Blog

How We Accidentally Became Pandemic Communication Researchers

13. July 2020

(Authors: Daniela Stoltenberg, Maya de Vries Kedem, Hadas Gur-Ze’ev, Neta Kligler-Vilenchik, Annie Waldherr, Barbara Pfetsch.) With the Covid-19 pandemic touching all parts of life, academic research has not been an exception. Even for researchers who are able to maintain access to their field – for instance, through online research – considerable changes in the objects of study force them to rethink their research questions and study designs as they go along. The team behind CRC project B05 “Translocal Networks” reflects on their experiences of conducting a survey of intense Twitter users at the height of the first Covid-19 wave in Jerusalem.

Academic research is a slow process. The proposal for our “Translocal Networks” project was largely conceived in 2016 and 2017 – its goal: Studying the spaces and relational arrangements of routine communicative action on social media. Besides a large-scale analysis of Twitter communication in two contrasting cities – Berlin and Jerusalem – the project was to include a comparative survey study in the two cities, which aimed at understanding active Twitter users’ subjective perceptions of their communicative behavior and network. To this end, we were going to ask participants a variety of questions: Where were they at the time of media use? Where were their interaction partners or where did they imagine their audiences being located? How close did they feel to their interaction partners? Did the answers to these questions depend on the topic of their communication?

To approach these questions, we planned to use the Mobile Experience Sampling Method (MESM), a repeat survey technique where participants receive short questionnaires via text message several times a day over a multi-day period. MESM is recommended for studying routine media use that people do not think about much and would therefore have trouble giving accurate information about in retrospect.[1] We spent the better part of 2019 developing the questionnaire, pretesting it, translating it into four languages (English, German, Hebrew, and Arabic), and getting it up and running on the survey system.

While academic research is slow-moving and our survey was designed to capture routine behavior, a pandemic unfortunately is neither of those things. Still, having already wrapped up the Berlin survey in January and early February 2020, a pandemic is exactly what we found ourselves in the middle of the Jerusalem survey. We knew all along that Jerusalem is a contested space and its inhabitants are used to communicating during violent political crises. Yet COVID-19 is a different animal: it is a health emergency, rather than a political crisis, and its implications on the population cross the regular lines of spatial and political division that characterize the city (that is, the virus doesn’t “care” if it infects East Jerusalem Palestinians or West Jerusalem Jews).

Watching the circumstances shift in real time

Recruitment for the Jerusalem survey began on February 19 and we gave ourselves three weeks to motivate members of our target group – intense Twitter users located in the city – to participate, before starting the survey on 9 March. In a twist that seems funny in hindsight, we decided on relatively short notice to push the start of the survey back by a week – from March 2 to March 9 – to avoid the Knesset elections on March 2 having an undue influence on participants’ communicative behavior and make sure to capture a routine period in Jerusalem daily life. After that, a second cohort of participants was going to be recruited for interviews starting two weeks later. As of 19 February, neither Israel nor the Palestinian territories had confirmed a single COVID-19 infection, so we simply did not realize the jeopardy for our project. The first case of COVID-19 in Israel was confirmed just two days after we started recruitment and thereby set the survey process in motion. Still, we did not smell trouble.

That changed over the next couple of weeks, first gradually and then rapidly. Amongst the first countries in the world, Israel began locking down international travel into the country, culminating in mandatory home isolation policies for all foreign travellers in early March. This led to the last-minute cancellation of Daniela’s research visit set to take place between March and May. On March 2, Israel held national elections and set up secluded voting booths in which quarantined citizens could cast their vote. Polling staff watched in full protective gear, producing images that reverberated in the international press. 

On March 5, four days before the start of the survey, Neta, the Israeli team lead, emailed the group: “Please note that Israel is en-route towards full blown emergency mode with the Corona. There is (still unconfirmed) talk about limiting public transportation and possibly even closing schools! I’m noting this because it probably means we’ll see a bias towards health-related discourse on Twitter that is not necessarily representative of regular circumstances.”

To our recollection, this is the first time anyone acknowledged that the comparative dimension of our survey was in trouble. But even then we couldn’t have imagined just how far removed from routine things were about to get. Absent other options, we began texting participants links to our repeated short survey on March 9 as planned. The next day, Israel limited public gatherings to 2,000 people and just one day later again to 100 people and ultimately – before the week was out – to ten people. On March 12, schools and universities were closed, followed three days later by malls, restaurants, pools, beauty salons, zoos, entertainment venues, and cultural sites. On March 19, the final day of our first survey cohort, Prime Minister Benjamin Netanyahu declared a state of emergency, stating that Israelis were not allowed to leave their homes except for a limited list of essential activities. Meanwhile, Palestinian authorities in the West Bank had already declared a state of emergency and limited movement on March 5.

