The working paper Investigating Transnational Communication on Social Media is now available on OSF.
This paper was presented and discussed at the workshop titled Transnationalization of the Far-Right, held at The Johns Hopkins University, School of Advanced International Studies, Bologna, Italy, on Tuesday, April 8th, 2025. It marks the beginning of an effort to systematically conceptualize and operationalize the study of transnational communication on social media.
The Pearson’s r coefficient is the most popular metric of correlation. In the social sciences, researchers not attuned to the nuances of statistics may be tempted to use it every time they need to compute the correlation between variables, including time series. However, several aspects should be considered when applying this method to time series data.
I present three scenarios where the correlation coefficient can be misleading without deeper data analysis:
Series with a common long-term trend but divergent short-term behavior.
Series with common seasonality or cyclical fluctuations.
Time series with regime-switching characteristics, exhibiting different behaviors at different stages.
These examples illustrate why calculating correlations between time series requires careful consideration and a thorough understanding of the underlying processes.
Case 1: Common Trend
Two series may have a common trend, for example, increasing or decreasing together. Such behavior returns a very high value of Pearson’s correlation coefficient. In the case of the series below the correlation coefficient is 0.99, indicative of an almost perfect positive correlation.
TIme series sharing a common trend.
However, if we look at how the series proceed point by point, we can see even antithetical behavior (one goes up, the other goes down) or completely random behavior. In the case of this example, if we subtract the trend from the series and calculate the correlation coefficient again, we find a value indicative of an inverse rather than positive correlation: r=-0.38.
The two series above, after being detrended.
Trending series are rather frequent. For example, measuring the correlation between scientific publications in two completely different and unrelated subject areas can return a high correlation coefficient simply because the global scientific productivity increases over time due to the ‘publish or perish’ culture, but without any other notable relationship between the two specific series.
Case 2: Common Seasonality
Two series that follow a common seasonality are generative of a high correlation coefficient. As with regard to trend, such correlation due to seasonality may mask opposite correlation or nonexistent correlation. The series represented below show a remarkable correlation of 0.70, using Pearson’s r coefficient.
Series sharing common seasonality.
However, when we go to subtract the seasonal component and recalculate the correlation coefficient, we find that the series are substantially uncorrelated (r = -0.04).
The two series above, after having seasonality removed.
Seasonality is common in social science data. For example, the daily frequency of social conversations on two topics may manifest a seasonal component with more conversations on weekends than on weekdays, producing a high correlation coefficient despite the fact that the two conversation topics are not correlated.
Case 3: Regime‐Switching Time Series
In the third case, we find two time series that are characterized by a sequence of different processes. In the example, they are first characterized by a common positive trend, then proceed flatly with negligible variability, and in the third phase take on opposite trends. Such series can be called regime-switching time series because they are characterized by parameters taking different values in each of a series of regimes or phases.
Calculating the correlation coefficient between the two series, we find an almost complete lack of correlation (r=-0.07) despite the fact that the human eye may suspect the presence of common determinant factors. In fact, if we break the series into their three phases and calculate the correlation coefficient for each of them, we find an r=0.99 in the first phase, r=0.00 in the second phase, and an r=-0.99 in the third phase.
In the paper Protest and repression on social media: Pro-Navalny and pro-government mobilization dynamics and coordination patterns on Russian Twitter1, we used a preliminary changepoint analysis to differentiate between different phases of social media mobilization for and against Alexey Navalny before proceeding to analyze them individually.
Conclusions
Time series are data with a special nature and therefore require specific statistical tools. It often happens that correlations due to common seasonality or trends patterns produce high correlation coefficients, even though such patterns are not determined by the processes the analyst intends to measure. Time series can also have varying behaviors over time: new factors may come into play, representing different processes. This happens all the more easily as the series covers larger time frames and in the case of communicative and social processes.
Misleading time series correlations are what is commonly referred to in the field of time series analysis as spurious correlations, an unclear term that I hope I have helped, in part, to clarify. The topic is indeed complex and multifaceted. To delve a bit deeper into the main concepts of classic time series analysis, a primer can be found in the online handbook on the topic that I wrote for my Master’s students in Communication Science at the University of Vienna2.
