NI PoliGraph, a new site for NI politics data…

NI PoliGraph is a new site created to provide ongoing data-driven analysis of Northern Irish politics. The site tracks data from multiple sources – Twitter, news outlets, the Assembly, and opinion polls – for the NI political parties, and our MLAs and MPs.

What the site contains

Twitter is used to some degree by the majority of MLAs and MPs, and although there is a considerable amount of ‘noise’ in the content of what members post or retweet, we can nevertheless learn something by tracking this activity: who is using Twitter the most, who is having the biggest impact, how party members are interacting with one another, and (to an extent) what is being tweeted. Below shows the most impactful members on Twitter, as measured by their average number of retweets per tweet in the past month (as of late November 2020).

The richest set of data on the site is on NI Assembly activity: ministerial questions and answers, committee activity, transcripts of debates, and votes, obtained through the AIMS Portal. Natural language processing is used to track the topics that are debated in plenary sessions, and statistics are calculated on how members and parties have voted. This section includes the option of viewing previous Assembly sessions, back to 2007, for the chance to be reacquainted with some forgotten names from the recent past.

Also included are mentions of the politicians in a range of online news sources, and an opinion poll tracker running from 2015 to present.

Finally, these pieces of data are collated into a ‘report card’ for each politician, to put the member’s overall activity in context, relative to their peers: how effectively they use Twitter, how often they are mentioned in the media, and the amount and nature of their contributions to the Assembly, including their voting record in the current session (for MLAs). This is inspired, in part, by similar functionality provided for Westminster by They Work For You, and for the US Congress by GovTrack. The Assembly activity portion of the report card can highlight a member’s favourite topics of discussion, and, to some extent, the tone of their debate contributions, as seen in this example (for Paul Frew MLA, as of late November 2020).

Assembly voting trends

The site is also home to occasional analyses, using the various data being collected.

The first instance of this uses Assembly voting data from 2007 to present. One trend that we find here is that the rate of ‘Aye’ votes has increased during this period, resulting in motions more often finding majority support. However, the requirement of cross-community support has also increased, leading to only a slight increase in the frequency of motions passing.

We can also look at how each individual MLA has tended to vote, depending on whether the motion was proposed by Unionist or Nationalist members (excluding motions proposed by members with designation ‘Other’, or by a mixture of designations). In the plot below, points in the lower-right corner represent members that have usually voted to support a motion proposed by Unionist members and voted against motions proposed by Nationalist members; points in the upper-left represent the reverse scenario. Some interesting outlier MLAs are highlighted.

Clearly, the designation of the motion proposer(s) is a strong indicator of how a member will vote on the motion, but there is a spectrum of strictness of adherence to this tribalist practice. Alliance members sit mostly in the centre of the plot, meaning that they have supported both Unionist and Nationalist motions about half of the time. There are some members and parties – those situated towards the upper-right corner of the plot – that show a greater tendency than others to vote in support of a motion in general, whether it has been proposed by members of the same or a different designation.

See the full article at NI PoliGraph.

What NI PoliGraph is for

Through the individual member report cards, in particular, I hope that this site can increase the scrutiny that our elected representatives receive, by allowing readers to focus on their MLA’s actual performance, rather than simply going by party label. It could provide useful reference material for ‘traditional’ political journalism, such as is featured on Slugger O’Toole. At the party level, the voting records and social media data can clarify what each party ‘means’ and how they relate to one another, which is particularly useful information in the context of the ranked choice voting system that we use in NI for Assembly elections. Finally, it is for people like me that like looking at graphs and are curious about long- and short-term trends in politics that might be uncovered through data.