Project Description

In this assignment you will explore the relationships among various network-based factors and the emotional and personal health of individuals. This context is broad and vague. As part of this assignment, you will decide what type of social network to analyze, as well as which emotional or personal health issue to investigate. For example, ethnographic networks and the impact on happiness. We will examine such networks characteristics as cycles and redundant ties and explain how their presence (or absence) affect one’s personal heath. Your hypotheses and findings should be supported by formal statistical testing and analysis.

Topic Explanation

Social networks influence physical health. “A stream of research has examined the importance of social integration—one’s membership in a diverse social network—for health and longevity (Berkman & Syme, 1979; Cohen & Janicki-Deverts, 2009). Studies in this line of research have found that individuals with diverse social networks live longer (Berkman & Glass, 2000), retain their cognitive skills when aging (Fratiglioni, Paillard-Borg, & Winblad, 2004), have greater resistance to infectious disease (Cohen, Doyle, Skoner, Rabin, & Gwaltney, 1997), and enjoy a better prognosis when facing chronic life-threatening illnesses (Kop et al., 2005; Rutledge et al., 2004)” (Social Network Analysis: Methods and Examples, 163).

Alcohol

“Two of the most studied social behaviors relevant to physical fitness are smoking and alcohol use” (Social Network Analysis: Methods and Examples, 164).

For drinking, the right social ties can help resolve drinking problems. “Being involved with others and receiving high levels of support from even one person prior to treatment leads to better outcomes for quitting drinking” (Social Network Analysis: Methods and Examples 164). This implies that the reverse is also true. If someone surrounds themselves with other people who drink, then they themselves are more likely to drink.

Smoking

They type of behavior associated with social networks and alcoholism is also commonly seen with smoking. “Christakis and Fowler (2008) also studied the smoking behavior among the individuals of the Framingham Heart Study. Their analysis once again revealed discernible clusters of smokers and nonsmokers who tended to stop smoking around the same time” (Social Network Analysis: Methods and Examples, 164). This study also found that smokers tend to move further to the periphery of the social network over time.

Drug Use

“Another health area in which SNA has been often applied is the use of illicit drugs. Illicit drug abuse is a significant public health problem because of its numerous negative health and social consequences” (Social Network Analysis: Methods and Examples, 166).

There is a significant amount of research that has already been done that supports the idea that social networks play a significant role in illicit drug use. “Differential association theory (Sutherland & Cressey, 1974) maintains that adolescents learn such behavior from close friends or family who use or have a favorable attitude toward drugs. Empirically, illicit drug use is often initiated and supported in social networks that already involve drug use, providing a context that reinforces and sustains the behavior (Valente, Gallaher, & Mouttapa, 2004)” (Social Network Analysis: Methods and Examples, 166).

Peer influence is one of the main causes for an individual’s initial drug use.

Data Explanation

“The social network data were collected in the Teenage Friends and Lifestyle Study (West and Sweeting 1995, Michell and Amos 1997, Pearson and Michell 2000, Pearson and West 2003). Friendship network data and substance use were recorded for a cohort of pupils in a school in the West of Scotland. The panel data were recorded over a three year period starting in 1995, when the pupils were aged 13, and ending in 1997. A total of 160 pupils took part in the study, 129 of whom were present at all three measurement points. The friendship networks were formed by allowing the pupils to name up to twelve best friends. Pupils were also asked about substance use and adolescent behavior associated with, for instance, lifestyle, sporting behavior and tobacco, alcohol and cannabis consumption. The question on sporting activity asked if the pupil regularly took part in any sport, or go training for sport, out of school (e.g. football, gymnastics, skating, mountain biking). The school was representative of others in the region in terms of social class composition (Pearson and West 2003).” -https://www.stats.ox.ac.uk/~snijders/siena/s50_data.htm

In this report, we will look at the behavior of people in this network in terms of alcohol, smoking, and drug use.

Matrix Representation

This is a sample for the adjacency matrix for the data:

library(knitr)

raw.data = read.csv("s50_data/s50-network1.dat", sep = " ", header = FALSE)
raw.data = raw.data[,-1]
rownames(raw.data) = paste0("P", 1:50)
colnames(raw.data) = paste0("P", 1:50)

kable(raw.data[1:10,1:10])
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10
P1 0 0 0 0 0 0 0 0 0 0
P2 0 0 0 0 0 0 1 0 0 0
P3 0 0 0 1 0 0 0 0 1 0
P4 0 0 1 0 0 0 0 0 1 0
P5 0 0 0 0 0 0 0 0 0 0
P6 0 0 0 0 0 0 0 1 0 0
P7 0 1 0 0 0 0 0 0 0 0
P8 0 0 0 0 0 1 0 0 0 0
P9 0 0 1 1 0 0 0 0 0 0
P10 0 0 0 0 0 0 0 0 0 0

