Monday, June 23, 2003

Standard Errors

Those who cling to their faith in demography should consider the sad story of the 2000 National Jewish Population Survey, which though still unreleased has been badly undermined. Or perhaps more tellingly, the faithful should ponder implications of the 1990 and 2000 U.S. decennial censuses. These were mandated by the Constitution and conducted with the full authority of the federal government, so that all were required to participate. There has been much controversy over the data, as well as litigation all the way up to the Supreme Court.

The problem with demography is not that it is error-prone. A more serious affliction is immodesty, the failure to acknowledge that what is being presented is nearly always based on assumptions and extrapolations that certainly contain errors and that the final set of statistics is, at best, an approximation of a social picture that cannot be fully rendered. This failure is enhanced by the tendency of the media to highlight the data and to gloss over the obvious warts and shortcomings.

We now have a study of New York’s Jews, the first in nearly a dozen years. There are, we are told, 972,000 of us in the city and 1.4 million in the metropolitan area. There is no way of knowing how accurate these figures are. They may be off by 100,000 or 200,000. My guess is that they are on the low side and yet I recognize that those who conducted the study went about their work honestly and developed several attractive conceptual twists that are improvements.

The primary methodological tool of Jewish demography is Random Digit Dialing (RDD). A huge number of telephone calls are made according to a pre-determined geographic pattern, the aim being to locate households with at least one Jew who is willing to respond to questions relating to Jewish identity. While RDD is an acceptable method for political polling and marketing surveys, it is highly suspect when employed to root out Jewish respondents, if only because it is often difficult to determine who exactly is a Jew. Furthermore, as Americans have become more resistant to telemarketing, the utility of RDD has been sharply compromised.

In order to plan the distribution of calls, it’s necessary to make in advance assumptions as to where Jews live. This means that a survey which is designed to determine how many Jews there are and where they live is directly based on earlier decisions as to how many there are and where they live. If this doesn’t raise eyebrows, Jewish demography is beset by difficulties and disagreements in identifying Jews. The household and not the individual is the standard unit in population surveys. This is fine when questions relate to gender, age and other neutral or readily ascertainable matters. But in contemporary American Jewish life, there are a great number of so-called Jewish households in which the majority of the residents are not Jewish. One doesn’t need to be a demographer to know that this circumstance can foul up the data.

The New York study defines a household as Jewish if it includes “an adult who self-identifies as a Jew, or a child who is being raised Jewish.” This definition has to result in the inclusion of persons of dubious Jewish status and also in the exclusion of persons who are Jewish according to religious law but who do not regard themselves as Jewish.

In order to reduce reliance on RDD, the survey contacted a considerable number of persons known to be involved in Jewish life who are included on lists provided by Jewish agencies. The device increased the response rate and it also means that the current study is not fully comparable with its predecessor, a point that has already engendered controversy. Another probably consequence is that the survey gives greater weight to involved Jews and has undercounted secular/advanced assimilated Jews.

This helps to account for the unexpectedly low intermarriage rate suggested by the data. Whether measured by individuals or couples, the rate among New York Jews is, I believe, considerably higher than what we have now been given. I am also skeptical about the claimed number of Russian Jews, a figure that is inconsistent with U.S. census data and what information we have about immigration patterns. The subject deserves additional scrutiny.

In a throwback to a bygone era when some Jews actually kvelled when they heard bad news, our media have pounced on the dubious finding of many more poor Jews than was previously believed to be the case. Our povertycrats and others have reacted with glee, probably because they believe that this will result in greater philanthropic and governmental funding. While there are substantial pockets of poverty among New York Jews and some are very severe, the claim of more than 200,000 poor Jews is absurd. It needs to be underscored that RDD is a fraudulent way to study money matters, whether income or wealth or poverty, as the respondents on the other end of the line are apt to be unwilling to disclose such information, assuming that the person who answers the telephone knows the facts, something that often is not the case.

For forty years or since the Great Society of Lyndon Johnson with its layers of self-serving agencies acting as intermediaries between the poor and government, povertycrats and their agencies have had a field day milking the poverty issue, with scant benefits going to the poor. When we fall prey to exaggerations – and especially among Russians – about income, we are opening the door wider to those whose rhetoric speaks a good game about the poor, while in fact the hard core poor continue to be neglected.

We will have to wait about six months for more comprehensive data and a fuller discussion of certain technical issues. From what we have seen so far, we have a fine effort to get at an elusive subject.