Jerusalem Center for Public Affairs
Daniel Elazar Papers Index

Jewish Community Studies

Enhancing Jewish Population Studies

Daniel J. Elazar

Since 1980, at least 33 Jewish communities, including the largest ones, have conducted population surveys, so that about 73 percent of the total American Jewish population has been surveyed within the last five years. A growing number of the 1980 surveys represents the second or even the third round of studies for their communities, attesting to the value of the first rounds and pointing to the increasing prevalence of regular collection of population data on the local plane. Judged, therefore, by the number of studies, by the percentage of total Jewish population covered and by the growing number of repeat surveys, the demographic study of America's Jews has advanced remarkably.

Ironically, despite the proliferation of these surveys and some advances in their level of sophistication, their quality and usefulness for research purposes still varies considerably. Problems of comparability and generalizability persist. However, despite these shortcomings, the literature on the demography of American Jews (Goldstein, 1981; Goldscheider, 1982; Cohen, 1983a) clearly documents that in this last quarter of the 20th century the Jews of America are characterized by the following facts:

  1. low fertility and decreasing household size.

  2. high levels of intermarriage,
  3. considerable population movement and redistribution,
  4. a substantial aging of the population, and
  5. high levels of education and occupational achievement concurrent with shifts in the types of work and levels of self-employment.

Striking as these underlying similarities are, considerable intercommunity variations still exist in the specific levels of demographic processes and in demographic and socioeconomic composition.

These variations suggest that a community's size, its age, and its regional location may all have important effects on its population structure and dynamics. This is a key reason why individual community studies are needed in addition to reliable national data. All too often, evaluations of the viability and the vitality of the American Jewish community as a whole, based on the demographic features and indicators of Jewish identity for individual communities, have led to very different conclusions about future prospects. In part, these differences reflect real variations among the particular communities; in part, however, they may also result from variations in the quality of data and the sophistication of the analysis. They may even reflect differences in the proclivities and perspectives of individual researchers that affect their optimism or pessimism in interpreting the future of the Jewish community. Isolating those effects that emanate from the research design and data from those reflecting real differences is one of the major challenges researchers face. Eliminating the problems associated with the former rates the highest priority.

Identifying the Universe and Choosing a Sample

Defining who is to be regarded as a Jew is a basic research task, regardless of data source and who is to be included in the sample. Identification and/or inclusion should be based on the Jewish identity of the parents or even of grandparents at the time of the birth of the respondent? Is it to be based on the religion in which the respondent was raised or in which his or her children are being raised? Is it to be the Jewish identity of the head of the household, regardless of the respondent's identity? Will it follow halachic standards, or is it to be based on self-identification or on selected behavioral tests?

As we know, a definition based on household criteria can make a vast difference in the number enumerated as Jews; in National Jewish Population Survey of 1970-71 the differential amounted to 430,000 persons (Massarik and Chenkin, 1973). The inclusion of non-Jewish spouses or children of a Jewish parent considerably distorts selected characteristics of the population being studied. Yet, given the situation in the United States, is it possible from a research perspective to completely exclude all household members who do not either identify themselves as Jews or qualify halachically as Jews at the time of the study; many play an important role in influencing the intensity of Jewish identification of other members of the household (Mayer, 1978, 1983).

It is necessary to develop a standardized classificatory scheme with information collected on individuals from all the categories. For some purposes certain persons may be included in the analysis while for other purposes, they may be excluded. A standard set of criteria and availability of data on all categories would also enhance comparability across communities and allow aggregation of results from various studies.

Selecting the sample is another delicate task. Here again, a host of alternative methods have been developed. The literature is replete with examples of reliance upon federation lists, distinctive Jewish names, language spoken, residential clustering, and even friendship networks as mechanisms for identifying and/or screening Jewish households and Jewish individuals (Himmelfarb, Loar, and Mott, 1983; Kobrin, 1983; Lazerwitz, 1983; see also Cohen, Woocher, and Phillips, 1984). Serious questions remain about the representativeness of samples that result from reliance upon these methods for identifying Jewish respondents. A particular method that seems to work well in one community may be far less appropriate in another because of differences in the generational and ethnic composition of the Jewish population, the levels of intermarriage, and the rates of affiliation. All too often, one or another of these approaches has been used almost indiscriminately in choosing samples and at least partially determining who is Jewish, despite the obvious biases such approaches may have.