Within the ten-day period of our first survey cohort, Jerusalem had gone from mostly routine life to virtually complete lockdown. This largely remained the situation for our second survey cohort, which started on March 23 and ran through April 2. Over the course of the three-and-a-half-week survey period, case numbers in Israel had increased more than hundredfold, from 61 to 6,857. We had arguably found the worst possible time to study routine behavior.

Adjusting the design within three days

On the one hand, the situation was a researcher’s nightmare. Years of planning, conceptualizing, and implementing a comparative design were moot because the data in our two cities simply did not reflect comparable circumstances. 

On the other hand, we may consider ourselves lucky. While many of our colleagues in the Collaborative Research Center – not to mention the wider academic community – had to cancel ethnographic fieldwork and interviews, our online survey design allowed us to still collect data during an unprecedented time. It just wasn’t the data we set out to collect.

We had to ask ourselves how to proceed with this knowledge. Ultimately, the direction emerged straight from the field. Having answered about half of our repeat surveys, one participant, who was stuck in home quarantine, sent a message to her recruiter, Hadas: “I was just thinking that the corona situation may taint your data – I keep answering that I’m tweeting from home at times when I would not normally be home. Just something to consider. :)”

On Monday, March 16, three days before the first survey cohort was set to wrap up, this message provided the tipping point to adjust the design. The comparative dimension was dead in the water and our survey was not designed to capture communicative behavior during an unprecedented public health crisis. However, we could still try to understand the impact of the current circumstances on our object of study, that is, Twitter communication. For that, we realized we would need an additional questionnaire.

The following days passed in a frenzy. Having already communicated to our participants that the survey would end on Thursday, we were determined to keep our promise and have our new “Corona questionnaire” ready to send out by that day. Setting up our initial questionnaire took us the better part of a year – now we had three days.

Monday was spent emailing back and forth ideas for what we needed to ask. Lacking time to pretest the survey extensively, let alone review the relevant crisis communication literature for concepts and scales, we opted for a rather simple questionnaire. Using standardized measures, we were going to ask participants about the degree to which they stayed home more than usual and whether their Twitter use frequency had changed. We also included two item batteries from our recruitment survey measuring how central Twitter use was to participants’ day-to-day life and what motivated them to use it. Finally, we posed an open-ended question: How did the coronavirus situation shape participants’ Twitter use?

By Monday afternoon, we had agreed upon an English questionnaire, and, by Tuesday, had translated it into Hebrew and Arabic. By evening, the survey was up on the system. This left us Wednesday to detect and fix any major bugs and by the end of the day, we were happy with our product.

Still, as the survey links went out to the participants on Thursday afternoon, we were biting our nails: Would the survey run smoothly? And were participants willing to share their experiences and feelings about the circumstances? Fortunately, we encountered no serious problems with the questionnaire and our participants were happy to reflect on their situation. Response rates were high, probably reflecting both the boredom of being stuck at home and the fact that they had a lot to share about what was happening to them. Almost all participants also took the time to write about the experience of the COVID-19 crisis and how it impacted their Twitter use in response to the open-ended question. We felt we had salvaged the project as much as possible.

Differential effects: A first look at Twitter use during Covid-19

What can our accidental crisis communication data tell us about Twitter use in times of COVID-19? In a paper for the COVID-19 and Media special issue of the open-access journal Social Media + Society[2] we provide some first answers. Analysing the open-ended responses and items on Twitter use motives leads us to one primary conclusion: It’s complicated!

Many of our Collaborative Researcher Center colleagues have used this blog to discuss the presumed increased importance of digital communication media in a time when embodied encounters with all but members of the same household have been restricted.[3] While this motive was certainly present among our sample of high-intensity Twitter users, a highly differentiated overall picture emerged. While virtually all participants stayed home much more than usual – or completely, as some were in formal quarantine at the time of the survey – not all of them increased their Twitter use during this time. To the contrary, some reported being less active due to feeling overwhelmed by the influx of information or because they were too busy caring for children who were out of school and too preoccupied with their immediate situation. Others did not feel that their use intensity had changed much at all. The largest group, however, did report spending more time on the platform.

This pattern of differential, partly even opposite, effects also emerged when looking at the reasons why people turned to Twitter. Roughly equal numbers reported turning to Twitter more for receiving or sharing information as those who did so less than usual. The same was true for the social motive of connecting to others through the platform. The only motive that saw a very clear overall increase in importance was using Twitter to pass the time. This is not a motive that has received prior attention in crisis communication research,[4] but it points towards the particular nature of a pandemic. While finding oneself in the middle of a hurricane or earthquake requires coordination and action, spatial distancing as a countermeasure to COVID-19 is equally likely to entail boredom, loneliness, and the search for entertainment.