References
Kulichkina, A., Righetti, N., & Waldherr, A. (2024). Protest and repression on social media: Pro-Navalny and pro-government mobilization dynamics and coordination patterns on Russian Twitter. New Media & Society, https://doi.org/10.1177/14614448241254126.
For verification professionals and disinformation researchers, it is crucial not only to access existing verification data but also to identify and monitor the networks that may disseminate problematic content. This webinar will discuss ways to identify coordinated networks on social media, exemplified by the Coordinated Sharing Detection Service, a prototype for analysing and visualising social media activity through patterns of coordinated sharing. By uploading their data, users can explore and uncover networks of coordinated behaviour, offering actionable insights for diverse research and monitoring purposes. Key features include dynamic network visualisations, exploration of coordinated content, and downloadable outputs for further exploration.
This webinar, which targets anyone interested in the technological aspects of AI-based fact-checking, is open to EBU Members and Associates, as well as other media professionals.
The veraAI project is co-financed by the European Union, Horizon Europe programme, Grant Agreement No 101070093, with additional funding from Innovate UK grant No 10039055 and the Swiss State Secretariat for Education, Research and Innovation (SERI) under contract No 22.00245.
I recently had the pleasure of collaborating with Phillip Ayoub and Kristina Stoeckl on their influential work, “The Global Fight Against Lgbti Rights: How Transnational Conservative Networks Target Sexual and Gender Minorities” (NYU Press). This project involved creating data visualizations that captured the analytical and communicative intent of the authors, starting from relatively complex datasets. The process, which included rigorous qualitative reviews by NYU editors, was both challenging and rewarding.
Visualizing the networks on the maps was challenging due to the relatively high number of nodes and relationships. Finding the right combination of graphical elements to allow for a clear representation required evaluating a variety of alternatives. I used R with rnaturalearth and the ggplot2 function geom_sf().
Visualizing the roles of actors and organizations in the various World Congress of Families required combining visual and analytical logic. The decision was made to position actors and organizations with a more constant presence at the base, visually emphasizing their foundational role in the events, with others gradually placed in higher positions. The use of color transparency further highlights both the constant presence (represented by less transparent colors) and the fluctuating or occasional presence (represented by higher transparency) at the events. For this purpose, I used a heatmap with the geom_tile() function in ggplot2. All data preprocessing and visualizations were done in R.
The paper “CooRTweet: A Generalized R Software for Coordinated Network Detection” (Righetti & Balluff) describing our CooRTweet software, has been accepted for publication on Computational Communication Research and will soon be available in the journal. The pre-print version is available on OSF: https://osf.io/zya2x_v3/
[Note added April 17, 2025] In the past few days, this blog post has received an unusual spike in visits, linked to distribution via a mailing list I was able to trace. I also received a violent and hateful email, sent from a mobile iPhone app and an IP address I identified and reported to the Italian and Austrian authorities. I stand by the scientific work discussed here and will not tolerate coordinated online harassment, threats, or intimidation. Academic dialogue and dissent — as well as support for any moral values — must be grounded in reason, not hate.
[Original blog post] My latest work, now online in Information, Communication & Society, explores how social media is used to amplify moral conservative advocacy in the digital space.
In the paper “Mainstreaming and Transnationalization of Anti-Gender Ideas through Social Media: The Case of CitizenGO”, along with an international group of talented young researchers, we examine how major social media platforms like Facebook serve as powerful tools for disseminating and globalizing radical ‘anti-gender’ ideas. Despite the growing transnational appeal of these ideas, the role of digital networks in this process remains largely underexplored. By analyzing a decade of multilingual social media activity (2013–2022) from the leading conservative organization CitizenGO using advanced computational methods, we shed light on key strategies that drive the digital presence of the anti-gender agenda.
Our findings reveal that CitizenGO strategically employs a diverse array of social media accounts to coordinate and amplify moral-conservative messages across different languages and regions. An amplification network sharing their content also facilitates their rapid transnational expansion.
For those interested in a deeper dive into our research, you can read the full paper here.
Righetti, N., Kulichkina, A., Almeida Paroni, B., Cseri, Z. F., Aguirre, S. I., & Maikovska, K. (2025). Mainstreaming and transnationalization of anti-gender ideas through social media: the case of CitizenGO. Information, Communication & Society, 1–24. doi.org/10.1080/1369118X.2025.2470229
First publication of 2025 on a highly unique case: a coordinated attack by pro-Vietnam activists on a post from the Facebook page of the Chinese Embassy in Italy. The post, which is still accessible (link), incidentally featured a map that included the Spratly and Paracel Islands—disputed by Vietnam—as part of Chinese territory.
