Hi everyone… In the second semester of my graduate study, I got a midterm assignment about Social Network Analysis Movie Producers – Composers Analysis. In this blog post, you will expect some terminologies about social network, the tools and techniques to solve the problems, and interpretation about the findings. Keep in mind that this was a midterm assignment, so we will only use social network theory that had been taught to us.
This was a group assignment. I teamed up with Joshua Kuntandi (Indonesia), Harry Nguyen (Vietnam), and Luxi Vy (Vietnam). It was a pleasure to work with them and I really had fun times doing this assignment with them. We also ended up to be the best team for this assignment. Yeay!!
Keep in mind that this assignment only covers the first half of the techniques that were taught. So, you will see that we only use the basic technique. To make this post easier to read, I will explain about the description, objective, result, and the final takeaways. Happy reading!
We were given some information about the movie producers and composers network.
Data Information (Movies.paj)
- Two-mode network with 102 vertices: 62 producers and 40 composers
- 192 valued edges (cooperation of producer and composer; line values represent the number of films cooperated on)
- Partition identifies the composers and the producers (1-composers, 0-producers)
- in Pajek, ‘rows’ are the projected producer network and ‘columns’ are the projected composer network.
- In Hollywood, composers of soundtracks work on a freelance basis. For each movie, a producer hires a composer and negotiates a fee. The earnings of composers is highly skewed: a handful of composers earn a lot, whereas most of them have moderate or low revenues. This is characteristic of artistic labor markets. Why do some composers earn much more money than their colleagues in Hollywood?
- The network Movies.net contains the collaboration of forty composers and the sixty-two producers who produced a minimum of five (completed, shown, and reviewed) movies in Hollywood, 1964–76. This is a 2-mode network: a line between a composer and a producer indicates that the former created the soundtrack for the movie produced by the latter. The line values indicate the number of movies by one producer for which the composer created the music in the period 1964–76.
- Among the composers, Goldsmith, J., Schifrin, L., Grusin, D., Mancini, H., and Williams, J. are the most successful, each of whom earned 1.5 percent or more of the total income of Hollywood movie score composers in the 1960s and 1970s.
- What are the characteristics of theses successful composers?
Tips from Professor
- This is an open analysis competition. Please get as much as you can from the information provided (through the SNA techniques you have already learned).
- “Weighted Degree” centrality rather than “Degree” centrality provide more interesting results.
- To analyze producer and composer network (“Movies.paj”) and present the analysis results.
- The main question to answer is “what are the characteristics of the successful composers?”
How did we do?
- Use Pajek
- In this course, we used a software called Pajek. It is free and very easy to use. Some features might not be as powerful as other software, but Pajek is more than enough to do social network analysis.
- Hypothesis 1: A composer who works on many projects will be more successful
- Hypothesis 2: A composer who works with the same producer frequently and the highest number of producers will be more successful
- (additional from Professor) A composer who has a high “average collaboration per producer” value will be more successful
- Hypothesis 3: A composer who has high degree centrality / closeness centrality / betweenness centrality will be more successful
- Hypothesis 4: A composer who works with top producers will be more successful
Our analysis and findings (pictures are our actual presentation slides)
Hypothesis 1: A composer who works on many projects will be more successful
For the first hypothesis, we found that 4 out of 5 composers have many projects during the period except Grusin, D. One possible explanation about this is that Grusin, D. started his career in early 60s and only has a few projects. However, he worked in successful movies.
Hypothesis 2: A composer who works with the same producer frequently and the highest number of producers will be more successful
The producers that they are working with are not the top producers. So, this means their connection and exposure to other top producers is smaller.
These producers have more connection, therefore it promotes exchanging resources and information.
Most of the successful composers work with many producers. Thus, the average collaboration per producers is low.
Hypothesis 3: A composer who has high degree centrality / closeness centrality / betweenness centrality will be more successful
4 most successful composers share the same characteristics. They have high centrality in all measurements. What about Grusin, D.? We found that he started his career in early 60s. So, he may know many people (high degree centrality), but that doesn’t mean he is close with others (low closeness and betweeness centrality).
We found that all top composers have connection to each other. In the network, we can see that Grusin, D. has a connection to Goldsmith, J. (previously, we found that Goldsmith, J. has the highest centrality in all measurements). This could probably be one of the factor of his success.
Hypothesis 4: A composer who works with top producers will be more successful
Top composers work with top producers on top movies that most likely to produce expensive movies. Hence, these composers could get more exposure or higher pay.
- Successful composers work with top producers (most likely to produce expensive movies)
- They generally work on many projects and they work with many producers
- Successful composers have high degree centrality (Number of supporters, confidants, trading partners) which means they connect to many of their peer
- The successful composers have a strong connection with each other
- They generally have high closeness centrality (Access to and spread of information, opinion formation), betweenness centrality (have higher ability to control the flow of information or exchange resources)
I hope that this post could help some of you. I really enjoy this course. If you are currently taking this kind of course, I hope you could enjoy it too.