paint-brush
Decoding the Popularity of TV Series: A Network Analysis Perspective: Resultsby@kinetograph
New Story

Decoding the Popularity of TV Series: A Network Analysis Perspective: Results

tldt arrow

Too Long; Didn't Read

In this study, researchers attempt to understand the psychology behind TV ratings through character networks, and the different interactions between characters.
featured image - Decoding the Popularity of TV Series: A Network Analysis Perspective: Results
Kinetograph: The Video Editing Technology Publication HackerNoon profile picture

Author:

(1) Melody Yu, Sage Hill School, California, USA.

IV. RESULTS

In this study, we calculated network metrics for three popular TV series: Game of Thrones, House of Cards, and Breaking Bad.


A. Network Metrics


We studied 22-26 episodes from the first three seasons of each show and calculated various network metrics for each episode. These metrics include density, efficiency, and transitivity, as well as node-level metrics such as degree, harmonic closeness centrality, and eigenvector closeness centrality.


For each metric, we also calculated the maximum and standard deviation values for each episode’s character network. The results of these calculations are shown in Appendix Table V, VI, and VII , with each row representing a single episode and the first two columns indicating the episode number and review. The remaining columns display the various network metrics for that episode.


B. Correlation between network metrics and episode reviews


To determine whether there is a relationship between the network metrics we calculated and the IMDB reviews, we used correlation analysis. Specifically, we used the Spearman correlation method, as it is well-suited for analyzing ordinal data such as TV episode review scores. While the absolute value of a review score (e.g. 8.7) may not provide much insight on its own, it is generally accepted that an episode with a higher review score is of higher quality than one with a lower score. As such, the review scores can be considered ordinal, with higher values indicating better quality. By using the Spearman correlation method, we were able to examine the relationship between the network metrics and the episode reviews and determine if there is a correlation between them.


TABLE ISPEARMAN CORRELATION OF GAME OF THRONES EPISODE REVIEWS.


Table I shows the results of the Spearman correlation analysis for the TV series Game of Thrones, examining the relationship between the network metrics and episode reviews. The analysis found two significant correlations: a negative correlation between the number of active nodes and episode review scores, and a positive correlation between the standard deviation of eigenvector centrality and episode review scores.


TABLE IISPEARMAN CORRELATION OF HOUSE OF CARDS EPISODE REVIEWS.


Table II shows the results of the Spearman correlation analysis for the TV series House of Cards, examining the relationship between the network metrics and episode reviews. The table indicates that there are six significant correlations between the two variables. However, we note that two of these correlations may be influenced by influential outliers, as visualized in Figure 6. After accounting for these outliers, the analysis found four significant correlations: a negative correlation between network efficiency and episode review, a negative correlation between maximum node degree and episode review, a negative correlation between node degree standard deviation and episode review, and a negative correlation between maximum harmonic centrality and episode review.


Fig. 6. Density vs Review for House of Cards


TABLE IIISPEARMAN CORRELATION OF BREAKING BAD EPISODE REVIEWS.


Table III shows the Spearman correlation between episode network metrics and episode reviews for the TV series Breaking Bad. One significant correlation between the network metrics and episode reviews is a positive correlation between network transitivity and episode review.


This paper is available on arxiv under CC 4.0 license.