How useful was social science during the pandemic?
Plan
The pandemic was useful for social science, but how useful was social science for the pandemic?
A look at (a small!) selection of work in political science and economics on:
- Predicting
- Tracking
- Solving
- Critiquing
Was this useful? Where could we have done better?
General remarks
Rapid responses, broad responses
- Oxford policy data tracker
- Tracking dashboards
- Pooled surveys: RECOVR, Y-RISE
- Special issues (e.g. POP) and funding opportunities (e.g. IPA, Mercury)
“Staying in lane” issues fairly well negotiated
Predicting: Where will the Corona burden lie?
- Lots of epi models: does social science have value added?
- Social scientists are particularly good at predicting regular events after the fact
- Real challenge to predict outside the routine
- Is it too much to ask?
Predicting
Bosancianu et al 2023: lots of available theory (in all directions)
Predicting
Capacity arguments mixed: astonishing unconditional relations, capacity correlates early on, trust later on
Predicting
Not much traction from political variables
Predicting
Trust and inequality stand out
Implied learning
Expert predictions (generally very diverse) and implied updating
Model challenge
Over 100 entries across challenges: Golden, Slough, and Zhai (2023)
Model challenge
Social scientists better than chance but only in the tails
Model challenge
Trust and inequality again
Summary
- Not clear much demand for social science predictions
- Quite modest traction: but still, value added over epidemiological and machine learning models
- Purely political arguments did not do a lot; theoretical logics thin
- Trust and inequality consistently important
Tracking
A lot of social science is focused on measurement.
Did social science effectively help track the situation?
Dashboards
Sierra Leone dashboard making use of existing sample and team (esp Voors and Mereggi in country): resources
Dashboards
Drawn on by MoF
Dashboards
Similar dashboards in Uganda and elsewhere
Vaccine acceptance
Despite bad press: broad willingness to use vaccines. Solı́s Arce et al. (2021)
Many many more
Many observational papers identifying who takes what kind of action
- e.g. low trust Goldstein and Wiedemann (2022)
- POP special issue
Summary
- Social scientists stepped up to address measurement issues
- Made use of existing infrastructures
- But also flexible
- Engaged with policy in developing areas
Solving
Are social science contributions helpful for formulating policy?
- Survey work
- Nudge work
- Field work
Surveys
Incentives to take up vaccine, Klüver et al. (2021)
Surveys
Attitudes to restrictions (Germany), Hartmann et al (2023)
Surveys
Broad (non-strategic) support for vaccine sharing (Germany), Geissler et al. (2022)
Nudges
A focus for many:
Mixed messages on messaging: Dai et al. (2021)
But failures to replicate (Rabb et al. 2022)
Nudges
Regret lotteries: positive news at first, but no replication Milkman et al. (2022)
Field: Masks
Perhaps the largest masking study done by economists (Abaluck et al. 2022; Jefferson et al. 2023); WHO change in policy. Wild heterogeneity across studies.
Field: Last mile
Key study of vaccine delivery Mobarak et al. (2022)
Taking costing seriously
Summary
- Some weaknesses: survey dependence, relying on easy fixes
but
Social science methods performing well (theory contributions less obvious)
Striking agility of some (often US based) social science researchers: access to data, willing to move quickly at scale, “shoe leather” research.
Critiquing
Some points of intersection some points of tension seen in lane crossings
Critiquing
- Social science reading of health analyses suggests scope for gains
- Statistical interpretations
- \(p\) values
- Weak cost - benefit analysis
- Interpretation of nulls
- Baseline balance
- Selection biases
- More attention to contextual heterogeneity: coordinated trials
- Weak between community connections (Sunstein et al. 2022; Lees et al. 2021)
Critiquing
Null interpretations, experimental design, theory
Future
Implications for the next time.
Lessons
Infrastructure
- Reliance on spotty coverage from project specific surveys: nothing like a global barometer. Can we imagine this?
Networks
- Strengthen networks, share methods
- Need: Consumer (government) side analysis
Questions
- Health consequences of social divisions
- Optimal responses model taking account of economic and political costs
Social costs
Huge costs cross-nationally from policies: Egger et al. (2021)