Network analysis on the voting behaviour of members of the Hong Kong Legislative Council
A network analysis on the similarity of LegCo members' voting behaviour. The analysis examines the voting record of the Fifth LegCo (2012-16) and the Sixth LegCo (2016-19) up until May 2019.
- preprocessing.R: Scripts to fetch the voting record using the LegCo API, compute the similarity between different lawmakers and export as vote_table5.rds and vote_table6.rds
- network.R: Scripts to plot network graphs in English and Chinese as 5th_all_en.png, 5th_all_zh.png, 5th_pro_est_en.png, 5th_pro_est_zh.png, 5th_non_pro_est_en.png, 5th_non_pro_est_zh.png, 6th_all_en.png, 6th_all_zh.png, 6th_pro_est_en.png, 6th_pro_est_zh.png, 6th_non_pro_est_en.png and 6th_non_pro_est_zh.png
- pro_est5.rds & pro_est6.rds: Lists of pro-establishment lawmakers in the Fifth and Sixth LegCo
After collecting all votes casted by every member in the Fifth and Sixth LegCo meetings, a correlation matrix containing the similarity of the voting behaviour between every member is generated for each term, which can then be used to plot weighted network graphs with each node representing a lawmaker and the correlation coefficient indicating the colour depth of each edge. This can provide a glimpse of the level of agreement between lawmakers across different political parties and camps.
Members who quitted early or joined late as lawmakers or have casted less than 50% of valid votes (i.e. yes or no only, not abstaining or absent) overall are excluded from the analysis. A number of motions submitted for the sole purpose of filibustering are also removed, e.g. member's amendment bills on the Approporiation Bill in 2013-16.
You need the R packages elgarteo/legco and elgarteo/legcoplus to fetch data from the LegCo API.
All other packages are obtainable from CRAN.
The method of constructing a network graph by correlation matrix is adapted from jmbh/bundestag.