Comparability and Standardization

It is clear that more careful attention needs to be paid to standardizing the procedures used in identifying our universe, designing samples and questionnaires, tabulating data and analyzing and reporting the results. The goal should be to ensure maximum comparability among community studies while still meeting the unique needs of individual communities.

Gary Tobin and Julie Lipsam (1983; see also Tobin and Chenkin, 1985) undertook a most ambitious effort to assemble and, where possible, to compute comparable data collected since 1979 from population surveys in 13 American cities. Their task was not a simple one; their efforts and frustrations in interpreting some of the differences among communities make clear the problems of comparability that arise because different communities use different sampling techniques and interviewing methods, define the same variables in different ways, ask the same basic questions in varied forms, and tabulate and report the findings in different formats or not at all. A few examples will illustrate the problem.

Why does Cleveland's 1981 survey show an average household size of 2.8 and Minneapolis 2.6, compared to only 2.3 for St. Paul and 2.2 for Denver in the same year? Why are Cleveland and Minneapolis characterized by the same high level that characterized the 2.8 average of NJPS ten years earlier, while Los Angeles, St. Paul, and San Diego households averaged almost half a person less in 1981? Have the latter declined? Was their situation already different from the national average, or are the changes due to differences in coverage and/or definition? Did some surveys include college students living away from home, often in other locations, while others, like the census, did not count such absent students as members of their parental households? Did some count Jewish college students living off-campus in their own units, thereby inflating the percent of small household units, while others missed college students altogether? Did some include institutional populations, such as students and residents of homes for the aged, as household members or perhaps even count them as individual households, while others overlooked them entirely? Why were 30 percent of the households in Denver and 33 percent in Los Angeles one-person units, whereas Cleveland had only 19 percent and Chicago 21 percent in the same period? Is it a function of differences in the general community, in age composition, in living style, or is it a definition/measurement problem? We rarely are given enough information either on study design or in the interpretation of the results to gain insights on such issues.

Do we know why in Denver, Seattle, and Rochester 20 percent of the Jewish population was under age 20, while in San Diego, Los Angeles, and Nashville it was about 30 percent? Footnotes in a few studies suggest that problems of categorization may account for some of the difference. For others, researchers can only speculate whether differentials in enumeration procedure or differentials in migration, fertility, or mortality account for the variations. Unfortunately, the general absence of direct data on fertility, and inadequate information on migration, with proper controls for age composition, preclude gaining further insights on the roles of these factors.

Similar comparative assessment can be made for marital status. Not surprisingly, 23 percent of Miami's relatively aged Jewish population is widowed, but it is not at all clear why in St. Louis 17 percent are widowed since only 22 percent of the population is aged 60 and over while in Minneapolis, with about the same percentage aged 60 and over, only 7 percent are widowed. Comparable questions could be raised about the percentage single, which ranges from lows of 7 percent in Miami and 9 percent in St. Louis to a medium level of 15 in New York and highs of 22 percent for Minneapolis and 23 percent in Chicago and Denver. Chicago's may be explained, in part, by inclusion of all persons 18 years old and over in the statistics on marital status, but what accounts for the highs in Denver and Minneapolis? Are there differences because some refer to all adults (Chicago), some to household heads only (New York), and others to respondents only (St. Louis), or do the patterns vary because the communities actually differ in the marital status of their populations?

