2022 Australian Federal Election: Update 1

By Daniel Angus and Axel Bruns

It’s finally here: with the Prime Minister’s visit to the Governor-General on 10 April, and after months of preliminary quasi-campaigning, the 2022 Australian federal election campaign proper has begun. As we have done with past campaigns including the 2019 federal election and the 2020 Queensland state election, our team of DMRC researchers is once again tracking the social media campaign. This time around, as before, we are monitoring activity by and around the candidate accounts and pages on Twitter and Facebook – but through projects with the ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S) and the Social Media Lab at Ryerson University, we are also paying detailed attention to the parties, candidates, and other groups as they run election-related advertising campaigns on Facebook. In this first post from the digital campaign trail, we’ll offer a first overview of what we’ve seen so far. We’ll provide further updates on the election as it unfolds on a weekly basis.

Bring out your wares

While the digital political campaigns of the various candidates and parties have been gradually expanding to include more platforms, resourcing and sophistication, researchers too have been honing our own tools and methods to track and analyse these activities. While researchers have and continue to perform some of this work at odds with the platforms themselves due to restrictive terms of service, some notable tooling has been developed in time for this current Australian election specifically for sponsored content.

PoliDashboard is an initiative from the Social Media Lab at Ryerson University, which was used to track Facebook advertising in the recent Canadian election. Our Ryerson colleagues have collaborated with the DMRC to extend PoliDashboard to Australia to provide a single source of information on ads being placed by authorised and properly declared parties and interest groups on Facebook. PoliDashboard enables aggregation of targeting information obtained from Facebook’s transparency library (ad spend, binary gender, age brackets, state).

Colleagues at the University of Queensland have launched a similar ad tracker dashboard aimed at providing breakdowns and insights into campaign spending. A considerable amount of work has occurred behind the scenes to collate and match candidates to their Facebook pages, and to their respective electorates. The interface is really clean and provides some great insights into which electorates are getting the most significant spends and from whom.

Like PoliDashboard, UQ’s tool accesses and accumulates data directly from Facebook’s Ad Transparency library, but neither of these tools are able to capture political advertising if it isn’t properly declared or authorised. This is where a third tool steps in, the Australian Ad Observatory. Through our involvement in the ARC Centre of Excellence for Automated Decision-Making and Society, we have built the Ad Observatory as a citizen data donation platform which enables Australian Facebook users to anonymously donate sponsored content (and only this content) that they encounter on Facebook via their laptop or desktop computer. By capturing ads at the point of consumption we will be in a better position to locate any ads that may fall foul of Facebook’s own advertising standards, but which still exist on the platform. In partnership with the ABC we are encouraging Australian Facebook users to download a simple plugin, register basic demographic details (if they choose), and join a growing group of citizen scientists in trying to locate any ‘dark’ ads that may be circulating.

Advertising in the first weeks

In terms of Facebook advertising spend, there are some early trends emerging that are altogether unsurprising. In the Senate race the most significant ad purchases have been made by Rex Patrick (IND – South Australia) ~$8000, Clive Palmer (UAP – Queensland) ~$4,600, and Jacqui Lambie (JLN – Tasmania) ~$4,200. In the lower house race Kooyong is the electorate which has had the most cash spent so far at ~$50,000, which is almost double that of the next-ranked electorates of North Sydney and Wentworth at ~$20,000 each. In Kooyong it is the incumbent Josh Frydenberg who has put around $30,000 into this blue-ribbon seat in a bid to hold off Teal Independent challenger Dr Monique Ryan, who has spent about half as much at $18,000. Overall, Labor have spent ~$395k, while the Liberal Party has spent ~$270k, Independents ~$115k, UAP $106k, and Greens $29k.

Figure 1: The lower house contest has seen a significant early spend in Kooyong, and some consistently high spending by Independent candidates (figure generated from https://www.election-ad-data.uq.edu.au/).

The content of advertising in the first week focussed significantly on voter enrolment, with many candidates spreading messages encouraging constituents to enrol, and to ensure that their enrolment details were up to date. This messaging seems to have worked as the AEC reported a record number of voter enrolments over the Easter weekend, and projected total enrolment of eligible voters will be just shy of 97%.

Figure 2: An example ‘enrolment’ ad run by the Australian Labor Party. This ad was exclusively targeted to 18-24-year-olds in NSW.

The Liberal Party have already run a considerable number of attack-style ads, questioning the credibility of the opposition leader Anthony Albanese on several issues such as the economy, and accusing Labor of lying. The Labor Party have instead focussed on positive policy messaging, outlining value statements and plans for supporting health, education, jobs, and clean energy.


Figures 3 & 4: These ads from the Liberal Party and Labor Party highlight a stark campaign difference in these first weeks, with the Liberals pursuing an attack strategy while the ALP adopt policy-focussed messaging.

We have not caught any ‘dark’ ads via the Ad Observatory yet, but if any are located we will share them in coming weeks.

