2020 Queensland State Election: Week 3 Update

By Daniel Angus and Axel Bruns.

Given that we are now only one week from polling day, we start this week’s update by looking into the themes of the campaign to date. We focus our thematic analysis on content we have obtained from candidate Facebook pages, tweets sent by candidates, and tweets directed towards candidates from 3 to 21 October inclusive.

Beginning with Facebook, we used the Leximancer text analytics package to derive the top 25 high-level concepts from all of the text located in Facebook posts made to candidate pages. As we observed last week, it’s hard to escape the parochialism of this state, and as expected Queensland and Queenslanders are both trending high in posts. We also observe substantial amounts of self- and cross-references to political parties (LNP, Labor, and Greens) amongst Facebook activity. We note that the economy has been a massive focus for all parties – with talk of jobs, plans, work, and investment all featuring prominently. We have also seen a strong focus on social support, with health, schools, and community all trending high.

At least at this high level, we haven’t observed a significant degree of change over time when it comes to the content of Facebook posts. Most tend to relate at least partly to the themes above, inflected with geographic or policy-specific information such as an announcements about a local hospitals, schools, roads, factories, etc.

For Twitter, we examined all tweets from candidate accounts and those that were directed to candidate accounts (either through retweets of candidates or @mentions of them). Within this it is a similar story to Facebook, but in this dataset we also gain access to the voice of the public (Facebook does not provide access to user comments). For the Twitter data, we note four significant themes relating to discussion of the economy, of Scott Morrison’s appearance on the campaign trail, of the former Newman government, and of the Townsville night-time curfew proposed by the Liberal National Party.

The economic discussion is framed through words such as economy, jobs, open, plan, grow, rebuild, and is the most politically neutral of all topics, shared by most of the parties and supporters. The topic relating to PM Scott Morrison relates to ongoing discussions around his appearance in Queensland, which included an altercation at the University of Queensland, a party fundraiser, and a plane break-down. The Newman government topic featured early in the campaign, largely driven by the Queensland Labor Party’s campaigning with keywords that included cut, axe, sack, and ‘last time’. The final topic only emerged this week, and relates to the LNP’s controversial curfew proposal to address youth crime in Townsville. This generated significant opposition to this on Twitter, with users accusing the party of racist, dog-whistle politics.

These thematic patterns are also reflected in the overall engagement with parties and candidates on social media. During the past seven days on Twitter, we observe a growing volume of tweets (including @mentions and retweets) directed at LNP candidates: from Monday 19 October, around 50% of all tweets to candidate accounts are targeting LNP candidates, and this increases even further from Wednesday morning, when LNP candidates command up to 90% of all the attention directed at Queensland election candidates.

This substantial volume of tweets is prompted predominantly by the LNP’s curfew proposal that morning, and indeed the vast majority of all tweets at LNP candidates that day are focussing on Opposition Leader Deb Frecklington. The reaction to the proposal is mixed at best; we also observe a substantial number of tweets containing the hashtag #DictatorDeb, and pointing out the similarity between the Townsville curfew and the continuing coronavirus lock-downs in Melbourne, which the Victorian state Liberal Party opposes vehemently. #DictatorDeb itself is of course also a play on the #DictatorDan hashtag used by opponents of Victorian Labor Premier Dan Andrews.

This negativity (at least on Twitter) is also reflected by the overall interaction patterns for the week: while LNP candidates receive a substantially greater proportion of all tweets at candidates, very few of those interactions are helping to amplify the LNPs messages through retweeting. Less than 2% of the tweets received by LNP candidates are retweets, compared to some 25% for Labor and 39% for the Greens. Even if the curfew proposal is received well in Townsville electorates, then, it may well have a significant negative impact elsewhere in Queensland.

Patterns on Facebook are considerably different. Once again ignoring simple ‘likes’, and treating ‘sad’ and ‘angry’ as negative, and the other reactions as positive, we see that the overall sentiment in user reactions to the candidates’ posts is mainly positive for all parties. There are substantial differences, though: Greens and Labor candidates receive more strongly positive reactions overall, while One Nation candidates attract a considerably larger proportion of ‘sad’ and ‘angry’ reactions than the candidates of the other parties.

As we noted last week, however, such negativity is not necessarily always directed at the party candidates themselves: users often simply share the candidates’ professed sadness or anger about the issues raised in their posts.

LNP candidates, meanwhile, have thus far failed to generate much in the way of clearly positive or negative responses: no more than 10% of all reactions to their posts are clearly positive, and very few are clearly negative; the vast majority are simple likes. This may indicate a failure to energise the LNP base: to date, the posts by its candidates appear to have neither particularly enthused them, nor inspired antipathy towards the other side.

The campaigning strategies that produce such reactions are illustrated by examples from different parties. The posts by Premier Palaszczuk that generate the greatest number of reactions tend to be positive – from campaign announcements about plans to provide free sanitary products to schoolgirls to a snapshot of her mother; meanwhile, some of her colleagues are generating more negative reactions by directly attacking the LNP, One Nation, and Clive Palmer.

For Opposition Leader Deb Frecklington, the picture is more mixed: an anniversary message to her husband is received positively, while a post about youth crime in Townsville and Cairns produces highly negative reactions. These are likely to consist of a mix of reactions that are angry about the situation in the region, and about the LNP’s proposed solutions.

In the case of One Nation, many of its candidates’ most successful posts elicit predominantly negative responses: this points to a more negative campaigning style which seeks to position the party as an alternative to both Labor and the LNP, but attacks the sitting Labor government most aggressively.

Notably, such aggressive campaigning has generated a greater amount of reactions, comments, and shares for posts by One Nation candidates during this past week than for those by another minor party, the Greens; One Nation posts receive a very substantial number of shares, in particular. It remains to be seen at the ballot box whether this negative campaigning approach serves only to energise the faithful or also manages to attract a substantial number of additional voters.

Lastly an update on advertising spending on Facebook. The Clive Palmer’s United Australia Party is still trending as the highest spender of all of the registered parties in this election ($84.8k), with the LNP ($59.3k) and Labor ($57.9k) nearly tied in second place.

Lobby groups, unions, and issue advocates have also been spending up considerably on Facebook advertising, with the likes of GetUp!, Advance Australia, Queensland Resources Council, Shooters Union of Australia, Workers’ Health Alliance, and Queensland Unions amongst the top spenders for the campaign to date.

And that’s it for this update – one more week to go!

Facebook data from CrowdTangle, a public insights tool owned and operated by Facebook.

X