Well, I promised myself I wouldn’t do it again, but here I am, caught in another round of mud-wrestling over the minutiae of Jewish population statistics. You’re in for a treat.
If you’re just joining our broadcast, here’s a quick recap: Pew Research Center published a demographic “Portrait of Jewish Americans” on October 1 that was widely interpreted as depicting a community in rapid decline. I wrote a column that went on line October 13, arguing that one of Pew’s most alarming assertions, a dramatic increase in Jews “having no religion,” was mistaken. Pew came back with a rebuttal that suggested I didn’t know what I was talking about. One old friend told me via Facebook that I’d been “eviscerated.”
Frankly, I figured I’d leave it there. I’d made my case. I didn’t want to get into a back-and-forth that could get personal. Overall, I found the Pew survey important and enlightening. My online version might have inflamed things by unintentionally implying that the whole survey was off base. If the Pew folks overreacted to that, I sympathized. But then I heard from friends who wanted to know if I was conceding that I was wrong, and if not, what was my reply? So here goes.
For starters, I had written that Pew got a false picture of current trends by comparing its count of Jews of No Religion, 22% of all Jews, with National Jewish Population Survey 2000-01, which found 7%. I wrote that NJPS 2000-01 was notoriously unreliable, and Pew should have looked back to the better-received NJPS 1990. As it happens, the 1990 survey found 20% Jews of No Religion. That’s statistically the same as 22%. In other words, I wrote, there has been no increase.
Pew replies that drawing comparisons among the three surveys “without further analysis” is “highly misleading” because they use different methods, ask different questions and so on. “We make comparisons only sparingly, always with appropriately cautious language,” Pew says.
But the Pew report opens in its very first two paragraphs by describing trends that rely on comparisons with earlier surveys: the increase in Jews of No Religion (increased compared to what?) and a decrease in Jews-By-Religion as a share of the overall American population. Pew’s most quoted trend — the increase in No Religion — is based mainly on the 2000-01 NJPS. The appropriate cautions come several chapters later. They’re now offering other, non-NJPS evidence. That doesn’t work either, as we’ll see.
As for Pew’s cautionary note about the pitfalls of drawing those comparisons, it appears in a “sidebar” sandwiched in between Chapters 4 and 5. The impression given out — indeed, the impression that virtually every reporter picked up and passed along to the public — was that the explosive growth of No Religion was not a cautious estimate but solid fact.
And what makes the 1990 survey, which I invoked, a worse basis for comparison? Pew says in its rebuttal that the 1990 survey, unlike 2000-01 and Pew itself, never asked respondents “do you consider yourself Jewish?” — that is, “not asked in the same way, at least not of all respondents and about all Jewish adults in surveyed households.”
Not asked “in the same way”? “Not of all respondents”? That raises more questions than it answers.
In effect, we’re told that the 22% “Jews of No Religion” reported by Pew in 2013 was a straightforward group made up of people who consider themselves Jewish, but not by religion. By contrast, the 20% Jews-No-Religion in 1990 might be inflated by including nonreligious persons with Jewish parentage who don’t consider themselves Jewish. At least, not “in the same way” and “not all respondents,” whatever that means.
For a valid comparison to 1990, Pew says, we’d have to readjust the 2013 “No Religion” category to include those nonreligious persons with Jewish parentage who don’t consider themselves Jewish. If you do that, then the 20% found in 1990 rises to 26% in 2013, not 22%. So there has been a change.
But this raises two problems. First, it doesn’t make sense to assume that the 1990 “No Religion” category includes any and all interviewees with a Jewish parent and no religion who did not consider themselves Jewish. Too little is known about them to say for sure what category they ended up in. Too many doubts are raised by “not in all cases” and “not of all respondents.” It seems equally plausible that if we add the entire 2013 “I don’t consider myself Jewish” pool into the No Religion category, as Pew suggests, it might inflate the 2013 No Religion pool and falsely increase the difference between 1990 and 2013. The safest conclusion seems to be that an “appropriately cautious” reworking of the 2013 categories to compensate for the question that was sometimes missing (but sometimes not) in 1990 would result in a rise to somewhere between 22% and 26%. After factoring in a margin of error, that’s not much.
Remember, the published Pew report said that the proportion of Jews claiming “No Religion” had risen from 7% in 2000 to 22% in 2013, more than tripling in the space of just 13 years. I had argued that such a dramatic change in just over a decade is implausible. I looked to an earlier survey, from 23 years ago, which indicated the increase was a negligible two points. Now, to defend its report, Pew argues that the increase over 23 years might be as high as six points. How does that substantiate the claim of tripling in 13 years? My claim that “nothing happened” might — might — overstate the case, but it’s closer to the truth than the published claim that Pew is trying to defend.