Social media engagement around the candidates

Nominations to stand as a candidate in the 2022 election closed yesterday, on 21 April, so there may still be some surprises in who has chosen to run for office that will be revealed when the full candidate rolls are published. However, many candidate announcements especially by the major parties were made some time ago, and we have begun tracking their Twitter and Facebook activities already. Here we’ll provide an overview of the broad patterns to date – with the caveat that we are focussing for now only on tracking House of Representatives candidates, and may still be missing some minor party and lower-profile Independent candidates in our data. We’ll add those and the Senate candidates in future iterations of this analysis.

For this first update, we’ll cover the period from 4 to 21 April, and focus on activity and engagement patterns on Twitter.

Labor and Greens candidates have traditionally been the most enthusiastic Twitter users, and this trend is continuing in the campaign to date. Even their efforts are outdone, however, by the candidates fielded by Clive Palmer’s United Australia Party, who have already posted nearly 3,600 tweets since 4 April. This is due predominantly to one particular UAP candidate, however, who has produced well over half of those tweets, mainly by retweeting misinformation and conspiracy theories from US-based Twitter accounts. (We won’t name that account here, to avoid further amplifying this content.)

The performance by Independent candidates is also notable: in keeping with the media debate of a ‘teal wave’ facing major party candidates at the polls, Independent candidates have been highly active at posting on Twitter, too. Coalition candidates, by contrast, have traditionally been only reluctant Twitter users, and this trend is continuing to date: Liberal candidates have posted less than 10% of the number of tweets published by their Labor counterparts, and LNP as well as National Party candidates have been even less active.

Figure 5: Tweets by and at candidates, aggregated by party, on 4-21 Apr. 2022.

However, such inactivity does not turn these candidates into smaller targets for engagement by other Twitter users. On this metric, Labor and Liberal candidates are broadly even, with around 500,000 tweets directed at each side to date. They differ markedly in terms of the type of tweets they receive, though: while some 17% of all tweets directed at Labor are retweets (resulting in the further amplification and perhaps even endorsement of their messages), for Liberal candidates that figure remains below 1%. This may result in part simply from the relative scarcity of retweetable posts from Liberal candidates, of course, but this notable difference nonetheless seems to point to the general public’s profound reluctance so far to share on the Liberals’ messaging. (The same is true also for engagement with the other Coalition parties.)

While Labor does well on this front, the Independents and UAP candidates do even better: almost one third of all engagement with Independent candidates is by retweeting their posts, and this rises to nearly 40% for UAP candidates. This aligns well with indications of growing support for Independents and minor parties since the last election.

But such attention is very unevenly distributed, of course: as in previous elections, the two major candidates for the Prime Ministership receive most of the attention on Twitter. Here, Labor’s Anthony Albanese is slightly ahead of the Liberals’ Scott Morrison, with between 350,000 and 400,000 tweets directed at each man, and other notable candidates lagging far behind.

Figure 6: Tweets at key candidates, on 4-21 Apr. 2022.

Notable here are again also the performance of United Australia Party candidate Craig Kelly and the two prominent Independents Monique Ryan and Zoe Daniel, standing respectively against Liberals Josh Frydenberg and Tim Wilson in the electorates of Kooyong and Goldstein. While the two Independents receive somewhat fewer tweets overall than their Liberal opponents, they benefit from a far greater number of retweets, repeating overall patterns at the party level. Frydenberg and Wilson each barely reach a single-digit percentage, while 40% of Ryan’s and 26% of Daniel’s engagement is in the form of retweets. (What remains to be seen on election day, however, is whether this reflects their national visibility, or genuine support by the local electorate.)

At the top, PM Morrison is similarly starved of retweets: less than one quarter of a percent of his Twitter engagement is in the form of retweets. Opposition Leader Albanese does somewhat better (11%), but this too lags well behind even the retweet averages for the Labor Party as a whole. This seems to indicate that the Twitter public has yet to completely warm to either leader. We will continue to monitor these figures over the remainder of the election campaign.

Next steps

The early stages of a campaign are always messy – not just for the candidates, as Albanese and Morrison have already found, but also for the general public as well as us researchers seeking to track campaign activities and engagement. Now that the candidate rolls have closed, we’ll complete our review of the candidates’ Twitter and Facebook presences and extend our tracking to the full list of candidates across both houses. For our next update, this should mean a considerably more comprehensive picture of their social media performance to date.

Meanwhile, our analysis of advertising patterns also continues at speed. With the campaign now in full swing, we’ll see how quickly candidates and their parties may respond to new developments, gaffes, policy initiatives, and other campaign events in their advertising strategies – and we’ll keep a close eye on any nefarious ‘dark ads’ that may appear in the mix, too.

Indeed, with Anthony Albanese now in COVID quarantine, online campaigning is likely to play an even greater role in the overall election campaign over the coming week: Labor will be forced to adjust to the absence of its Prime Ministerial candidate from public events and promote the ‘virtual Albo’ instead. This time next week, we should be able to tell how well that strategy has worked out.