A new study from the Steinhardt Social Research Institute says there could be over seven million Jews. Why are there so many different numbers, and why are Jews so hard to count.

One of the biggest reasons is that it’s hard to determine just who is a Jew. People identify as Jewish on religious, ethnic, and cultural grounds, while others identify by birth, and still others were clearly born to Jews but do not personally identify as Jewish. This makes picking out who is Jewish in a survey extremely difficult. You have to ask the right questions, the right ways, and then figure out which answers mean someone is a Jew. Changes in how you define a Jew can cause swings of literally millions of people.

But once you come up with a definition for who’s Jewish and who isn’t, you’ve still got trouble counting people. That’s because no matter how you define it, there just aren’t that many Jews in the population. Think of a jar with 1000 marbles in it. Of those, somewhere between 975 and 985 are yellow, while the rest are green. The green marbles, between 15 and 25, represent the size of the Jewish population. Now imagine that you have to count them by taking a small sample, say 100 marbles, out of the jar. If the jar were half green and half yellow, this would work pretty well, but with only a small number of green marbles, your sample of 100 could give you one, two, or three, each of which would give you radically different estimate of the number of green marbles in the whole jar. Counting Jews works the same way - the statistical sampling methods used in normal social science break down when trying to reach a very broad but very small group.

Further, Jews are less likely than the population at large to respond to random digit dial phone surveys. In other words, when you try to take the marbles out of the jar and count them, some marbles won’t let you, and the green marbles are less likely to let you. Given the huge swings a few marbles can give you, this makes them even more difficult to count. Because Jews won’t talk to the survey interviewers, we don’t really know how much less likely they are to answer the phone, so we don’t really know how to weight things. We also don’t know the characteristics that make one person more likely to answer than another.

Each of these caveats makes counting Jews extremely difficult, and leads to the controversy around the numbers given by Jewish demographers. So what are the different techniques that are used? Random digit dial surveys, like the massive National Jewish Population Survey use a weighting technique, similar to what political pollsters use, to try and bring the numbers in line with what they know to be true about the population as a whole. These can be controversial, whether for counting Jews or voters, as they start to tinker with the underlying science of public opinion research in ways that not all social scientists or statisticians agree. The NJPS also needs an extremely large sample to cover the whole country, and this means it’s prohibitively expensive to conduct accurately.

Another technique, used by Dr. Ira Sheskin at the University of Miami, takes a collection of community studies and combines them all together to get a national account. Since most Jews live in urban areas, Dr. Sheskin’s community studies, of which he has done many, are more likely to target those numbers where Jews are concentrated. He also over-samples in “core” Jewish areas, increasing his response rate. However, this method has its own shortfalls. For instance, many elderly Jews are “snow birds,” with summer homes in the Midwest or Northeast and winter residences in southern areas like Florida and Arizona, making it likely that they’ll be double counted. The same thing happens with younger Jews who may go to college in one city while claiming residence in another. And finally, it ignores the “long tail” of Jewish communities, smaller cities that still have Jewish populations.

Researchers at Brandeis’s Steinhardt Social Research Institute used a third method. They took publicly available survey data, rather than data specific studies on the Jewish community, and conducted a “meta-analysis,” trying to merge it all together to come up with a good accounting. In other words, instead of taking 100 marbles out of the jar, they found twenty people who had each taken 100 marbles out of the jar for different reasons, and used their counting. Not only does this give them more samples, but it also allowed them to save money for expensive polls with large samples that are required for the other two techniques. Unfortunately, there are a number of risks here. Different studies use different samples, questions, methodology, making it complex to determine how to weight any specific data point, and restricting the available data to what those other studies are interested in. Nevertheless, it provides an innovative, and significantly more cost effective, technique for gathering data not only about how many Jews there are, but information about them as well.

So which of these methods works the best, and how many Jews are there? Because of the problems described above, there’s no way to really know. The best answer is that it’s somewhere between four and eight million, and that whatever number you pick is probably off by a million. And like with most things Jewish, there’s going to be a healthy debate around the issue so that any two Jewish demographers will probably give you three different numbers.

Categories: America, Community, Information Asymmetries, Jewishness, Metrics, Polling