They then suggest that the comparable No Religion category for 2013 might be as high as 31% if “we also include those who have a Jewish parent or were raised Jewish and who decline to specify their religion or who report practicing a religion other than Christianity (as the Born Jews With No Religion category in 1990 seems to have done).”
As a matter of fact, I put that very question to the director of the 1990 survey, Barry Kosmin, shortly after it was released. That survey had included, in addition to the 5.5 million “core Jewish” population, two additional categories labeled “Jews Converted Out” and “Jews of Other Religion,” roughly analogous to Pew’s “People of Jewish Background.” They totaled about 600,000 persons, and were not included in the core Jewish population — with or without religion. I asked Kosmin if “Jews Converted Out” (those with a Jewish parent who currently practiced a religion other than Judaism) might include persons who practiced Buddhism but continued to consider themselves Jewish. At the time this was a much-discussed group. His answer was a flat Yes. Persons who practiced non-Christian religions in 1990 were categorized as Jews Converted Out (not included in the core Jewish population), and not as Jews of No Religion (included in the core). So the conjectural boost in the 2013 figure to 31% doesn’t work.
Bottom line: Reworking the 2013 No Religion definition to match the 1990 definition produces a rise of at most 6%, and probably less.
Pew offers two other pieces of evidence to show that I was wrong about the No Religion increase. One refers to Barry Kosmin’s American Jewish Identification Surveys of 2001 and 2008, which used the same methods he employed in 1990. Those surveys, Pew reports, showed the Jews-No-Religion category increasing steadily “from 19% in 1990 to 28% in 2001 and 32% in 2008.”
That sounds bad for my side. On the other hand, Kosmin’s AJIS surveys also show the total American Jewish population declining steadily, from 5.515 million in 1990 to 5.34 million in 2001 and between 5.2 million and 5.4 million in 2008. Pew finds a current Jewish population larger than Kosmin’s by at least a million. Talk about a poor basis for comparison.
What’s more, Pew’s total population figure closely matches three other counts, which use three different methods but all reach roughly the same conclusion: the Steinhardt Social Research Institute at Brandeis University, the Institute for Jewish and Community Research in San Francisco and the work done by Ira Sheskin and Arnold Dashefsky for the American Jewish Year Book. All four studies produce figures between 6 million (in 2000) and 6.4 to 6.7 million (today). Kosmin’s figure is an outlier, and must at a minimum raise questions about the compatibility of his work with the rest of the field.
This is not to disparage Kosmin’s work. He’s been a pillar of the Jewish demography field for decades, a pioneer in many ways. But the AJIS studies raise a host of questions that haven’t been fully addressed.
Among other things, Kosmin finds that Jews-By-Religion comprise a shrinking share of the already shrinking Jewish population, which is becoming increasingly Jews of No Religion. Not surprisingly, the Jews-By-Religion share of the overall U.S. population is shrinking even more dramatically. He finds Jews-By-Religion (adults and children combined) totaling just 3.16 million in 2008, or 1% of the U.S. population. The 1990 NJPS under his supervision had found between 4.3 million and 4.4 million Jews-By-Religion, or 1.8% of the overall U.S. population.
Amazingly, 1.8% is precisely the share of Jews-By-Religion in the general U.S. population that Pew found in 2013.
This suggests two things. First, it’s another clue that the relative shares within the overall Jewish population of Jews-By-Religion, and presumably Jews-No-Religion, might have remained stable over the last quarter-century, as I’ve argued. Definitions of Jews-No-Religion may be elusive, but the Jews-By-Religion category hasn’t changed much.
Second, it gives a new context to another of Pew’s major claims. In the opening paragraphs of their study, along with the alarming figures about the rise in No Religion, we’re told that the “percentage of U.S. adults who say they are Jewish when asked about their religion has declined by about half since the late 1950s and currently is a little less than 2%.” That’s a reference to the the U.S. Census Bureau’s Current Population Survey for 1957, which included a question about religion. Those who said their religion was Jewish made up 3.2% of the total U.S. population (age 14 and over) in 1957. Comparing that to today’s 1.8% reinforces the impression that we’re currently in free fall.
But if the share of Jews By Religion fell from 3.2% in 1957 to 1.8% in 1990, as Kosmin’s NJPS indicated, and remains 1.8% today, as Pew found, then the dramatic changes in American Jewry took place in the 1960s, 1970s and 1980s and have stabilized since then.
