Social media sparks conversation, breaks news, and brings people together. It can also inform research in new and innovative ways.
We recently used data from Twitter to explore how spatial movements and friend networks align with existing patterns of residential segregation in metropolitan Chicago.
Previous Urban Institute research demonstrated that high levels of racial and economic segregation in the Chicago region were associated with higher homicide rates and lower levels of four-year college degree attainment for black and white residents and with lower median income for black residents.
In this new study, we wanted to understand whether residents’ social media use reflected or overcame those segregated patterns. Within our sample of tweets, the accounts people followed and the accounts that followed them had little overlap with people in demographically similar neighborhoods.
This brings up several interesting questions we hope to explore in subsequent research:
- Does social media break the bounds of place and afford people viewpoints from outside the neighborhoods they live in or visit?
- Does social media encourage other forms of digital segregation, such as ideological differences and access to information, not reflected in their friend networks?
Combining novel private-sector data with existing administrative data can help researchers and policymakers explore these questions and gain greater insight into critical issues like segregation.
The promise of private-sector data
Private-sector data are becoming increasingly vital to the study of important public policy issues. Government agencies and research groups investigate these data because they have the potential to provide more timely and detailed insights, which can aid in answering big questions at potentially lower costs.
Federal agencies like the Bureau of Labor Statistics and the US Department of Transportation are using retail scanner data, smartphones, and traffic sensors to improve the accuracy and timeliness of government statistics and traffic management.
Internationally, research on satellite imagery and mobile phone data paired with small-scale surveys are being used to create more reliable, real-time estimates of population data at dramatically lower costs.
But analyzing private-sector data also comes with challenges. In our exploratory analysis of data from Twitter, we discovered issues of both bias and accuracy. We found that white, higher-income, and more highly educated areas in Chicago were more likely to produce tweets in our sample and that many users registered check-ins for the City of Chicago in general, not a specific location within the city.
Private-sector data have the potential to provide more timely, detailed, and cost-effective insights. And as the use of such data to inform policymaking grows, we should ensure it is paired with rigorous investigation and understanding of the underlying data.
The authors of this post have been updated to include Solomon Greene and Joan Wang.