INTRODUCTION AND OBJECTIVES

Every time an event of global scope occurs, whether it is something unexpected or if it is an event organized in advance, social networks become an active forum for the exchange of information and debate, a phenomenon approached from multiple approaches (Kim; Hastak, 2018; ) (Demszky et al., 2019; Guallar, Suau, Ruiz-Caballero, Sáez, & Masip, 2016; Hassan, Gomaa, Khoriba, & Haggag, 2018) and through concrete examples of viralization in social events and movements, either specifically in previous climate summits (Segerberg; Bennet, 2011) or any other topic of debate (Shiau Chou et al, 2020) or global event, such as the recent covid-19 pandemic (Noor, Guo, Shah, Fournier-Viger, & Nawaz, 2020).

This flow of information is especially striking on Twitter, especially if the topic is perceived as a matter of general interest and has the participation of actors who stimulate the discussion and gather attention around them. An example of this has been the COP25 World Climate Summit held in Madrid (Spain) in December 2019, in which the two aforementioned elements were present. On the one hand, climate change, which has settled at the center of the political and citizen agenda, and whose footprint on social networks has already been studied in previous summits (Hopke; Hestres, 2018; Tyagi; Babcock; Carley, 2019). On the other hand, the irruption in the conversation of the Swedish media activist Greta Thunberg, present at the summit, whose relevance is unquestionable both on Twitter and in the social and political debate.

In the first case, her verified account accumulates 4.5 million followers and close to 9,000 tweets published in two and a half years (data as of January 2021). Furthermore, in the weeks before the summit, this social network was one of her main communication channels, such as during her boat trip, which was followed through the account @Sailing_LaVaga and the hashtag #LaVagabonde, the name of the catamaran in which she was traveling.

Regarding her social and media repercussion, Thunberg has been nominated twice (2019 and 2020) for the Nobel Peace Prize, she was one of the winners of the Right Livelihood Award 2019 (also known as the “Alternative Nobel Prize”) three months Before COP25, and two days before the end of the Madrid summit, she was named the person of the year by Time magazine, the youngest to receive this recognition in the 93-year history of this award. Furthermore, she has spoken before the main leaders and with some of them, she has held official meetings, such as with the Chancellor of Germany, Angela Merkel.

Beyond the individual contributions of Twitter users, a good part of the discussion on this social network revolves around what is published by the media, which see more and more on this social network (Nielsen; Schroder, 2014 ) an interesting channel to increase the traffic of their information, within their role, adapted to the new times, of setting the public agenda (Skogerbo; Krumsvik, 2014) selecting the debate topics -gatekeeping- (Mcquail, 2016) and adapting the political concept of agenda-setting (Alonso-Muñoz; Casero-Ripollés, 2018) to the new formats and the proposals of topics made by the users themselves ( ; Coddington; Holton, 2014) (Singer, 2014). And as for the activism aspect, social networks are undoubtedly first-rate speakers for NGOs (Blight; Ruppel; Schoenbauer, 2017; Carrasco-Polaino; Villar-Cirujano; Martín-Cárdaba, 2018).

With these four elements (the social network Twitter, the Climate Summit, the media, and the media-figure/activist), this research aims to answer a series of questions:

To what extent did the media publish content related to Greta Thunberg?

How was Greta Thunberg's interaction with the media through Twitter, if any?

METHODOLOGY

To answer these questions and hypotheses, all the tweets and their respective interactions (responses, mentions, and retweets) containing the hashtag #cop25 between December 2^nd and 12^th, 2019, dates on which the summit was held, were downloaded through the NodeXl pro program (Smith, 2010).

From the database generated with all the interactions (n=67,431), those messages and interactions published by the media and journalists who had their Twitter user account verified (n=350) and Greta Thunberg's interactions with those of the issuers were filtered. With these data, a graph was generated to determine the relationship that the different media had with Greta Thunberg and vice versa (mentions, responses, and retweets) regarding the topic of the event.

In a second step, the engagement generated by each of the actions was calculated. Engagement on Twitter is generally defined as the interaction rate. It is an index that measures the ability of a tweet to generate interest or action on Twitter. It is a relevant index because besides establishing an emotional relationship with users, it has value when the Twitter algorithm organically shows content on the timelines of the different users with whom there has been interaction (Inza & E, 2020). It is not about the number of followers, but the degree of involvement or interest that followers present to certain publications (Montells, 2016). There are two most widespread ways to calculate engagement or interaction rate: in the first of them, the total of interactions received by a tweet is divided by the number of impressions (the number of people who have received it on their timeline). In the second, the total number of interactions received by the tweet is divided by the number of followers of the author of the tweet (Inza & E, 2020; Montells, 2016). In this way and due to the type of data collected and its structure, it was decided on the second mentioned formula. Thus, the operation used to calculate engagement was established as engagement = ((total interactions received) / number of followers) *100 (Herrera-Torres, Pérez-Tur, García-Fernández, & Fernández-Gavira, 2017).