Pew contains other evidence of that post 1990 leveling-off. It shows the intermarriage rate among Jews rising from 17% before 1970 to 55% around 1995, and leveling off since then. The biggest increase came during the 1970s, when it doubled from 17% before 1970 to 36% by 1979.
(Incidentally, Pew now accuses me of comparing their intermarriage figures to some other survey in order to show a leveling off. I did not. I relied entirely on Pew’s own figures, which show the leveling-off quite clearly.)
Other sources have shown a pre-1960 intermarriage rate in the single digits. That means the rate skyrocketed over two decades during the 1960s and 1970s, doubling from 9% in 1960 to 17% in 1970 and again to 36% in 1980. It then rose more slowly during the 1980s before leveling off sometime in the mid-1990s and has been stable for the last two decades.
Pew offers one more piece of evidence that No-Religion is on the rise. Sorting their respondents’ answers by age, they find that among those born between 1914 and 1927, fully 93% say their religion is Jewish and just 7% say they have no religion but consider themselves Jewish in some other sense. The percentage claiming no religion rises steadily by generation, reaching 32% among Millennials, born after 1980.
This looks like strong evidence. If the other indicators were stronger, this might confirm them. But given the weakness of the other arguments, this one ought to be examined for other possible explanations. Specifically, it’s possible that identification with religion grows as the individual Jew ages.
This isn’t just fanciful conjecture. Brandeis has studied levels of attachment to Israel, which increase by age cohort, much like attachment to religion. It found, however, that the overall attachment has remained stable over time. If attachment were declining over the generations, the overall rate should drop over time. There should be some resemblance between 65-year-olds today and 40-year-olds in 1990, since they’re the same people. But there isn’t. On the contrary, Israel attachment among 65-year-olds today resembles attachment among 65-year-olds in 1990. Ditto 40-year-olds.
What has changed in recent decades, incidentally, is Israel attachment among the under-30s. It’s gone up, apparently because of Birthright. That doesn’t necessarily teach us anything about attachment to religion, though it’s worth recalling that fasting on Yom Kippur has more than doubled among Jews-of-No-Religion since 1990 (from 10% in NJPS to 22% in Pew).
The point is that if attachment to Israel is related to life-cycle rather than (or in addition to) generational cohort, then the same might possibly be true of attachment to religion.
What’s more, if the percentage of Jews-No-Religion has been rising steadily with each successive generation, as Pew’s numbers seem to indicate (1914-1927, 7%; 1928-1945, 14%; 1946-64, 19%; 1965-1979, 26%; post-1980, 32%), it seems highly improbable that the overall average among all generations in 2000 could have been 7%, the same as the oldest cohort. Averages must by definition fall in between the two extremes. They can’t be identical to one of the extremes.
Once again, Pew’s new survey is a compelling document on the whole — a landmark in some senses — with a wealth of important information on the state of Jewish life today. It appears they made one noteworthy mistake in interpreting their results, tripping over the very pitfall they warned against — drawing too-easy comparisons with earlier surveys. That shouldn’t detract from the overall value of the work.
It’s unfortunate that their one mistaken assertion becomes the one thing that strikes a chord in the public mind. But it happens again and again, like the 1990 survey with its inflated 52% intermarriage rate and the 2000-01 survey with its false decline in Jewish population. We find the bad news to obsess over, even if we have to invent it. The good news washes off our backs.
This is not to say that Jewish life in America is becoming steadily richer and the Jewish future more promising. The point is that we don’t know. We’re in the middle of a historic transformation. The present moment is a shaky vantage point from which to try judging the outcome.
One vitally important thing that Pew should teach us is that American Jewish life has stabilized after an explosive transformation in the 1960s and 1970s. It’s still too early to tell where we’ll end up. It’s worth remembering, though, that there was a time when the priests in the Holy Temple in Jerusalem looked around and saw Judaism going down the drain as new generations replaced good old animal sacrifice with this newfangled prayer stuff. You never know how things are going to turn out.
And one of the critical lessons concerns researchers and how they frame their reports. One eye needs to be trained on the press and the public. One of the great sages of late Temple era, Avtalyon, is remembered as having given a warning to academics about how they frame their messages (Pirkei Avot 1:11): “Scholars, be cautious in your words, lest you be banished to a place of muddy water, and the students who come after you drink it and die, and the name of Heaven be desecrated.”
J.J. Goldberg is editor emeritus of the Forward, where he served as editor in chief for seven years (2000-2007).
Dear Pew: I Was Right. Here's Why.