human movement inequality

human interaction

matters for well-being.  Interacting frequently, particularly with a diverse collection of others, makes us more tolerant, increases information flow, improves mental and physical health, and increases labor market participation.  

The benefits of diverse interaction develop in two ways.  First, social isolation is detrimental.  Those with less social interaction experience more stress, have lower health outcomes, and are more likely to be hurt or die during periods of significant social disruption.  Second, homophily, the tendency to associate with people who are similar to us, is also problematic as it reduces our access to resources, makes us less empathetic, and reduces generosity. 

We are our best selves when we interact with diverse others.

This project utilizes anonymous GPS data to develop measures of the frequency, depth, and diversity of modern interactions.  We have four primary goals: (1) describe the patterns of human interaction prior to the pandemic; (2) understand the inequality consequences of these patterns; (3) map and explore changes in human interaction during the pandemic; and (4) monitor pandemic recovery to understand the ‘new normal’.

Individuals of different incomes are not equally likely to come in close contact across all spaces.  For instance, the figure shows average pattern of close contact for San Francisco and New York City, for a weekday in March 2017.  Purple areas are segregated interaction, where individuals encounter others like themselves.  In contrast, yellow areas represent extreme income differences of $5,000 or more in monthly rent (or equivalent home value).

Our research shows that people at the very high and low ends of the income spectrum are the most likely to experience exposure segregation.  They have the least diverse interactions in residential neighborhoods, workplaces, and leisure sites.  This segregation comes about because:

  • People may actively seek out those of similar economic standing (the homophily effect).
  • Some people are more likely to come into contact with very heterogeneous visitors to their neighborhood or workplace (the visitor effect).
  • People may choose to travel outside of their neighborhood or workplace. Where people choose to go might bring them into contact with people like them, or people who are very different.

For example, the golf course is so segregated not only because people from similar socioeconomic status groups are more likely to choose that as their leisure (e.g., golf instead of bowling, shopping at Target instead of Nordstrom’s, etc.), but also because one’s own playing group is protected from meeting anyone outside that group (golf courses are a site of low density).  In contrast, when going to a stadium to watch a baseball game, even though people likely attend with friends, the stadium is densely populated and visitors cross paths with a large collection of others.  Thus, the type of leisure we select has a large impact on our experienced segregation. 

with, David Grusky, Jure Leskovec, Emma Pierson, Blanca Villanueva, Nic Fishman, Wenli Looi, and Hamed Nilforoshan.

For more information see the Human Movement Inequality Laboratory.

The benefits of diverse-human interaction are on stark display in the COVID world.  Before the pandemic, different racial and income groups tended to travel the same distances even if they were largely isolated from one another.  The pandemic changed this, and the resulting limitations of movement and mobility create differential impacts depending on the kind of jobs individuals hold, the neighborhoods they live in, and their access to essential services. 

The pandemic has the potential to further isolate the isolated.  It also has the potential to make our world smaller.  It decreases the distance we travel, the number of others we see, and the number of contexts we experience daily.  Resources differ substantially by neighborhood and as our worlds shrink, our local contexts matter more.  Thus, the long-term consequences of the pandemic, and our ability to rebound as a society, will be deeply intertwined with interaction patterns that develop during the pandemic.

With the rise of COVID, the human movement lab pivoted to assist with epidemiological modeling of spread, and estimations of mechanisms.  The methodological paper “Supporting COVID-19 policy response with large-scale mobility-based modeling” recently won an award from the Association for Computing Machinery

Mobility network models of COVID-19 explain inequities and inform reopening

Using anonymous cell phone data to map the hourly movements of 98 million people to places like restaurants, gyms, and churches, a team of Stanford and Northwestern researchers has created a computer model that accurately predicted the spread of COVID-19 in 10 of the largest U.S. cities this spring.