Some of the concepts discussed in this article were presented during my research colloquium as a Visiting Research Fellow at the ZeMKi Centre for Media, Communication, and Information Research at the University of Bremen on May 15, 2024. I also benefited from discussions with Nguyen-Phuong Tran, currently a doctoral student in Urbino, who provided valuable insights into social media in Vietnam. Our conversations on the concept of online firestorms were particularly enriching.
Many thanks to the editors of ISR – Italian Sociological Review for their valuable work and for making this high-quality journal freely accessible.
For those interested, the full paper is available here: link.
In recent years, an increasing number of studies have focused on analyzing coordinated communication networks across social media platforms. These networks use coordination among various social media accounts to influence and manipulate the user base and the platforms themselves (Chan, 2022). This phenomenon is linked with several problematic activities, including the propagation of disinformation (Giglietto et al., 2020) and more extensive state-backed information operations.
Coordinated behaviour can be undertaken by both covert and less covert actors with varying objectives, ranging from mimicking organic engagement and support for content, to distributing content across social media pages through multi-actor broadcasting and re-posting, for example. In this project, we aim to perform an analysis of the diverse array of coordination forms, highlighting the analytical ambiguity of approaches that measure inauthenticity through the use of platform research affordances such as (timed) link-sharing and reposting.
Coordinated inauthentic behaviour campaigns on social media are driven by actors (and perhaps bots) pushing the same or related content in synchrony and causing it to gain virality, or some threshold of interactions and impressions to indicate a degree of popularity. The primary purpose is ambient. It is to ‘flood the space’, thereby exerting or appearing to exert a large measure of influence. The broader aim, for state and other political actors, could well be to develop a full-fledged counter-program to accrue symbolic power and assert political dominance (McIntyre 2018).
Research into coordinated inauthentic behaviour has demonstrated its operational as well as geographical breadth but also its platform dependence and orientation toward single platform studies (Thiele et al. 2023). It has been tied to corona-politics (Magelinski & Carley 2020), election misinformation (Nizzoli et al. 2021), protest repression in authoritarian regimes (Kulichkina et al. 2024), and cryptocurrency manipulation (Terenzi 2023). It is far-flung geographically but rather platform-specific in its targeting. Research has described coordinated networks on social media in Australia (Graham et al., 2021), Nigeria (Giglietto et al. 2022), South Korea (Keller et al., 2020), Philippine (Yu 2022), Brazil and France (Gruzd et al. 2022).
The algorithmic architecture of each platforms suggests coordination efforts around certain digital objects. For example, coordination on Twitter typically aims to push a hashtag into the trending topics, occasionally with ‘weaponised bots’ (Graham et al. 2021). On Facebook, website URLs are typically placed in posts on Pages and Groups, where the idea is to elicit emotive reactions, long comment threads and further sharing, which is the platform formula for algorithmic amplification (Merrill & Oremus 2021). The Facebook Feed is thereby persuaded to elevate these shared links now charged with emotive currency.
Through a thick description of a set of international coordinated networks on Facebook, this project advances the discourse on coordinated inauthentic behaviour on social media platforms by extending its study beyond influence operations. Based on techniques developed for its technical analysis – namely CooRnet (Giglietto, Righetti, Rossi, 2020) – we undertake an empirical study on Facebook that surfaces not only such operations but also activist networking, viral marketing, fan support, analytics-driven publishing and others in the service of mobilising attention.
Coordinated Inauthentic Behaviour on Facebook, Richard Rogers (UvA), Nicola Righetti (Univ Urbino Carlo Bo), Digital Methods Summer School and Data Sprint 2024
In this paper we analyze 200,000 posts published on Facebook pages and groups mentioning veganism, articulating a typology of social media functions for veganism.
In addition, we develop a critical methodological reflection on the limitations and potential of “big data” for studying the phenomenon.
I have made available on GitHub the slides used in the Research Design and Data Collection course I taught at the University of Vienna, in the international master’s program in Communication Science (2023/2024).
