The COVID-19 Real World Worry Datasets

This repo contains the data and variable descriptions on the Real World Worry Datasets.

Papers on this dataset:

• [1]M. Mozes, I. van der Vegt, and B. Kleinberg, ‘A repeated-measures study on emotional responses after a year in the pandemic’, arXiv:2107.03466 [cs], Nov. 2021, Accessed: Nov. 24, 2021. [Online]. Available: http://arxiv.org/abs/2107.03466 (in press at Scientific Reports)
• [2]B. Kleinberg, I. van der Vegt, and M. Mozes, ‘Measuring Emotions in the COVID-19 Real World Worry Dataset’, presented at the ACL-NLP-COVID19 2020, Online, Jul. 2020. Accessed: Nov. 02, 2020. [Online]. Available: https://www.aclweb.org/anthology/2020.nlpcovid19-acl.11
• [3]I. van der Vegt and B. Kleinberg, ‘Women Worry About Family, Men About the Economy: Gender Differences in Emotional Responses to COVID-19’, in Social Informatics, vol. 12467, S. Aref, K. Bontcheva, M. Braghieri, F. Dignum, F. Giannotti, F. Grisolia, and D. Pedreschi, Eds. Cham: Springer, 2020, pp. 397–409. doi: 10.1007/978-3-030-60975-7_29.

Contact: bennett.kleinberg [at] tilburguniversity.edu

These projects are also available with on the World Pandemic Research Network:

• Measuring emotional responses to COVID-19 https://wprn.org/item/459652; Presentation on YouTube
• Gender Differences in Emotional Responses to COVID-19 https://wprn.org/item/547252; Presentation on YouTube
• Worry, coping and resignation - A repeated-measures study on emotional responses after a year in the pandemic https://wprn.org/item/545152; Presentation on YouTube

---

Repo structure

• main
  • data_phase_1
    • files
      • meta
      • texts
        • short (n=2500)
        • long (n=2500)
    • files_clean
      • liwc
      • meta
      • texts
        • short (n=2491)
        • long (n=2491)
  • data_repeated_measures
    • meta (n=1698)
    • texts
      • phase1
        • long
        • short
      • phase2
        • long
        • short

The directory ./data_phase_1/files contains all 5000 texts, while ./data_phase_1/files_clean contains those after the exclusion of 9 participants (as reported in the paper). Note: the LIWC data were only extracted on the "cleaned" data.

---

Variable codebook

Variable name	Type	Meaning
startdate	time stamp	Time of starting the task
enddate	time stamp	Time of submitting the task
duration	numeric	Duration in seconds
worry	numeric	9-point scale rating (1=low, 9=high)
chosen_emotion	string	Choice of emotion that best represented their feeling
anger	numeric	9-point scale rating (1=low, 9=high)
disgust	numeric	9-point scale rating (1=low, 9=high)
fear	numeric	9-point scale rating (1=low, 9=high)
anxiety	numeric	9-point scale rating (1=low, 9=high)
sadness	numeric	9-point scale rating (1=low, 9=high)
happiness	numeric	9-point scale rating (1=low, 9=high)
relaxation	numeric	9-point scale rating (1=low, 9=high)
desire	numeric	9-point scale rating (1=low, 9=high)
text_long	string	Text input on long text (500 chars. minumum)
text_short	string	Text input on Tweet-sized text (280 chars. maximum)
self_rating_general	numeric	9-point scale rating (1=low, 9=high) - how well they could express their feelings in text in general
self_rating_short	numeric	9-point scale rating (1=low, 9=high) - how well they could express their feelings in the short text
self_rating_long	numeric	9-point scale rating (1=low, 9=high) - how well they could express their feelings in the long text
twitter_general_often	numeric	9-point scale rating (1=rarely, 9=very often) - how often they are on Twitter
twitter_tweet_often	numeric	9-point scale rating (1=rarely, 9=very often) - how often they tweet on Twitter
twitter_participate_often	numeric	9-point scale rating (1=rarely, 9=very often) - how often they participate in conversations on Twitter
eng_native	string	Whether English is their native language (Yes, No, No but fluent)

Note: in the repeated measures version of the dataset, the variables from phase 1 (April 2020) are denoted as "[VARIABLE]_wave1" and those from phase 2 (April 2021) as "[VARIABLE]_wave2". The "delta" variables are the difference between the values from these two measurements (i.e. phase 2 - phase 1).