Yesterday, an article by Molly Boigon was published in the Forward, with maps and data on the impact the coronavirus had on Haredi and Hasidic communities in NYC. Unfortunately, Ms. Boigon missed some very important details. These details completely change the way we should look at these data.
Boigon’s method for defining which Zipcode Tabulation Areas (ZCTA) are Haredi/Hasidic, was by looking at the percentage of Yiddish speaking people in a ZCTA.
We defined Haredi neighborhoods using census data that show the percentage of people in each tract that speak Yiddish. This is not a perfect measure, but it led us to categorize three New York neighborhoods as Haredi: Williamsburg, Borough Park and Crown Heights. Together, they encompass 12 ZIP codes.
This is wrong, and ultimately leads to wrong conclusions. To illustrate why, I superimposed the maps of the eruvs in those three neighborhoods. To be clear, using eruvs as a measure of where haredim live is not perfect either. But they underline two big flaws in Boigon’s method. Eruvs are built specifically to accommadate Orthodox Jews, and its boundaries tightly hug the areas in which Orthodox people live. This is especially true in a place like Williamsburg where almost everyone living within the eruv boundaries is haredi (except for some public housing projects here and there). If you look at the map, you can see how much of Williasmburg and Crown Heights are simply missing from the analysis. But more importantly, you can see how there’s almost no ZCTA that can accurately represent the haredi statistics. In fact, huge parts of what Boigon considers to be the hardei areas in brookly are heavily populated by other communities (such as the African-Americans who live in Bedford Styeuvsant between Williasmburg and Crown Heights).
Interestingly enough, the only ZCTA which seems to be mostly haredi is 11219. Indeed, 11219 has an higher than average case rate and death rate.
There’s an expression in the world of data science: “Garbage in, Garbage out.” Using a wrong metric for where Jews live, leads to a wrong conclusion:
This suggests that the much-reported incidents of large funerals or weddings that disregarded social-distancing, and of underground schools and synagogues continuing to gather in person, may have been relatively isolated incidents — or, that similar violations of the guidelines were prevalent in other neighborhoods but garnered less attention.
There’s a further issue with the data used in Ms. Boigon’s analysis, on which she touches only lightly but it cuts to a very important issue. Looking at death rates simply by population is very misleading for two reasons. The first one, as Ms. Boigon rightly points out, is that hasidic communities tend to have a very young population. The second issue is that we have a smaller than average older population due to the Holocaust. Almost no Hasidic Jew living in Williasmburg or Boro Park today was born in the early 1940s and the late 1930s. To draw any conclusions from the death rate, we need to look at the death rate among the elderly population. An analysis on the Yiddish forum Kave Shtiebel, shows how Kiryas Joel, where the 60+ population was at an estimated 702 in 2018, there were between 10 and 15 deaths in that age group, a rate many times higher than the state’s average.
Boigon is absolutely right with her conclusions. We need more data to reach any meaningful conclusions. Unfortunately, she did jump to conclusions which are already being cited authorativley and uncritically.