Earlier this week, Gothamist shared an interactive scatterplot graph showing pronounced associations behind high rates of COVID-19 cases and neighborhood median income, ethnic diversity, and age.
Today, we’ve added another chart, showing a positive association between a zipcode’s average number of people in a household and infection rate:
(See a larger version of the scatterplot here.)
When we first got the zip code data from the city’s Department of Health, we tried to find a correlation between each neighborhood’s density per square mile and infection rate, but we didn’t find one- many of the densest zip codes are in Manhattan, where there are low infection rates in most neighborhoods.
Then we had the idea to look at household size as a better indicator of density, because it made sense to us that infections would pass more quickly in neighborhoods where families were living in more crowded conditions. When we analyzed the data that way, we found one of the highest associations we’ve seen so far—higher even than the one we found for neighborhood median income.
Andrew Beveridge, a sociologist professor at Queens College, agreed with Gothamist’s analysis.
“If you had a situation where there are a lot of kids and people are just piled on one another, of course you would have more transmission because you would have to basically be around these people if they’re living with you,” said Beveridge, adding that the theory in isolating someone in a cramped room inside a tight conditions is “not very easy to do.”
That position is supported by a new study of coronavirus spread in Iceland showing that the top means of transmission in the later stage was within families.
Beveridge added that the low median incomes coming out of hard-hit Elmhurst, Queens correlate with greater household density, with many residents pooling their resources to cover the rent.
Zach Gottehrer-Cohen / WNYC
(See a larger interactive version of our map here.)
Dwayne Baker, an urban studies professor for Queens College, called the data “incredibly interesting" and concluded that smaller households sizes have an advantage in following social distancing guidelines.
“We’re social distancing, and if you’re the only person who lives in that household, you’re not interacting with people, right?" Baker said. "And the chances of you going out and talking with others are really slim. But if you’re in a house of four or three people that are living in there, that’s three times as much social interaction and multiplied over and over."
Annetta Seecharran, the executive director at Chhaya, a group based in Jackson Heights that works with South Asians, said that family size is a distinguishing feature of the Bangladeshi community.
According to 2015 census data, Bangladeshi New Yorkers average more than four people per household. They are also more likely to be in overcrowded housing than other city households in general, with 42 percent of Bangladeshi households compared to 9 percent of households in general having more than one occupant per room.
Not surprisingly, she has heard stories of Bangladeshis in central Queens that have been sickened or hospitalized by the virus.
Seecharran, who is Guyanese, said the same was true for Guyanese New Yorkers. Within her own extended family, four members in their 60s and 70s were hospitalized. One relative died.
Organizers like herself have had to rely on anecdotal data collected from their various networks. She expressed a frustration that the city was not providing more granular ethnic data.
“We need the data now to hear where the greatest need is,” she said.
While household density clearly shows a correlation, Baker theorized that it’s also likely that residents in specific employment sectors that require physical contact with strangers on a daily basis, specifically taxi drivers and delivery workers, make you more prone to catching the disease.
“All those things could be a factor that creates pockets of social vulnerability,” said Baker.
Zach Gottehrer-Cohen of WNYC contributed to this report.
Note 4/17: we've updated this post to be more precise with statistical terms including "correlation" and "association."
