WA Votes 2013 – the #wapol context

Over the next couple of months, I will be collecting and analysing Twitter data (technology permitting) around the WA state election, to be held on 9 March 2013: see this post for an introduction to this project.

As the election nears, now is a good time to look at the WA political landscape as it appears on Twitter, before the heightened activity that usually accompanies election campaigns. As I mentioned in my previous post, setting the foundations for this project, my QUT colleagues and I tracked several hashtags pertaining to Australian political discussions over the course of 2012. These hashtags included all-purpose signifiers such as #auspol, #qldpol, and #wapol – hashtags denoting political topics (at a federal or state level) without being issue-specific.

A preliminary overview of activity around these three hashtags was presented in October last year in a paper by myself, Axel Bruns, and Stephen Harrington. In this post, though, I want to expand on the #wapol analysis from that paper in order to contextualise WA politics on Twitter.

the #wapol year

2012 saw the final year of the 38th Parliament, formed following the state election in September 2008. Late in 2011, the Parliament passed a bill introducing fixed terms for future parliaments and state elections, establishing the date for the next state election as 9 March 2013. This meant that 2012 provided a full year of government business, without the possibility of an early election being called. The approaching election also influenced events during 2012; in January, for example, the ALP leadership changed hands, with Mark McGowan taking over from Eric Ripper – the change gave the Opposition a year to promote McGowan as a viable alternative to the Liberal Premier Colin Barnett, as part of formulating its strategy for attempting to win back office, which it had lost in the 2008 election.

The change of Opposition leader, provoked by Eric Ripper’s resignation on 17 January, provides the first spike in tweets containing the #wapol hashtag during 2012. The following graph shows the total tweets per day for 2012, from 15 January to 8 January 2013. In general, the hashtag does not generate extensive tweeting activity, in comparison to other political signifiers such as #auspol or #qldpol. Even when breaking news causes increased activity, the peak activity recorded is still under 500 posts per day.

On average, #wapol is featured in 61.5 tweets per day – this amount varies dramatically depending on the day of the week, though. Tuesdays, Wednesdays and Thursdays see the highest averages, each with over 70 #wapol tweets per day respectively. Conversely, Saturday and Sunday are the quietest #wapol days, averaging around 35 tweets per day each. This pattern is in keeping with behaviour from other online political discussions, such as the Australian political blogosphere, where activity markedly decreased on weekends. What this year-long activity shows is that there is some consistent interest in tweeting about WA politics, with spikes around major news stories, even if the level of activity is lower than other state-oriented hashtags (again, the limits of the single hashtag approach should be noted here – this is not representative of all discussion around WA politics, but just the tweets containing #wapol).

The next question, then, is who is tweeting about #wapol? Below are the 25 user accounts who included #wapol in their tweets (or retweets of other users) between 15 January and 12 December 2012. Featuring among the list are journalists (the ABC’s Oliver Peterson and Alisha O’Flaherty), media outlets (WA Business News), political consultants and strategists (Darren Brown), politicians (led by Ken Travers), environmental groups (Conservation Council of WA (@nevillenumbat), Forest Rescue WA), and one of the fake Colin Barnett accounts (YourMateColin). The low level of overall #wapol activity is put further into context by the appearance of the @wa2013vote account as the 12th most active user – its 228 tweets gathered here were all published from 26 November onwards:

[Full disclosure: I am responsible for the @hansardblog tweets]

the #wapol networks

Tweeting a lot about #wapol does not necessarily mean that anyone is reading or responding to a given user, though; the final part of this contextual review investigates the connections between users through #wapol tweets by visualising the network of @mentions, @replies, and retweets.  These network maps are created by extracting any users mentioned (preceded by an @ symbol) in each #wapol tweet. A two-column matrix is then generated, showing each mention as a from-to relationship: the user who published the original tweet appears in the ‘from’ column, the user mentioned in the tweet is in the ‘to’ column. This information is used to create a directed network illustrating the shape of #wapol discussions. In the following visualisations, the size of each user (node) is based on their out-degree (the number of links to other users). I’ve also provided some initial colour based on party affiliation for the politicians, candidates, and party accounts represented in the network – blue for the Liberal Party, orange for the Nationals, red for the ALP, pink for independents, and green for the Greens. For visual clarity, I have also filtered the network so that only those nodes with a degree range of 12 or more (giving and/or receiving a combined total of at least 12 connections to/from other nodes) are shown in the map.

