Authors: Esben Kran & Jonathan Hvithamar Rystrøm

Started: March 2021

Data analysis project on a whole Cognitive Science student year before and during the covid-19 pandemic lockdown in Denmark.

During 1.5 years, 28 students communicated intensely on the social messaging app Facebook as part of the cognitive science bachelor’s program at Aarhus University. These students voluntarily gave us their highly anonymized data (see “Anonymization”). In a post-experimental questionnaire, 100% reported that it was their main communication tool in the study group and 87,5% reported that it was their main communication tool with everyone from the study.

The anonymized data contains a message per row between two of the 27 users (28 minus 1 dropout). The weight of a message is 1 for a direct message, while a group message is weighted as 1/(n-1) where n is the amount of users in the group (the sender inclusive).

The information present for each message is the date, time and the anonymized names of the receiver(s) and the sender. The participants downloaded their private messaging data (1-3GB) from Facebook in the period between the 1st of August 2020 and the 1st of January 2021. They personally ran a script that anonymized and extracted the relevant data to create a first-step encrypted dataset using the same hash key of their messages (2-10MB). These were sent to the authors that performed a second-step anonymization via another hash key. Both hash keys were subsequently deleted.

Released under MIT by @rysias and @esbenkc.