In arnoblalam/risk_sharing: Risk Sharing Networks in Tanzania

name : data_check.R

description : Check the data

maintainer : Arnob L. Alam [e-mail]

updated : 2016-05-26

Check the raw data (exported from Stata) and the data provided

by the author on the paper's website to see if we can recreate summary

statistics provided in the other paper

nyakatoke_raw_data.csv is the export of the STATA dta file

provided to Heath by the author. We read it into the data frame

called raw_data

raw_data <- read.csv("data/nyakatoke_raw_data.csv")

tanzania_data.csv is the data provided along with the paper

on the Wiley (publisher's) website. We read it into the

data frame paper_data

paper_data <- read.csv("data/tanzania_data.csv")

According to th STATA variable labels direct_link_1 and direct_link_2

specify whether hh1 is linked with hh2. These columns are referred to as

willingness_link1 and willingness_link2 in the paper_data. Below I show that

these two colums are the same

Check if direct_link_hh1 is equal to willingness_link1 for all rows of data

all(raw_data$direct_link_hh1 == paper_data$willingness_link1)

perform same check for directness_link_2 and willingness_link2

all(raw_data$direct_link_hh2 == paper_data$willingness_link2)

Let's see if we can recreate the results of DeWeerdt (2006)

library(igraph)

First let's work with the paper_data and use the underreporting model

g1 <- graph_from_data_frame(paper_data[paper_data$willingness_link1 == 1 | paper_data$willingness_link2,], directed = FALSE) g1 <- simplify(g1)

g1_order <- length(V(g1)) g1_size <- length(E(g1)) g1_mean_degree <- mean(degree(g1)) g1_median_degree <- median(degree(g1)) g1_min_degree <- min(degree(g1)) g1_max_degree <- max(degree(g1)) g1_distances <- distance_table(g1)$res

Now let's try the same thing with the raw_data

g2 <- graph_from_data_frame(raw_data[raw_data$direct_link_hh1 == 1 | raw_data$direct_link_hh2,], directed = FALSE) g2 <- simplify(g2)

g2_order <- length(V(g2)) g2_size <- length(E(g2)) g2_mean_degree <- mean(degree(g2)) g2_median_degree <- median(degree(g2)) g2_min_degree <- min(degree(g2)) g2_max_degree <- max(degree(g2)) g2_distances <- distance_table(g2)$res

Check if the results are the same

g1_order == g2_order g1_size == g2_size g1_mean_degree == g2_mean_degree g1_max_degree == g2_max_degree g1_median_degree == g2_median_degree g1_min_degree == g2_min_degree all(g1_distances == g2_distances)

arnoblalam/risk_sharing documentation built on May 10, 2019, 1:46 p.m.