Other socio-economic indicators show similar high degrees of variation. The percentage with college education varies from a low of 57 percent in St. Paul to a high of 78 percent in Chicago; the percentage of professionals ranges between a low of 23 in Minneapolis and a high of 45 in Rochester; and managers/proprietors vary between Los Angeles' low of 16 and Minneapolis' high of 42. Even blue collar workers show a ratio of almost 3:1 between Los Angeles' high of 11 percent and Nashville's low of 4 percent. Yet, despite the low percentage of professionals/managers (49 percent) in St. Louis in 1982, the city was reported to have 40 percent of its households with incomes of $40,000 and over, compared to only 25 percent with such incomes in Rochester in 1980, where 64 percent of the work force were professionals/managers. The mysteries of intercommunal variation thus persist.

This comparative evaluation is, of course, oversimplified, and the stress on differences is perhaps exaggerated. What needs to be emphasized is that all too often, despite our generalization about such characteristics as family status, socio-economic status and age structure (many of which may be generally correct), individual community variations seem to be substantial. In the absence of standardized definitions, coverage, and tabluations, however, the inconsistencies may be spurious and lead to erroneous conclusions about the nature of the specific community, about differences between communities, and about change over time. Even more serious, using data based on one or two communities to generalize about the total American community or to test national data derived from alternative sources, as was done for the AJC studies, can be dangerous.

The foregoing discussion clearly indicates that standardization of procedures in all phases of research is the key to enhancing the quality and the usefulness of future Jewish population studies. All studies have in common a large number of variables that are recognized as crucial to understanding the community and to planning for its future. The difference lies in how they are identified and how to sample the population to be studied, and how to define these variables and handle them in the data gathering, tabulation, measurement and analysis stages. Recognizing the need for standardization and for comparability does not deny the importance of tailoring studies to the needs of individual communities. With proper attention to defining concepts, choosing samples, asking questions, measuring variables, and tabulating data, inconsistencies due to varying procedures and definitions can be eliminated, and higher levels of comparability can be achieved at the same time that information is obtained to meet the needs of individual communities. Both the larger community and the local community can then be better served, and both the resulting studies and community planning based on them will be greatly enriched.

Interpretation of Data: The Fertility Debate

Even more frustrating than the unexplained inconsistencies found in results obtained from different community studies are the inconsistencies obtained in a single community even in the same data set when different authors interpret the data, especially when the same author does so at different times. The current debate raging over whether the Jewish population is in danger of seriously shrinking and whether ferrtility is at replacement or non-replacement levels is a case in point (cf., Schmelz and Della Pergola, 1983, and Goldscheider and Goldscheider, 1985). As before, the problem of inconsistency is illustrated by citing some of the literature, especially those papers authored by Steven Cohen and Calvin Goldscheider. Both scholars have come to assess the demographic situation of American Jews quite differently both from their own earlier evaluations and from those offered recently by other researchers working in the field. The inconsistencies among these assessments lend added weight to the need to overcome the limitations inherent in our current research designs and data sets and to achieve greater standardization and comparability in our research efforts. Only by doing so, can we hope to assure a full and correct evaluation of the dynamics of demographic change in the Jewish community and the implications of these changes for the future.

In his comprehensive, insightful analysis of American Modernity and Jewish Identity, which relies heavily on data from the 1965 and 1975 Boston Jewish Community Surveys, Steven Cohen (1983a:118) concluded that

"On the basis of past experience, it does seem safe to say that the completed Jewish birthrate for today's Jewish parents may remain well below the number needed for replacement....Barring a significant rise in national fertility, Jewish birthrates--if they follow historic patterns--should continue to reside in the region of NGP (Negative Population Growth)."

Following review of a number of factors to explain this situation, Cohen concluded further,

"Undoubtedly, these several aspects of Jewish distinctiveness aside, so long as middle-class, urbanized Americans experience low birthrates, so will comparable Jews. Jewish birth patterns will generally follow those of the larger society, as they have in the past. If anything, advancing assimilation may well bring Jewish fertility behavior into even closer alignment with that of their non-Jewish contemporaries." (Cohen, 1983a:120)

The data cited by Cohen seem firmly to support this conclusion. Boston's ever married Jewish women aged 25-34 averaged only 1.5 children in 1965; in 1975, this age group had an average of only 0.7 children. The national data cited for Jews, although somewhat higher at 2.1 and 1.2 in the 1960s and 1970s, respectively, pointed in the same direction.