Our first dive into the data therefore suggests an intriguing picture, but no straightforward single answer as to how our participants used Twitter, a social media platform that was generally important in their pre-COVID-19 communication, during the crisis. One idea we emerge with, that we think may be applicable beyond social media use specifically, is the salience of differential effects of COVID-19. In contrast to natural disasters or violent conflict, which most people see as having a harrowing effect on their lives, the Coronavirus shaped the experiences of different people in various, sometimes contrasting, ways: For some it was a time of stress, for others of relief; for some a time to connect to others, for others a time to disengage due to a sense of overload; for some it meant a complete overhaul of life as they knew it while for others it changed very little in their everyday routine. Hashtags such as #InThisTogether might suggest that COVID-19 is a great equalizer. Instead, as has already been pointed out by our colleagues for the offline world,[5] also in the realm of networked communication on Twitter, we find that COVID-19 affects people in different situations very differently. In this sense, we might even speak of differentiating and disintegrating effects. 

It will be exciting to further examine these patterns as we draw on the rich MESM data and will be able to study also how the interactions of Twitter users were spatially structured during the pandemic. Did those who intensified their Twitter use reach out translocally to the world? Or were they, too, preoccupied with connecting to others in their close surroundings?  A deeper dive into the use situation data will allow us these insights, while we are still getting used to our unexpected role as inadvertent pandemic communication researchers… 


About the Authors

Daniela Stoltenberg (M.A. Free University of Berlin) is a researcher of communication at the University of Münster and a research associate in the CRC 1265 project B05 “Translocal Networks”. Her research interests include digital public spheres, communication in cities, and computational communication science.

Maya de Vries Kedem  (Ph.D. The Hebrew University of Jerusalem) is a lecturer of Communication at the Hebrew University of Jerusalem. Her research interests focus on political participation and digital activism in the context of the Israeli-Palestinian conflict. 

Hadas Gur-Ze’ev  (M.A. The Hebrew University of Jerusalem) is a Ph.D. student in Communication at the Hebrew University of Jerusalem. Her research interests include negotiations of gendered power relations in digital environments.

Neta Kligler-Vilenchik  (Ph.D. University of Southern California) is Assistant Professor of Communication at the Hebrew University of Jerusalem. Her research interests focus on political expression and participation in the changing media environment.  

Annie Waldherr  (Ph.D. Freie Universität Berlin) is Assistant Professor at the University of Münster. She co-leads the project B05 “Translocal Networks” in the CRC 1265.  Her research interests focus on the structures and dynamics of public spheres under the conditions of digitalization and datafication. 

Barbara Pfetsch  (Ph.D. University of Mannheim) is Professor of Communication Theory and Media Effects Research at the Freie Universität Berlin and Principal Investigator at the Weizenbaum-Institute for the Networked Society in Berlin. She co-leads the project B05 “Translocal Networks” in the CRC 1265. Her research interests focus on comparative political communication, digital issue networks, transnational and European public spheres.


[1] Karnowski, V., Kümpel, A., Leonhard, L., & Leiner, D. (2017): From incidental news exposure to news engagement. How perceptions of the news post and news use patterns influence engagement with news articles encountered on Facebook. “Computers in Human Behavior”, 76, 42-50. https://doi.org/10.1016/j.chb.2017.06.041

[2] Kligler-Vilenchik, N., Stoltenberg, D., de Vries Kedem, M., Gur-Ze’ev, H., Waldherr, A., & Pfetsch, B. (2020, forthcoming). Tweeting in the Time of Coronavirus: How Social Media Use and Academic Research Evolve during Times of Global Uncertainty. “Social Media + Society.”

[3] see: Blokland, T., Krüger, D., & Vief, R. (2020). “Just because we have to do it, doesn’t mean it is right.” https://sfb1265.de/en/blog/just-because-we-have-to-do-it-it-doesnt-mean-it-is-right/

Knoblauch, H. & Löw, M. (2020). “Dichotopia – The Refiguration of Spaces and the Security Society in Times of Corona Risk.” https://sfb1265.de/en/blog/dichotopia-the-refiguration-of-spaces-and-the-security-society-in-times-of-the-corona-risk/

Sayman, V. (2020). Corona und der öffentliche Raum. https://sfb1265.de/en/blog/corona-und-der-oeffentliche-raum/

[4] Takahashi, B., Tandoc Jr, E. C., & Carmichael, C. (2015). Communicating on Twitter during a disaster: An analysis of tweets during Typhoon Haiyan in the Philippines. “Computers in Human Behavior”, 50, 392-398.

[5] Blokland, T., Krüger, D., & Vief, R. (2020)