Avec près de 400 000 gazouillis depuis le début de la campagne électorale en Ontario, il peut être difficile de départager la teneur des conversations. L’analyse des réseaux permet justement de décerner qui influence les conversations en ligne, les communautés qui se forment et les enjeux qui motivent ces communautés. En l’espèce, il s’avère que le soutien en ligne aux libéraux est le reflet de ce que nous voyons dans les sondages. De plus, on constate que les communautés d’intérêt partisanes sont moins motivées par les politiques de leur propres partis que par la critique des propositions (ou l’absence de propositions) chez leurs compétiteurs. (Le billet est en anglais.)
Since the writ dropped on May 7, 394,593 tweets have been written that mention either #OnPoli, @fordnation, @Kathleen_Wynne, or @AndreaHorwath in Ontario alone. With such a large amount of content, it’s difficult to identify the influential voices. This is where network analysis is useful.
What is network analysis?
In its most basic form, network analysis identifies connections between objects (see NATIONAL’s Kevin McCann’s piece for more information). A network analysis using Twitter data is created by visualizing all connections between authors of tweets and the sharers of tweets.
If you are struggling to visualize the process, picture this: when someone shares a tweet, the original author and the sharer appear in a network graph as two nodes (or circles) connected by a single line. As more people share the same original tweet that author becomes connected to each of their nodes and the size of the author’s node grows, indicating greater influence in the conversation.
In addition to identifying influential individuals, network analysis can show communities, or clusters, which are groups of people that typically interact with each other.
Network analysis of the Ontario Provincial Election
By conducting a network analysis using Twitter data from May 7th to May 25th, we found that there are four main communities in the Twitter conversation (represented by different colours in the visual below): media organizations (pink), the Progressive Conservatives (blue), the NDP (orange), and the Liberals (red). What may surprise you are the individuals who are leading the conversations in each community as well as the issues that matter to each community.
Conversation originating from media organizations, predominately @CBCQueensPark, @TorontoStar, and @nationalpost, accounts for about 21 per cent of the total conversation. Although media organizations receive criticism for being partisan, this network analysis showed that content from media organizations is being engaged with by people regardless of the political party they tend to engage with. Content from media organizations that received the most engagement focused on Hydro One board member salaries, indicating that hydro is a hot-button issue for Ontarians.
The Progressive Conservatives
Conversation in the Progressive Conservative community is lead by @fordnation, @sunlorrie (Twitter account of Lorrie Goldstein, former reporter at the Toronto Sun), @OntarioIsProud, and @DrJacobsRad. This community accounts for 18.5 per cent of total conversation and tends to engage with content that is critical of NDP budget proposals and previous NDP governments, as well as critiques of the Liberal Party’s lack of investment in healthcare and education. Most conversation within this group is reactive to the actions of other parties.
Similar to the results from recent polls, the NDP community accounts for slightly less conversation (17.6 per cent) than the Progressive Conservatives. Aside from @AndreaHorwath and @OntarioNDP, the individuals driving conversation in the NDP community are reporters who are close to Queen’s Park, including: @MariekeWalsh, @ChrisAReynolds, and @Tom_Parkin. These reporters are not leading conversation because they are posting pro-NDP content, rather they are leading conversation by posting content that discusses potential wrongdoings within the PC Party. Like the PC community, the NDP community’s conversation is generally reactive to the actions of other parties.
Much like recent polls, the Liberal community is trailing the Conservatives and NDP with 12 per cent of total Twitter conversation. Unlike the PC and NDP communities, there are no individuals leading conversation beyond the party leader (@Kathleen_Wynne) and the official party account (@OntLiberal), which likely indicates a general lack of public support for Liberal messaging. Unlike the engaging content from the PCs and NDP, engaging Liberal messaging focuses on policy, specifically healthcare and safe-injection sites, however, these policy-related tweets do not attract the same level of engagement as the potential scandals shared within other partisan communities.
Through network analysis, we can take a large and incredibly noisy issue and easily identify the conversation influencers, communities, and the issues that motivate communities. By applying network analysis to the Ontario election, we see the amount of support for the Liberals on social media resembles the support seen in the polls. Furthermore, we can see that partisan communities are motivated not by their own party’s policies, but by criticizing the policies, or lack thereof, of their competitors.
NATIONAL will be monitoring social conversation in the days leading up to the election. Be sure to look for an updated post on June 8 for an analysis of what people were discussing as they went to the polls.
——— Blaine Mackie était directeur adjoint, Données et analytique au Cabinet de relations publiques NATIONAL