CoVaxxy

Presentation by: Matthew R. DeVerna

What's in store for today...

Checkout the CoVaxxy dashboard

Cover some preliminary results

Talk about the CoVaxxy data

So lets check out the dashboard!

https://osome.iu.edu/tools/covaxxy

About the project

Started a COVID19/vaccines Twitter stream in early January
Details can be found here

Data set paper + data set
Collaborators
VaccinItaly

General

Quick jump to visualizations
More info buttons

Geography

Two data maps
Geo-locate Twitter users based on self-report
Vaccine data is reported at the state-level

Correlations

Same data as above, in a scatter plot
When we hover over dots...
Alabama, Texas, Arkansas, Kentucky

Hashtags

covidiots on the same level as pfiser & moderna
billgates
greatreset
covid19 for transparency/context

Vaccine uptake

7-day smoothed total differences --> due to population size!
Percent distributed is a fairere comparision
Georgia is a bit lower --> lets check out their level of vaccine hesitancy
Alabama, Texas, Arkansas, Kentucky
Many other options to select from...

Sources

Compare low-credibility and MAINSTREAM
When we hover over dots...

Most shared sources

Leaderboard
Can see low-credibility sites that were popular in the past week

What is the relationship between misinformation and vaccine hesitancy?

Twitter

CDC

FB Survey

Twitter +
CDC +
FB Survey

Combine different data sets

State-level vaccine uptake

Misinformation negatively correlates with vaccinations on the ground

race

age

politics

poverty

rurality

State-level vaccine hesitancy

Misinformation positively correlates with vaccine hesitancy

County-level vaccine hesitancy

Misinformation positively correlates with vaccine hesitancy

Available Data

CoVaxxy Data

Real-time data collection

- Constantly growing

- 120M+ tweets

- Two virtual machines ensure we have no data outages

Employ a snowball sampling technique to create a list of over 70 keywords related to covid-19 and vaccines
There are many, however, over 95% of our tweets so far include the first three words highlighted in red
We then set up a real=time Twitter stream that downloads any tweet containing any of these keywords - beginning on January 4th
This means we have:

Constanty growing
120M+ tweets
To ensure no gaps in data, we’ve set up two virtual machines in different geographical locations that collect the exact same data

Thank you!

- Twitter: @mdeverna2

- Email: mdeverna@iu.edu

- Site: matthewdeverna.com

CoVaxxy dashboard

Data collection paper

Vaccine hesitancy preprint