photo: A passenger in Brooklyn waits for a subway train.
A passenger in Brooklyn waits for a subway train. The transit system's role in the city's coronavirus infectious rate is now the subject of a political, but not scientific, debate. Holly Pickett/Bloomberg

A paper claims that the nation’s largest transit system made NYC a Covid-19 hot spot. But experts say there are too many unknowns to link ridership to infection rates.

On April 13, Jeffrey Harris, an economist at the Massachusetts Institute of Technology released a non-peer-reviewed study with a provocative title: “The Subways Seeded the Massive Coronavirus Epidemic in New York City.”

In the working paper now available at the National Bureau of Economic Research, Harris maps subway turnstile data against infection rate by zip code, and claims that the recent flattening of New York City’s epidemic curve is linked to the 65 percent decline in ridership that occurred in the first half of March. In an op-ed in the New York Daily News detailing the paper’s contents, he also points to the heavy death toll among MTA workers, which hit 79 on April 20, as another piece of evidence.

The study has been widely panned. Over the past week, mathematicians, infectious disease researchers, and transit policy experts have criticized Harris’ methods, warning that he fails to provide statistical evidence and ignores significant confounding factors. That hasn’t stopped its conclusions from now providing fodder for a political debate on social media, conservative talk radio, and among city policymakers about the appropriate public health response to coronavirus. On Sunday, a Queens politician cited the paper in a plea to New York Governor Andrew Cuomo to shut down all subway  service.

In this moment of profound uncertainty for public transit agencies, the dialogue around the paper shows how easily subways and buses can be scapegoated, even as they provide millions of essential trips during the pandemic.

To arrive at his deduction, Harris presents a series of thematic maps purporting to match busy train lines to infection rates, claiming that subway lines are the “correct unit of analysis” for studying the system’s role. The map that has been most widely shared online shows some of the stops along the 7 train — the Flushing Local Line — layered on top of a map of infection rates by zip code as it courses from Manhattan through the heart of Queens. He also charts ridership declines against the citywide epidemic curve.

Yet disease modeling experts say that the data visualizations that Harris includes do not appear to clearly support the correlation that he is trying to draw: For example, only 13 of the 22 stops shown on his map of the Flushing Local Line are located in zip codes with infection rates greater than 100 per 10,000 people.

Stops along the Flushing Local Line in the New York City subway system, superimposed on a section of infection rates by zip codes, with case data from April 8, 2020. The palest shade of green denotes areas with less than 70 cases per 10,000 people. The next darkest shade denotes areas with case numbers greater than 70 but less than 100 per 10,000 people. The darkest green denotes areas with more than 100 cases per 10,000 people. Harris writes, “the outer area of each point corresponds to the volume of turnstile entries during the first week in March 2020, while the inner area corresponds to the volume during the third week of that month.” (Jeffrey Harris/National Bureau of Economic Research)

The paper also does not attempt to disentangle the many confounding factors that could easily make the subway not the primary vector. With no statistical analysis to contextualize the significance of the numbers he presents, it’s impossible to say that ridership is more than a reflection of the other types of activity that declined in mid-March, after public schools and restaurants were closed and stay-at-home orders were issued.

“School closure, workplace closure and reduced community contacts have resulted in a reduction of transmissions,” said Philip Cooley, an emeritus fellow at the nonprofit research organization RTI International, who spent his career studying viral disease transmission through computational biology. “In other words, reduced subway ridership is a proxy for the increase in social distancing practices, and it is the culmination of all of those social distancing practices that have flattened the infection curve.”

Abbey Collins, the New York City MTA’s chief communications officer, called the paper “flawed — period.”

Transit researchers also warn that no such dramatic links between transit use and viral spread have been found elsewhere in the world. As one case in point, the city with the world’s longest subway system, Seoul, may also the among the very best at controlling Covid-19. The fact that the most recent rises in infections are happening in rural parts of the U.S. might also give credulous readers pause. Salim Furth, a senior research fellow in urban economics at George Mason University, dissected the paper on the blog Market Urbanism and determined that more attention ought to be paid to the infectious threat of car-based travel, considering (among other factors) that Staten Island, the least transit-reliant borough in New York City, now has the highest infection rate in the city.

In a withering blog post critiquing the paper, mathematician, transit analyst and CityLab contributor Alon Levy detailed several reasons that subway and track workers might have been affected worse than riders — “contamination at work is not the same as contamination during travel,” Levy wrote.

Harris maintains that the core premise of the argument is sound. “From the perspective of the highest standard of proof, this is a correlational study — there’s no way around that,” he said. “On the other hand, I can’t see how any serious practitioner of public health, having looked at facts, would recommended other than to focus sharply on the subway system as the fuse that lit the epidemic in New York City.”

