Staying at home
Modelling prosocial behaviour and its impact on COVID-19 spread
Jana Lasser | jana.lasser@tugraz.at | @janalasser
slides available at
https://janalasser.at/talks/staying_at_home/
Motivation
The influence of human behaviour on infectious disease spreading remains an underexplored topic:
Focus on vaccine uptake, not infection prevention (see for example [1]).
Little empirical evidence for assumed functional forms of behaviour [2].
(1) Why to people adhere to non-pharmaceutical intervention measures?
(2) How does the level of adherence change over time?
[1] Bauch & Earn: Vaccination and the theory of games, PNAS (2004)
[2] Funk, Salanthé & Jansen: Modelling the influence of human behaviour on the spread of infectious diseases: a review, Journal of the Royal Society Interface (2004)
Outline
Using prosociality as moderator for measure adherence
What is prosociality?
Prosociality relates to measure adherence
Measuring prosiciality
From concept to construct
Measure validity
Testing the causal relation to COVID-19 case numbers
Modelling prosociality dynamics
Collective emotions and feedback processes
The interplay between physical distancing and prosociality
Using prosociality as moderator for measure adherence
What is prosociality?
People have "prosocial motivations" when they have any motivation to promote the welfare of others.
[1] Zaki & Mitchell: Equitable decision making is associated with neural markers of intrinsic value, PNAS (2011)
[2] Fehr & Fischbacher: The nature of human altruism, Nature (2003)
[3] Barclay: Trustworthiness and competitive altruism can also solve the “tragedy of the commons”, Evolution and Human Behavior (2004)
[4] Nowak & Sigmund: Evolution of indirect reciprocity, Nature (2005)
Prosociality relates to measure adherence
People with a record of prosocial behavior have greater prevention intentions [5].
Prosocial motivation for social distancing is linked to health behavior adherence [6].
Prosociality is at least as strong a motivation as risk awareness. Motivations for social distancing [7]:
"I want to protect others" (86%)
"I feel a sense of responsibility for my community" (84%)
"I want to protect myself" (84%)
[5] Jordan, Yoely & Rand: Don’t get it or don’t spread it: comparing self-interested versus prosocial motivations for COVID-19 prevention behaviors, Scientific reports (2021)
[6] Nelson-Coffey et al.: Health behavior adherence and emotional adjustment during the COVID-19 pandemic in a US nationally representative sample: The roles of prosocial motivation and gratitude, Social Science & Medicine (2021)
[7] Coroiu et al.: Barriers and facilitators of adherence to social distancing recommendations during COVID-19 among a large international sample of adults, PLOS ONE (2020)
Measuring prosociality
From concept to construct
Operationalize "prosociality" using a list of words related to the concept constructed by domain experts [8].
Examples: aid, help, support, volunteer.
Extension of the LIWC (Linguistic Inquiry and Word Count) method [9].
Measure the prevalence of prosocial words in the public discourse (social media → Twitter) to approximate motives to help others in a population.
[8] Frimer, Schaefer & Oakes: Moral Actor, Selfish Agent, Journal of Personality and Social Psychology (2014)
[9] Pennebaker et al.: The Development and Psychometric Properties of LIWC2015 (2015)
Internal measure validity
LIWC's general performance is similar to other SOTA sentiment analysis tools [10].
[10] Ribeiro et al.: SentiBench - a benchmark comparison of state-of-the-practice sentiment analysis methods, Journal of Personality and Social Psychology (2016)
Internal measure validity
There is evidence that prosocial word density predicts motives to help others in individuals [8].
Twitter accounts of NGOs use prosocial words about three times as often as other accounts [11].
Planned validation study: large-scale survey on prolific
Prosocial Behavioural Intentions Scale [12]
+ Twitter handle to collect text and validate dictionary
+ Measure adherence intentions
[8] Frimer, Schaefer & Oakes: Moral Actor, Selfish Agent, Journal of Personality and Social Psychology (2014)
[11] SI to Garica & Rimé: Collective Emotions and Social Resilience in the Digital Traces After a Terrorist Attac, Psychological Science (2019)
[12] Baumsteiger & Siegel: Measuring prosociality: The development of a prosocial behavioral intentions scale, J. Pers. Assess. (2019)
External measure validity
... but Twitter isn't representative!
Garcia et al.: Social media emotion macroscopes reflect emotional experiences in society at large, arXiv. (2021)
Why digital traces are better than surveys
Testing causal relations of COVID-19 spread I
Granger causality: how predictive is one time series of another? [8]
[8] Granger: Investigating Causal Relations by Econometric Models and Cross-spectral Methods, Econometrica (1969)
Testing causal relations of COVID-19 spread II
Testing causal relations of COVID-19 spread II
Testing causal relations of COVID-19 spread II
Modelling prosociality dynamics
Which mechanisms could influence the prevalence of collective prosociality?
Positive feedback
Collective effervescence: A community or society may at times come together and simultaneously communicate the same thought and participate in the same action. Such an event then causes collective effervescence which excites individuals and serves to unify the group. (Durkheim 1912, Garcia & Rimé 2019)
Inhibition
Measures – especially physical distancing – hinder social exchange and inhibit effervescence.
Decay
It is exhausting to maintain a state of high emotional arousal over time.
Memory?
Is there such a thing as "measure fatigue"? (Abbasi 2020).
External driving (?)
Feedback between prosociality and severity of the pandemic (reported cases, ICU occupancy)
    → could be directly coupled in an extended SIR model
Feedback between prosociality and measure stringency
    → could be directly coupled in an extended SIR model
Feedback between prosociality and intensity of news reporting?
    → hard to measure and model
Simplest functional form
Competition between effervescence and prosociality decay produces oscillations in adherence independent of external circumstances.
What's next?
Measure prosociality using Twitter data
In individual states/regions in the US and/or Germany
Test causal relationship between prosociality and case numbers
Get inspiration for the functional form of prosociality dynamics from the empirical data
Couple prosociality to infection dynamics in an extended SIR model