January 2012-December 2012 (a year’s cumulative connections)

The first network map draws on the total links between users found in #wapol tweets between 15 January and 12 December 2012. What is immediately apparent here is the strong ALP presence, seen through the nodes coloured red and the red edges connecting them to other nodes in the network. This is accompanied by the presence of clusters of users around the Nationals (in orange, at the top of the map) and the Greens (to the right). In comparison, there is a more limited Liberal presence, and those politicians which are represented here are not among the most active users, based both on the total tweets published and the @replies/mentions received.

Looking into the structure of the network in more detail (apologies for the low quality version of the map uploaded here – will address this in future posts), what the visualisation demonstrates is that there is a central group of Twitter users either linking to, or being linked to, in the #wapol discussion. Among these accounts are media organisations (including the major WA media, such as The West Australian) and journalists, and prominent ALP members (such as Eric Ripper and Ken Travers). This central group is then connected to other clusters – to the left, additional ALP accounts appear, for example, while to the Greens grouping to the right forms part of a larger cluster around related issues (such as environmental groups and activists). The presence of this topical cluster, as well as the ideological clustering around parties for both the Greens and the Nationals, is in keeping with similar findings for both the Twittersphere and the political blogosphere.

Of course, it can also be argued that the ALP grouping is also an ideological cluster, at least in part, despite its greater size within the network than the clusters formed around other parties. The ALP section may also suggest a model of tweeting which also explains the clustering for the other parties, and which will be explored further during the campaign. The ALP cluster contains several prominent accounts close to the centre of the network, connected to the other accounts central to the network (central in that they connect, or are connected by, different clusters, not necessarily that they are hubs of information). These ALP accounts appear to be closer to the ongoing debate than others – the further out from the centre you go, the ALP accounts are less densely interlinked with the central accounts. The presence of some ALP accounts as more central than others may be because of the particular roles these members hold in the Opposition (shadow ministers, spokespeople, rather than back-benchers, for instance), or the electorates they represent. Their prominence may also be due to their tweeting patterns – a member who regularly tweets and replies to tweets might expect/attract tweets from others seeking further information/comments – however, these patterns may also have developed because of the other factors mentioned.

This is a directed network, which means that the edges shown have a definite ‘from’ and ‘to’ relationship – while the edge connects two nodes, this does not mean that it is a reciprocated connection. A tweet from one Twitter user mentioning another account will not necessarily be replied to by the cited user. There are also several further limitations to this map. First, it presents a year’s worth of tweets without any attempt at distinguishing between time of publication, intentions, or topic discussed; these are the prominent accounts within the #wapol dataset, on a quantitative level, but the activity around one user, for example, could all take place within one day out of the year featured here, or infrequently across several months – this map does not differentiate between these different activities. Second, as I noted in my previous post (and will continue to do so), using #wapol is not a requirement for tweets about WA politics; while it has been adopted as a general signifier, it is not universally used. Similarly, even if a tweet includes the hashtag, replies to it from other users might not. What this means is that the data represented here are only a small subset of the total tweets discussing, or published by key figures (politicians, journalists, commentators, union representatives, and so on) within WA politics.

Why publish this map, then? Because, despite these limitations, it still serves some important purposes – it is just necessary to also note that it does not explain everything. What the map does provide is an overview of who is involved in Twitter discussions of WA politics – prominent users in terms of @mentions, for example – who is citing/replying to them, and what kinds of linking patterns can be found through tweets, such as the topical clusters. These overall patterns can then be used as a baseline for comparison with subsets of the data, such as tweets within a given period of time, involving specific users, or around particular topics. For example, the map below covers the connections through @mentions, @replies, and retweets in #wapol tweets published between 12 December 2012 and 8 January 2013:

December 2012-January 2013

In this map, some of the aspects of the cumulative map are repeated – the number and prominence of some ALP accounts, such as Ken Travers, for example, as well as a Greens group. The overall clustering effect is not as apparent, though – however, it should also be noted that this takes into account the Christmas and New Year period, when there were some lulls in activity. Over the upcoming weeks, I’ll be examining the #wapol discussion further, investigating whether or not these patterns are replicated during the election campaign, as well as related hashtags, and the user-specific networks – those generated by all tweets posted by, or mentioning, a given user, rather than relying on the subset of them that contain a specific term.