For each publication, a content analysis was carried out (Piñuel-Raigada, 2002) through a previously designed file that consisted of the following items:

• Type of account: to determine if the issuing user was a corporate media outlet or a professional journalist from their private account.

• Type of publication: a distinction was made between published original tweets, mentions, replies, retweets, and mentions on retweets.

• Contains multimedia elements: In this way, a distinction was made between publications that had an image, a video, or an animated GIF inserted in their body and those that did not contain any of these resources.

• Contains URL: The publications that contained in their body a URL that could be clicked and it directed to a page or element other than the tweet and its interactions were distinguished.

• Mentions Greta Thunberg: Those publications that included Greta Thunberg's name in their body were identified from those that did not.

• @Gretathunberg is mentioned: In this case, the publications that directly included the identifier of Greta Thunberg's account (@gretathunberg) in their body were selected.

To analyze the engagement of each post according to the different variables, previously those publications that presented an atypical engagement (outliers) were identified and filtered through the calculation of the typical engagement scores of each post. As the sample was of more than 80 records, it was decided to filter those posts with a typical score with a value higher than the absolute value of 3 (Estrada, Batanero, & Fortuny, 2004).

Six records with atypical engagement were identified, which were filtered and not processed together with the rest for the inferential statistical analysis.

To determine the application of the type of statistical tests that allowed us to see how engagement varied following each of the other variables and if the presented differences were statistically significant, normality tests were performed using the Kolmogorov-Smirnov (Massey, 1951) or Shapiro-Wilk (Royston, 1992) systems, depending on whether the sample of each group within each of the variables was greater or less than 50 records.

The normality tests showed negative results, so it was decided to do a non-parametric analysis through the Mann-Whitney U test (Macfarland et al., 2016) if the variable to be analyzed presented two Kruskal-Wallis groups (Ostertagová, Ostertag, & Kováč, 2014) when the variables presented more than two groups.

ANALYSIS OF RESULTS

DISCUSSION AND CONCLUSIONS

Original tweets generated less engagement than any other type of interaction, although the differences were not significant. Despite these non-relevant differences, it was possible to have responded to user publications besides retweeting, seeking greater engagement in the accounts of media outlets and journalists. Posting original content on Twitter and not following up on it by interacting with users and responding to their observations and comments, does not allow us to squeeze the potential of the social network.

When preparing a post on Twitter, it was not relevant in terms of engagement if any multimedia element was included in it, be it an image, video, or animated GIF since this feature did not affect the achieved engagement. However, inserting a URL into the content did have negative effects. By inserting a URL, the recipient is being offered an alternative to leave the conversation or the social network and send them to an external website. As an informative strategy, it could be correct if these links are directed to information or content from the same media outlet or journalist, because despite reducing the interaction with the tweet, visits to the destination website to which the URL directs are increased, thus fulfilling the fundamental objective for which the media resort to this microblogging network.

Regarding this aspect, if the media and journalists planned to use Twitter as a tool to promote their own content hosted on their respective websites, the strategy of inserting links to their information could be considered correct. However, if the goal was to use Twitter as a standalone information-and-conversation channel, inserting a URL into posts might have been an error.

The topics covered by the information sector on Twitter during the celebration of the climate summit were varied and did not focus exclusively on the character of Greta Thunberg (she is mentioned in 19% of all publications).

Greta Thunberg did not react to any of the mentions received, nor did she respond, retweet, or mentioned on a retweet, drastically reducing the potential engagement of Twitter posts from news industry accounts. Surely, if any of the posts published by the media or journalists who had mentioned the @gretathunberg account had received a reaction from her, these tweets would have become much more viral. It should be noted that when a user is mentioned in any tweet, they receive an alert, whereas, if instead of the user's account identifier what is included in the body of the publication is their name, this alert is not executed, so it is difficult for the person mentioned to take any action on this publication (they do not know that they have been mentioned), thereby reducing its potential viralization. For this reason, it could be considered an error on the part of the information sector to have not included in the publications that dealt with topics related to Greta Thunberg her user account instead of her name as text. However, it is impossible to determine in which cases the absence of direct mention is a voluntary or involuntary act, even if logic leads to thinking about the latter regarding the media.

The action of journalists and the media on Twitter during the climate summit held in 2019 in Madrid focused on the dissemination of their own content through URLs inserted in their tweets, without making subsequent efforts to maintain the conversation about the published topics. The conversation did not focus on the character of Greta Thunberg, but when they did inform about her, the potential of having such a media figure in the event was not exploited, as her username was not included in the tweets.

Despite the obvious limitations implicit in every analysis of a specific event, such as the reduced time frame or the limitations of the tools for obtaining complete data that overcome the barriers imposed by social networks, the conclusions of this study show the convenience that for the media may have to focus their efforts on Twitter on simple decisions that do not require deep training in social networks, such as not including external URLs to avoid diverting traffic or directly mentioning the account of a media figure related to the topic of discussion to increase engagement, instead of making an indirect mention. Subsequent studies on other discussion topics in which the two elements analyzed now coincide (a global event and a figure with great media coverage) will show whether these patterns are repeated, nuanced, or modified.

REFERENCES