Yet, only two years after writing this, in a September 1984 interview in Moment, Cohen and Calvin Goldscheider argued to the contrary, claiming that the data of the 1960s and 1970s were a fluke of the particular cultural moment in American history and that the issue was one of time--reflecting late marriage rather than reduction in total fertility. Cohen suggests that "even if we assume that the women of the late '60s and early '70s were indeed less family oriented, less prone to have large families--which is very doubtful if you look at the data (a reference to New York data which appeared to show an average of 2.1 children for women 35-44) - there is no reason to assume that the same predilections are carrying over into the early and mid-eighties" (Cohen and Goldscheider 1984:41).

Goldscheider, in the same interview, questioned the validity of any reduced fertility conclusions, suggesting that this issue hinges largely on the proportion of Jewish women marrying and on the timing of their fertility in relation to age at marriage. He acknowledged, however, that it is not really known whether marriage patterns have changed, and whether, as a result, overall fertility levels have changed or whether they are below replacement, if one takes into account the non-married. As Goldscheider himself stated so well in an earlier interview with William Novak:

"What has changed? As I said, most Jewish couples tend to have two children; relatively few have none, or only one child, or more than two. But what is changing is the proportion of people who are getting married - it has declined significantly in the past 10 years. If everybody gets married, and has about two children, then we have replacement, and possibly even a little growth; certainly we have stability. But if suddenly 20 percent of the Jewish population doesn't get married--and that's a hypothetical figure--then you can see how that changes things. In the past, universal marriage has been a mainstay of Jewish life; suddenly, we can't take it for granted." (quoted in Novak, 1981:51)

Clearly then, the situation is very complex. All agree that in the United States, Jews have averaged lower fertility than non-Jews for a century or more. It is also true that at a time when most Jewish women married, Jewish fertility hovered at about 2 children per women and, therefore, met natural replacement levels. Few would dispute Goldscheider's (1985:12) strong belief that "projections about the drastic numerical decline of the American Jewish population in the next generation is demographic nonsense." I have criticized this overly pessimistic view, too (Goldstein, 1981). However, in the absence of a sharp reversal in trends, some decline still seems likely to me; the extent remains open to question until inconsistencies in evidence, in interpretation, and in ways of measurement are resolved. The argument (Goldscheider, 1985:12) that "neither the educational attainment nor the career orientation of younger American Jewish women poses a threat to the demographic continuity of the American Jewish population" also remains to be tested definitively to ascertain its generality. If Cohen's earlier claim that Jewish fertility will likely follow or align with that of the larger society proves correct, there is added reason to question the validity of Cohen's and Goldscheider's more recent optimism about the level of Jewish fertility and its adequacy for replacement.

In view of past patterns of lower Jewish fertility, why should the future situation be characterized by averages equal to or higher than that of the general population--which would be the case if the revised Goldscheider-Cohen view holds and if U.S. Census Bureau projections prove correct.

The Bureau of the Census (Das Gupta, 1985) has estimated that the average completed number of children born by the end of childbearing age to white women, who had no children at age 20 and who reproduced at 1980 rates during their reproductive years, would be only 1.48, and for women childless at age 24 (probably closer to the current Jewish age at marriage), the average would decline still lower to only 1.08, about 50 percent below replacement level. If, as data suggest, a high percentage of Jewish women marry late, and if they should follow such fertility patterns, these projections raise serious doubts about the persistence of replacement level fertility. Goldscheider is correct in stressing that marriage is a key factor, but his arguments are often misread by wishful thingers when they overlook that an average of 2.0 children on the part of the married is not sufficient to replace both themselves and the unmarried who have no children, especially if the latter constitutes a substantial group; an average below 2.0 compounds the problem.

