Pearl Street Cooperative is student housing cooperative in Austin, Texas with over 100 residents that I lived in from fall of 2016 to spring 2017. In an effort to better understand the social network of the house, I used a combination of tools visualize the network structure to uncover hidden themes and trends.


Using Google Forms to share a survey of the house, I asked the responders to rate each individual (n=108) in the house from 0 to 3 based on how well connected they are to the individual. 0 being, ‘I don’t recognize this name’, 1 being, ‘I know who this person is but we don’t interact’, 2 being, ‘we are friends, we eat together but don’t regularly hang out one on one’, 3 being, ‘we are very close friends, we hang out in each other’s rooms regularly’.

41 individuals responded, which fills out the 11664 cell matrix nicely if we assume that the network is non-directional. Using Rstudio packages ‘igraph’, ‘ggplot’, ‘network’, and ‘Matrix’ I was able to create the following visualizations from the data.



An initial visualization looking at scores of at least one shows just how well connected all of the individuals within Pearl Street are.


Taking one more step back, here is the visualization of the same data, only graphing connections when scores are 2 or greater.


Finally, here is the visual representation of ‘best friends’ within the house. Even at the most exclusive and highest score, the network is still very well connected.

Diving a bit deeper, I investigated the “coreness” of the graph, using the k-core algorithm.


The network has a very robust 35 core center. There are 48 individuals in the highest core, representing not only a very tight-nit, but a large tight-nit community. At this point, it is worth noting that within the house there are people who are known as “Ghosts”, meaning these people live in the house, but do not contribute socially, and are only seen entering and leaving their rooms. This k-core analysis is the first way to begin quantitatively identifying Ghosts.

Screen Shot 2017-06-06 at 4.58.59 PM

This graph can be easily sorted into 3 groups. One group is everyone in core 35, most integral to the social structure of the house. The second group, from cores ~24-34 are starting to trail off in terms of importance to the house. Finally, the last group in the tail, cores 0-24 are individuals who do not contribute to the house, informally known as Ghosts.

Another metric to investigate is differences between genders in the house, particularly in average scores given and average scores received. A total of 21 girls and 20 guys participated in the survey, making comparisons between genders relatively accurate. The average sum of scores given from boys was 159.55 compared with a score given by girls of 159.45, remarkably close with a difference of 0.06%. Scores received, on the other hand, was not as close. Boys received an average score of 57.41, while girls received an average score of 60.29. A difference of around 5% higher received score for girls. The discrepancy between these two difference in genders of given and received scores suggests that within the house, boys perceive their connectedness to girls higher than girls perceive their connectedness to boys. This doesn’t surprise me very much.

More investigation into visualizing the data can show us how individual “Pearlies” perceive their connectedness to the house, versus how the house views them. (n=41)Screen Shot 2017-06-06 at 5.18.34 PM

This graph is a simple representation of the sum of scores given by each individual who participated. I should note here that people who took the survey, were typically people who are socially well integrated into the house; people who were in the higher “coreness” of the k-core graph. Ghosts did not take the survey, further fulfilling their stereotype. There is one major group within this graph and outliers on either end.

Screen Shot 2017-06-06 at 5.23.44 PM.png

This suggests that there are individuals who both over and underestimate their importance to the houses social structure (there is one star here at the top who really thinks they have a lot of friends!).

Conversely, I created a similar visualization which shows the scores received by members of the house. (n=108)

Screen Shot 2017-06-06 at 5.18.40 PM

All things said and done, this basically shows how popular individuals within the house are, how many connections they have made with others. Similarly to the k-core graph, the tail here on the left side indicated who the “Ghosts” are. It is also interesting that there is a steady increase in the score received for individuals after Ghosts are accounted for. There are 3 individuals at the top who break this trend, although not particularly significantly. Other than the slight boost the top few people have, there is no one in the house who is ‘super popular’ or significantly more popular than others. On the other hand, there are a few handfuls of people who are very unpopular.

On a micro level, there are individuals who do not perceive their integrality within the house to be very high but were actually given quite high scores. Also, some individuals who gave very high scores of connectivity while not receiving such high scores themselves, with the difference between these two being significant.


  1. Out of the house of 108 people, 48 individuals represent the highest “coreness” or integrality to the social network of the house. It could be argued that somewhere around 50 is the ideal number of people for a maximum utopian social structure, which is ironic because there are co-ops in Austin of 10-20 people, and co-ops of ~100 people. Much more data would be needed to back this claim up.
  2. Within this house, boys perceive their connectedness to girls higher than girls perceive that to boys. This may have something to do with cooties.
  3. About 10% of the house are “Ghosts” and intentionally do not contribute to the social cooperativeness that the house culture tries to create.
  4. The house is very well connected socially. Even at the highest degree of connections between best friends, the average degree of separation between any individuals in the house is not greater than 2.

Final Thoughts

Is there enough data to really draw any significant conclusions from the data? I’m not sure but it makes for a good art project. I am interested in seeing how this information changes in both space and time. If you would like to recreate this data visualization for your living situation, reach out to me for more information on the methodology. I would love to learn how this compares with other cooperatives, fraternities, sororities, and apartment complexes. Furthermore, I am curious on how the social connections in a living situation like this contribute to the mental health of the individuals living there.

What are the implications that research like this has on revealing the social solidarity of both private and public organizations and how does this impact the success of the organization?

Github Link to the code that makes these informatics possible