But where the facts point, many infectious disease experts say, are to a few critical days in early March that daily life went on after the virus was established in the city of eight million, but before public health directives took effect. That lag is the leading explanation for why New York City wound up leading the world in coronavirus cases: well before subway ridership dropped off, “the damage was already done,” said Robyn Gershon, a clinical professor of epidemiology at New York University with a focus on occupational and environmental health and safety. “The virus was already spreading.”

That isn’t to say that germs did not spread in the subway. As one of many places in New York City where people crowd together, subway cars almost undoubtedly amplified the coronavirus pandemic, Gershon said. But to confidently measure the role that it played compared to taxis, schools, restaurants, workplaces and other shared spaces would take “a big, expensive study with a lot of grad students.” A researcher would likely need to swab a lot of subway poles, conduct a detailed survey about the commuting routes and habits of a representative sample of thousands of riders, and take antibody tests for all of them, similar to what Gershon did in a 2009 analysis of how subway noise impacts hearing loss among New Yorkers published in the American Journal of Public Health.

Gershon is gearing up for further survey-based research now: She says she hopes to partner with the Transit Workers Union to study how workplace conditions and labor practices could have led to the high infection rates and death toll from Covid-19 among MTA workers.

Alternatively, a scientist could build a simulation of virus transmission and run it on top of a synthetic population that reflects real commuter demographics over several weeks. That’s what Cooley did to study the subway’s role in the transmission of H1N1 in New York City in 2009, concluding that 4.4% of citywide infections were transmitted via the subway. If subway service had been suspended at at that time, he found, it could have reduced the total number of infections in NYC by 12.5%. Since coronavirus is far more contagious than H1N1, Cooley guesses that the subway infection rate for Covid-19 would be even greater.

However, the mysteries of coronavirus would present a major challenge to replicating that type of study, Cooley said. Crucially, with H1N1, Cooley knew the rates of virus immunity and reinfection among New Yorkers. Because coronavirus is so new, with no prior human immunity, an uncertain incubation period, and an unknown reinfection rate, a confident simulation of transmission on the scale of New York City might not be possible yet.

The Harris paper does prove at least one thing very clearly: how easily the subway can be blamed for a disaster of invisible origin, especially in a hotly partisan environment that pits cities against pretty much everywhere else. Public transit has long carried a social stigma in the auto-dominant United States, with low-income people of color and other marginalized groups making up an outsized share of riders. New York City was already a national outlier in many respects, including in terms of transit use. Now, with the help of the MIT paper, its famed subway has become ammunition in the mounting coronavirus culture war. “It is rather odd to see NY journos losing their shit about FL beaches when the NYC subway is still running,” Kyle Smith, a critic-at-large at the National Review, tweeted on Saturday. “Seems obvious that the subway was a leading means of spreading the disease, possibly the single most important means in the entire country.”

There are also now voices citing the paper in calls for a transit shutdown, including four New York City council members from outer-borough districts. They signed a letter to Governor Andrew Cuomo calling for a temporary closure of the MTA’s trains and buses, with riders diverted into for-hire vehicles instead, which Harris has also proposed. That seems unlikely to happen: Under normal conditions, nearly 60% of New York City's 4.2 million workers rely on buses and trains, and although ridership has plunged 90% from normal levels, the subway system still supplied some 2.5 million rides in the first week of April.

In response to the council members, the MTA defended its operations as essential transportation for “doctors, nurses, first responders, grocery and pharmacy workers, and other essential personnel to get to work and save lives.”

New York City leaders will soon face a reckoning for the belated response to the coronavirus, and for the transit workers who are dying in such alarming numbers. Before the pandemic, the MTA already had its work cut out to improve service delays and overcrowded trains, and it’s not clear how the agency will now move forward, facing an $8.5 billion shortfall and a prolonged need for heightened sanitation and health protocols. But in a city as large and densely packed as New York, transit makes the economy function; while some commuters with means may choose to buy cars when lockdowns lift, many workers and students will inevitably return to the rails. Said Gershon: “We can never re-open without the subway.”

About the Author

Most Popular

  1. A photo of a police officer in El Paso, Texas.
    Equity

    What New Research Says About Race and Police Shootings

    Two new studies have revived the long-running debate over how police respond to white criminal suspects versus African Americans.

  2. Equity

    The Problem With Research on Racial Bias and Police Shootings

    Despite new research on police brutality, we still have no idea whether violence toward African Americans is fueled by racial prejudice. That has consequences.

  3. A map of population density in Tokyo, circa 1926.
    Maps

    How to Detect the Distortions of Maps

    All maps have biases. A new online exhibit explores the history of map distortions, from intentional propaganda to basic data literacy.

  4. A map of Minneapolis from the late 19th century.
    Maps

    When Minneapolis Segregated

    In the early 1900s, racial housing covenants in the Minnesota city blocked home sales to minorities, establishing patterns of inequality that persist today.

  5. Maps

    Your Maps of Life Under Lockdown

    Stressful commutes, unexpected routines, and emergent wildlife appear in your homemade maps of life during the coronavirus pandemic.

×