The legislative elections in Portugal on 10 March took place within a very specific political context. They resulted from the unexpected resignation of a government with an absolute majority due to judicial investigations into corruption and influence peddling. Also, they took place amid a trend towards the growth of right-wing and far-right parties, which would be reflected in the final results.1
As far as the presence of parties and candidates on social media is concerned, the time was also ripe for expectation, given the trends detected of a decrease in the use of Facebook, traditionally the most relevant social network for political content in Portugal, and an increase in the use of other platforms, such as Instagram and TikTok.
In this context, MediaLab CIES Iscte and the Lusa news agency established a protocol to monitor and analyse the legislative elections on a weekly basis, between 1 February and 6 March, by collecting posts on Facebook, Twitter/X, Instagram and TikTok. The analysis was guided by four specific parameters: reach and impact of posts by the main candidates; reach and impact of posts by political parties with parliamentary representation; reach and impact of posts by other social media users mentioning the candidates; and reach and impact of campaign cases and disinformative content identified by fact-checkers. Therefore, the focus of analysis was the reach and impact of posts relating to the elections.
This report is divided into three sections. First, we frame online political discussion on the four main online social networks analysed (Facebook, Twitter/X, Instagram and TikTok). Secondly, the next section focuses on the online activity of political leaders and their parties, as well as the online discussion around them. In this section we analyse the evolution of the online activity over 5 weeks and highlight the main characteristics and dynamics detected on each social network.
Finally, we present a proposal consisting of four amplifying dimensions of online political discussion according to the trends observed in the online activity of the candidates, around them and around irregular and sometimes disinformative cases that occurred during the electoral campaign period.
In methodological terms, the social networks analysed are Facebook, Twitter/X, Instagram and TikTok. The data was collected using the CrowdTangle tools (for Facebook and Instagram), SentiOne (for Twitter/X and TikTok) and the TikTok Research API (for TikTok, naturally). In most cases (and where not stated otherwise), the data has been organised according to total interactions. The choice of total interactions as the main metric for the analyses is due to the fact that we wanted to analyse the messages that most captured the attention of the users of the social networks in question and which, therefore, may have had the greatest reach and impact.