The complex situation is further illustrated by data from the latest Bureau of the Census (1983) fertility report on births completed and birth expectations. It was found that for wives of all races who had married at ages 22-24 and who were already aged 25-29 in 1983, average expected completed fertility was 2.07; for those aged 30-34, the average was 2.08. Even if Jews follow these patterns exactly, the fertility of married women would be just at replacement level. The question remains: how many will not marry and how much higher would Jewish marital fertility have to be to compensate for the non-married? Furthermore, will Jewish women in fact follow these marital fertility patterns?

This census survey also showed that among white women aged 18-34 with four or more years of college, the expected average completed fertility, will be 1.93 children, with 14 percent being childless. For those with five or more years of college, the average will be 1.72 with 19 percent childless.1 Among women aged 18-34 surveyed by the Census who were not married at the time, the expected lifetime births averaged only 1.83 per woman, almost half a child less than the average for currently married women of similar age; for 25-29 year-old unmarried women the average was only 1.56. The expectations of the unmarried, therefore, provides no basis for believing that the averages will rise above replacement for all women if these single women eventually bear children.2 To the extent that Jewish women are characterized by both late marriage and high levels of education, it there any reason to believe that Jews in the future will deviate substantially from these general patterns and in ways different from the past?

When analyzing data on expectations, one should particularly keep in mind the assessment of childbearing intentions offered by Nathan Keyfitz's (1982) evaluation of the various efforts made since 1940 to rely on surveys to elicit childbearing plans. While recognizing that the Bureau of the Census relies on such procedures, Keyfitz (1982:741) concludes that, "after 40 years of effort, this instrument (survey) has come to seem uncertain, even controversial." Moreover, he stresses that extension of the questioning to women who are not yet married is not likely to add much information; "a girl of 15 can hardly give a meaningful answer to the question of how many children she intends to have" (Keyfitz, 1982:741). Clearly, data on expectations for the total population and for Jews need to be used with considerable caution, especially when they are dependent on expectations with respect to marriage behavior.

Although these data are obviously of very limited use in assessing current and future Jewish fertility because they refer to all whites and not just Jews and in a few instances to whites and blacks combined, and because they include inadequate controls for a host of factors, they do serve a purpose. To the extent that Jewish fertility in the United States has rarely if ever exceeded non-Jewish fertility and more often has been substantially lower, the data for total whites provide an upper limit - one which Jewish fertility would resemble if it came up to the national average in future years (or if the national average came down to Jewish levels). As such, it would seem more reasonable to hypothesize that Jewish fertility will not likely exceed replacement levels, and more likely will be below them. This interpretation, like Cohen's and Goldscheider's views, remains speculative. If we are to accept that Jewish fertility behavior will rise above the national levels, then the evidence and reasons for this very basic change in pattern has to be presented much more clearly and convincingly than has been the case to date.

Above all, these illustrations are intended to point out the inconsistencies that characterize both our data and especially our interpretation of them (cf., Schmelz and Della Pergola, 1983). Beyond this, they are intended to argue strongly for the need to include appropriate questions on marriage and fertility in Jewish population surveys and for appropriate measurement of the resulting data.3 One outstanding defect of Jewish population studies, both past and present, is the limited and often misleading analysis, due to faulty measures or comparisons, of fertility. It is ironic that the one demographic variable that may be of greatest interest to those concerned with the future of the Jewish community tends to be the most neglected of all. The current debate, I believe, is a healthy one if it serves to stimulate more careful research. It is dangerous only if the findings on either side are accepted uncritically as the basis for what may be false alarms or unjustified complacency. The inconsistencies challenge all demographers to greater efforts at resolution.


