The DHS Program User Forum: Child Health

Home » Topics » Child Health » SII, RII, CI, and Gini (Calculation of relative index of inequality, slope index of inequality, Concentration Curve and Index, Lorenz Curve and Gini Coefficient)

Show: Today's Messages :: Show Polls :: Message Navigator

SII, RII, CI, and Gini [message #26072]

Sat, 04 February 2023 03:19

Kazi_Salahin
Messages: 9
Registered: February 2023

Member

Dear Concern

I am using the Bangladesh data (BDHS-2017-2018) specifically the KR data file. I want to calculate the "Relative index of inequality." "Slope index of inequality," "Concentration index and concentration curve," and "Gini coefficient and Lorenze curve" in consideration of the wealth index (v190), maternal education (v149), and full vaccination coverage (ch_allvac_either). I'd like to compute each indicator by division (v101). I am using R. I want to make certain that I am on the correct truck. The following is my R code:

############################################################ ############################################################ ###############################
#Rcode
## Calculate ridit score
KRvac$ridit_score_wealth <- rank(KRvac$v190, ties.method = "average") / nrow(KRvac)

# Loop to create a data frame with division and its corresponding SII, RII, and CI
divisions <- unique(KRvac$v101)
sii_wealth_vec <- numeric(length(divisions))
rii_wealth_vec <- numeric(length(divisions))
ci_wealth_vec <- numeric(length(divisions))
gini_vec <- numeric(length(divisions))

for (i in 1:length(divisions)) {
KRvac_subset <- KRvac %>% filter(v101 == divisions[i])

model_wealth <- glm(ch_allvac_either ~ ridit_score_wealth, family = binomial(link = "logit"), data = KRvac_subset)

pred_wealth_top <- predict(model_wealth, newdata = data.frame(ridit_score_wealth = max(KRvac_subset$ridit_score_wealth)), se.fit = TRUE)
pred_wealth_bottom <- predict(model_wealth, newdata = data.frame(ridit_score_wealth = min(KRvac_subset$ridit_score_wealth)), se.fit = TRUE)

pred_wealth_top<-as.numeric(pred_wealth_top)
pred_wealth_bottom<-as.numeric(pred_wealth_bottom)

sii_wealth_vec[i] <- pred_wealth_top - pred_wealth_bottom
rii_wealth_vec[i] <- pred_wealth_top / pred_wealth_bottom

ci_wealth_vec[i] <- sii_wealth_vec[i]/(2*sqrt(rii_wealth_vec[i]))

n <- nrow(KRvac_subset)
y_values <- sort(KRvac_subset$ch_allvac_either)
x_values <- sort(KRvac_subset$ridit_score_wealth)
numerator <- sum((2 * (1:n) - n - 1) * y_values * x_values)
denominator <- n * sum(y_values) * sum(x_values)
gini_vec[i] <- numerator / denominator
}

results_df <- data.frame(division = divisions, sii = sii_wealth_vec, rii = rii_wealth_vec, ci = ci_wealth_vec, gini = gini_vec)

############################################################ ############################################################ ###############################

Here the plot code (Concentration Curve and Lorenze Curve) is not included. It would be greatly appreciated if anyone could assist me in ensuring that I am doing the right thing; if not, please correct my code; and I also require assistance with the plot code. I am a little bit confused about that.

Your early response will be appreciated

Thank you

Report message to a moderator

Re: SII, RII, CI, and Gini [message #26102 is a reply to message #26072]

Wed, 08 February 2023 09:44

Janet-DHS
Messages: 893
Registered: April 2022

Senior Member

Following is a response from DHS staff member, Tom Pullum:

We (DHS staff) cannot help with questions that are not specifically about DHS data and indicators. Perhaps other forum users can review your R code.

Report message to a moderator