Network Representation

This is a visualization of the network:

library(igraph)

net = graph_from_adjacency_matrix(as.matrix(raw.data))
net = as.undirected(net)
plot(net, vertex.label = NA, vertex.color = "red")

Network Analysis

Alcohol

This is a visualization of the network with the students, identifying those who drink at least once a month:

net.alcohol = read.csv("s50_data/s50-alcohol.dat", sep = " ", header = FALSE)
net.alcohol = net.alcohol[,2]
for (i in 1:length(net.alcohol)) {
  if (net.alcohol[i] <= 2) {
    net.alcohol[i] = FALSE
  } else {
    net.alcohol[i] = TRUE
  }
}

plot(net, vertex.label = NA, vertex.color = c("#C39BD3", "#76D7C4")[1+net.alcohol])
legend(x = -2.5, y = -1.0, c("At least once a month", "Less than once a month"),
       pt.bg = c("#C39BD3", "#76D7C4"), bty = "n", pch = 21, pt.cex = 2)

Looking at the network, it seems as though the students that drink alcohol more regularly than others are pretty well dispersed throughout the network as a whole. The exceptions are two outcroppings of friends who are exclusively friends with other girls that drink at least once a month. This supports the idea that people who drink alcohol are more likely to do so if people in their social network also drink alcohol.

Cycles and Redundant Ties

It seems as though most of the students who do not drink alcohol are not connected to a larger alcoholic network. Most of the alcohol drinkers are in self-contained networks with no attachments to any other group.

Healthy Modification

One way to modify the network in a potentially healthy way would be to integrate the lonely alcoholic networks that are only connected to themselves into the larger network as a whole. A more significant affiliation with students who do not consume alcohol may encourage some of the more alcohol inclined students to drink less.

Smoking

This is a visualization of the network with the students, identifying those who smoke:

net.smoke = read.csv("s50_data/s50-smoke.dat", sep = " ", header = FALSE)
net.smoke = net.smoke[,2]
for (i in 1:length(net.smoke)) {
  if (net.smoke[i] == 1) {
    net.smoke[i] = FALSE
  } else {
    net.smoke[i] = TRUE
  }
}

plot(net, vertex.label = NA, vertex.color = c("#C39BD3", "#76D7C4")[1+net.smoke])
legend(x = -2.5, y = -1.0, c("Smokers", "Non-smokers"),
       pt.bg = c("#C39BD3", "#76D7C4"), bty = "n", pch = 21, pt.cex = 2)

In this network, most of the students are smokers. The non-smokers, for the most part, seem to form their own cluster.

Cycles and Redundant Ties

The cluster of non-smokers are connected to the rest of the network via multiple redundant ties. If only one edge connected the smoking and the non-smoking clusters, then there would be no redundant ties, and instead there would be a narrow path connecting the two clusters. In this case, it doesn’t seem like any of the non-smokers are cut off from the smokers.

Healthy Modification

Since most of the students are smokers, there isn’t much to be done to improve the network. It might help to spread the non-smokers in with the smokers more, in order to spread good influence, but that runs the risk of the majority peer pressuring the non-smokers into smoking, so its probably best to leave the non-smokers where they are.

Drug Use

Smoking

This is a visualization of the network with the students, identifying those who have consumed cannabis:

net.drugs = read.csv("s50_data/s50-drugs.dat", sep = " ", header = FALSE)
net.drugs = net.drugs[,2]
for (i in 1:length(net.drugs)) {
  if (net.drugs[i] == 1) {
    net.drugs[i] = FALSE
  } else {
    net.drugs[i] = TRUE
  }
}

plot(net, vertex.label = NA, vertex.color = c("#C39BD3", "#76D7C4")[1+net.drugs])
legend(x = -2.5, y = -1.0, c("Has consumed cannabis", "Has not consumed cannabis"),
       pt.bg = c("#C39BD3", "#76D7C4"), bty = "n", pch = 21, pt.cex = 2)

Again, in the same network, it seems as though the rebels outweigh the do-gooders. People who have consumed cannabis outnumber those who have not. There seems to be one main cluster of those who have not consumed cannabis. They are less connected to the rest of the network.

Cycles and Redundant Ties

There are not many redundant ties connecting the non-drug users and the drug users.

Healthy Modification

This is the same as the smoker section. It might be helpful to disperse the non-drug users in with the drug users, but the risk of peer pressure is too great.

References

(n.d.). Retrieved February 26, 2018, from https://www.stats.ox.ac.uk/~snijders/siena/s50_data.htm

Yang, S., Zhang, L., & Keller, F. B. (2017). Social network analysis: methods and examples. Los Angeles: Sage.