Social Media and Digital Outreach │ Data Science

Harnessing social media for service delivery and to enhance social and behavioral research

Social media and digital outreach offer new opportunities for data collection and for improving service delivery. With a tap on a phone or a tablet, users can provide key information and help agencies better understand how to partner with individuals to improve social outcomes. However, working in this new realm creates significant challenges, such as determining how to measure changes in behavior and how to expand the reach of campaigns to reach different demographic groups equally. AIR’s digital technologists and social media researchers work hand in hand with our clients to address these issues.

Contact
Image of Amber Bloomfield

Amber Bloomfield

Senior Data Scientist

Featured AIR Projects

Black Americans and COVID-19: Social Listening Using Twitter

AIR used machine learning and text analysis to support research into how Black Americans view the COVID-19 vaccine. We used social listening methods to analyze Twitter data. The team used hashtags and keywords to identify Twitter users posting in the “Black Twitter” conversation space and searched their tweets for discussions about the COVID-19 vaccine. Using distilbert, the team created an advanced language representation model trained to identify tweets in support of and in opposition to the COVID-19 vaccine. The team then used topic modeling and K-means cluster analysis to identify pervasive arguments among the tweets. The results of these analyses will inform communication strategies to address COVID-19 vaccine distrust and stigma among Black Americans.

USAID Haiti Alliance for the Protection of Children Project for USAID

In Haiti, AIR examined and analyzed popular broadcast media (e.g., radio) and Facebook and Twitter to identify individuals who were influencing or following conversations on violence. AIR translated unstructured conversations into a common data scheme with tags describing violent events and sentiments. AIR used the structured data to create visualizations of social networks, time series displays, charts, and graphs to inform Haitian policy analysts and other stakeholders who design programming.

CMS/SIE Social Network Analysis and Monitoring

The Centers for Medicare & Medicaid Services (CMS) established the Strategic Innovation Engine (SIE) to rapidly move innovative, evidence-based quality innovation (QI) practices from research to implementation across the health care community. Through the SIE, CMS wanted to investigate the extent to which social network analysis and monitoring of Twitter could be harnessed to achieve two goals:

  1. Monitor social media more effectively to identify the newest and most important health and patient safety information that CMS and others should incorporate into quality improvement (QI) initiatives.
  2. Develop best practices for disseminating QI initiatives through social media.

Our team automated a process to extract tweets based on a set of keywords. We then mapped the social network channels resulting from the message and analyzed the characteristics and activity of the networks. We focused on tweets from 41 FDA patient safety announcements on medical device issues and 28 published articles on a variety of topics from the journal, Health Affairs. The analysis focused on identifying the extent to which conversations related to the article’s topic preceded its publication and the extent to which Twitter disseminated awareness of the announcement.