend.
Giovanni
In a repeated cross-sectional designed study I would like to calculate the relative and absolute indexes of inequalities for education and test their associations with the prevalence of hypertension across survey periods. To do so I am using riigen to obtaing relative the index of inequality and then poisson and linear regressions for calculating relative (RII) and absolute (SII) inequalities followed by margins.
Code:
riigen edu, riiname("hyp") varprefix("RII_hyp")
glm hypertension c.age i.gender c.RII_hyp_edu##i.survey, fam(poisson) link(log) eform vce(robust)
margins, at(survey=(1(1)3)) expression(exp(predict())) saving(RII, replace) vce(uncond)
regress hypertension c.age i.gender c.RII_hyp_edu##i.survey, robust
margins, at(survey=(1(1)3)) saving(SII, replace) vce(uncond)
combomarginsplot RII SII, labels("RII" "SII")
Giovanni
Comment