The study, published in the journal Nature, merges demographic data and epidemiological estimates with the cell phone location data. It appears to confirm that most COVID-19 transmissions occur at “super-spreader” sites, such as full-service restaurants, fitness centers, and cafes, where people remain in close quarters for extended periods. It also reveals why the coronavirus disproportionately affects low-income Americans of color.

In layering demographic data from a database of 57,000 census block groups—neighborhoods of 600–3,000 residents—into their model, the researchers show how minority and low-income people leave home more often because their jobs require it. They shop at smaller, more crowded establishments than people with higher incomes, who can work from home, use home delivery to shop, and patronize more spacious businesses.

For instance, the study revealed that it is roughly twice as risky for Black, Latino, and other underrepresented populations to buy groceries (compared with Whites).

“This study demonstrates that the way low-income neighborhoods are constructed, with smaller establishments that serve more customers, is one of the drivers of racial and economic inequality in infections,” Redbird said. “It also shows that, by reducing density in these locations, we might reduce this disparity.”

The model can be used to as a tool for officials to help minimize the spread of COVID-19 as they reopen businesses by revealing the tradeoffs between new infections and lost sales if establishments open, for example, at only half their normal capacity.

This research was supported by the National Science Foundation, the Stanford Data Science Initiative, the Wu Tsai Neurosciences Institute and the Chan Zuckerberg Biohub.

This research was Northwestern’s 14th highest reaching story in traditional media in 2020, earning more than 250 mentions for a reach of 35.4 million.

Top outlets include The Wall Street Journal, The New York Times, CNN, Bloomberg and CBS News, among many others.

Tools and data are publicly available on the COVID-19 Mobility Network Modeling site

For more information see the Human Movement Inequality Laboratory.

Segregation in movement also develops as people from racially segregated neighborhoods move through space in different ways.  The figure shows movement of Chicago residents from predominantly White, Black, and Latino neighborhoods during different times of the day.  This segregation reduces diverse interaction and has institutional consequences.

“I don’t think there is another city in the U.S. that has an extensive and integrated camera network as Chicago has.”  

-Michael Chertoff

former U.S. Homeland Security Secretary

By some estimates, the city of Chicago has between 32,000 and 60,000 cameras throughout the city.  Using FOIA requests, this project seeks to understand the placement, and privacy consequences of this network. 

Such cameras are significantly more likely to be placed in Black, Hispanic, and low-income neighborhoods, or in neighborhoods where people from these areas commonly pass though.

Also, by examining the movement of people within the video range of a camera, we can estimate how often individuals from various neighborhoods are monitored by the Chicago surveillance network.

with Andrew V.  Papachristos, Professor of Sociology, Director of the Northwestern Neighborhood & Network Initiative

For more information see the Human Movement Inequality Laboratory.

From decades of research, sociologists know a lot about human relationships.  We know about your friends, your co-workers, and your neighbors.  What we do not know is who you see every day or how well you know them.  In the course of your Saturday, you might spend five minutes with a neighbor and an hour with a friend, but what happened during the rest of the time?  When you exchanged money at the coffee shop or nodded at another parent at your child’s soccer game, you interacted with others, but we know virtually nothing about the collective mass of these interactions or how they shape your perception of the world.

Social network analysis was supposed to provide revolutionary insight into the cause and consequences of interactions, but massive data requirements limited its utility.  There is good news, though – the right tools exist to take apart these interactions and study them.  By capitalizing on the wealth of data sources inherent to modern technology, we can track the connections between people, not just relationally, but experientially.

The project, using a nationally-representative sample of 5,000 respondents, shows that high income Americans have the opportunity for more interactions.  They see people more often and have more frequent interactions with neighbors, friends, and family.  In contrast, low-income Americans have the least diverse interactions, and are more likely to see people in their same income category.  However, if we remove people who are serving or waiting on them, high income Americans have the least diverse interactions. 

This lack of diversity has striking consequences for information exchange.  For example, the average high-income American lists the minimum wage at nearly $40,000 a year, more than twice its actual rate.

For more information see the Human Movement Inequality Laboratory.