Visual Voting Analysis Shows Sinn Fein Enjoy Strongest Voting Discipline On Dublin City Council
One of our core beliefs at DataChemist is that viewing data visually can tell us things that otherwise remain concealed - or at least hard to intuitively grasp.
To show what we mean, we’ve taken a sample set of data - voting records of Dublin City Councillors - and shown visually how ‘close’ individual councillors are in terms of those records. The results can be seen here.
In the simplest possible terms, the closer two individuals are to each other, the more similar their voting records. By presenting the data in this way, we can see immediately ‘what’s going on’ in a way that would simply be impossible using any other approach.
Specifically, the data allows us to draw the following conclusions:
- Political parties do tend to vote in a reasonably homogeneous fashion (the distribution is not random, or anything close to it)
- Sinn Fein enjoy the strongest disciplinary record in terms of council votes, with all councillors clustered closely together (Noeleen Reilly, the ‘interloper’, is a former member of the party)
- Fine Gael and Fianna Fail are heavily inter-mingled, to the extent that they could almost be considered a single party
- The same could be said for the Green Party and Labour Party, who present as a single block in this analysis
These results are interesting in and of themselves. But in addition, the visualisation of data in this way enables us to see a continuum across the political spectrum, and accurately place parties on that spectrum based on real data rather than perception or prejudice.
In this analysis, it appears there is a quite identifiable left / right polarity. Fianna Fáil and Fine Gael occupy one end of the spectrum, and People Before Profit, Solidarity, The Workers’ Party occupy the opposite, while the Sinn Fein and Labour-Green clusters occupy the centre.
Lastly this data shows clearly the politics of any given independent just by looking which parties they are situated with. A few independents seem to occupy an equivalent status to Sinn Féin or Fianna Fáil-Fine Gael. A few hover around the Solidarity-People-Before Profit-Workers’ Party nexus. However, the voting positions of the Greens and Labour are not apparently so popular with the unaffiliated.
Housing Votes Show A More Complex Picture
Using precisely the same methodology, we limited data only to those votes relating to housing policy. The results, which can be seen here, showed that party discipline in general breaks down, or is at least less ‘tight’ when it comes to this particular area.
The left is apparently little moved, but individuals from Fine Gael, Fianna Fail, Labour, Sinn Féin and the Greens become more interwoven. For the most part, Sinn Féin does retain the centre, the usual suspects retain the left, but the position of a given councillor from the other parties becomes much more difficult to guess, and Labour look decidedly further right. This is perhaps food for thought given the scale of the current housing crisis.
Together with the data above, this is of course just one example of how any sample set of data can be shown visually to deliver new insights that escape conventional analysis. That is the promise - or one aspect of the promise - of DataChemist.
About The Data And Process
We used our tools to inspect the voting records of local government councillors in Dublin City Council. We produced a similarity measure, which related the number of times a given politician voted in concert with another politician. The similarity is a kind of distance, where the smaller the number, the more similar the voting behaviour. We then flesh out all the relationships between every councillor to create a fully-connected graph.
The graph layout model uses a physics engine which lays out our councillors with some repulsion (so they don’t all bunch up), but which tries to make equal weighted edges, equidistant. Since this isn’t in general possible to do perfectly in 2-Dimensions with N objects, we have to try to stretch some edges and compress others. The cost of this compression and stretching is computed incrementally, leading to a process of relaxation from the original arbitrary layout, to a low energy state.
The final result is such that councillors that vote in concert frequently will tend to be close to each-other, whereas those who voted infrequently will tend to be further apart.
The layout is inexact, acting as it does to find a low-energy state by a process of moving the councillors around small jumps. This process might never relax fully so a decay is over-layed in order to ensure that they eventually come to a halt. If one doubts that a given councillor is placed appropriately, dragging them to a new position can be instructive. Sometimes it’s possible to resituate them, but generally speaking things are not allowed to wander too far.
December 31st, 2018
Share this post:
Answering The Really Big Questions
Posted March 25th, 2019 by Luke Feeney
Gartner and DataChemist Agree: Graph Databases Are Set For Growth
Posted March 4th, 2019 by Kevin Feeney