Given the limitations of both methods and available data, there is evidence that the research done to date provides no firm basis for resolving the debate about the current or future number of Jews. The existing data sets raise and leave unanswered a number of key questions about success in identifying and measuring the total Jewish population and in assessing the interrelations between population change and the current and future character and strength of Judaism and Jews in the United States. What these studies do suggest, despite their limitations, is that the Jewish population has undergone dramatic changes and will continue to do so; that these changes have significant implications for the degree and character of Jewish identity and Jewish practices; that the extent of Jewish integreation into the larger American scene as well as the persistence of Jewish exceptionalism in certain areas will jointly determine future Jewish demographic behavior and patterns; and that while future Jewish patterns will therefore probably more closely resemble those of the American population as a whole with respect to many variables. It will not necessarily do so for all characteristics.

Both the changes occuring to both Jews and the national population generally and those unique to the Jews make it most important to continually monitor the demographics of American Jewry. It must be possible to assess their implications for individual localities and for the larger Jewish community, and to accomplish this whenever possible within the comparative context of the changes occurring in the larger community. Researchers therefore continue to be challenged by the necessity of insuring that their methods and the resulting data will allow the best possible assessment of the present situation of the community and where it is heading.

In-depth evaluation of studies already completed and of alternative data sources should rate the highest priority if we are to make our future studies more accurate and more useful. Fortunately, we have begun to see some such critical assessments, as, for example in the use of DJNs for sampling (Phillips, 1983; Lazerwitz, 1983; Massarik, 1983). These mark only a small beginning; they must be followed by concerted efforts to assess alternative approaches to sampling, data collection, measurement, data analysis and population estimates, and to resolve the differences in conclusions about the levels of fertility, intermarriage, converson, assimilation and overall growth. Researchers must strive to agree on standardized definitions and procedures and on a set of up-to-date procedural guidelines. Results must be obtained that allow comparative analysis over space and assessment of change over time, as well as cumulation of comparable local studies to obtain regional and eventually even national data sets.

One basic goal for such efforts is development of a good current standard against which to judge the representativeness or even the reasonableness of the results obtained. NJPS has provided some clues, but it is far too outdated to constitute an acceptable standard. Use of individual communities such as New York, Boston, or Los Angeles is questionable. In the absence of such a single standard, we must give much more thought to appropriate alternatives. This includes consideration of how one or more synthetic standards might be developed from existing, available information, and of whether aggregation of weighted data from individual community studies would better serve the purpose.

Under ideal conditions, a new, well-designed national study would provide such a standard. In its absence, achievement of a high degree of standardization and comparability among community studies might well provide a viable alternative, especially if a large number of communities and a high percentage of the total population continue to be covered by local sample surveys. It is, however, highly unlikely that we can obtain national results by aggregating indiscriminately the results of individual community studies. But with proper planning and centralized supervision, it may be possible to select a range of communities which are judged by experts to be representative of the country as a whole. By then using these particular communities as a kind of national sample, and aggregating their individual survey results, approximation of a national sample survey may be achievable. Moreover, by properly staggering the times at which communities of different size and type undertake their surveys, a more continuous assessment of the changing Jewish-American scene might also be possible.

In sum, we must continuously enhance and substantially modify our thinking about the design, conduct, and utilization of Jewish population studies (cf. Cohen, Woocher, and Phillips, 1984). Reevaluation becomes especially necessary as we gain new experience from ongoing studies and from analysis of data sets collected earlier; as our colleagues in the social sciences develop more sophisticated methods of analysis; and as we broaden our reliance upon varied sources of information. Such constant reassessment is especially critical because of constraints on undertaking new studies. Demographers must also inventory, evaluate and prescribe methods for exploiting every possible data source that allows them to make relatively reliable assessments of changes in the size, composition, and distribution of the Jewish population, as well as in its components of change. Only through such full and careful exploration of data will answers be obtained to the vital questions of growth vs decline, replacement vs non-replacement fertility, and the negative vs positive effects of intermarriage and assimilation. Only in this way will researchers enhance the likelihood of resolving the inconsistencies that have come to characterize their findings and assessments, and in so doing, provide the community with a firmer basis on which to plan its future.


1. The effect of education is similarly indicated by the combined data from recently analyzed 1973 and 1976 National Surveys of Family Growth (Mosher and Hendershot, 1984). They showed the total births expected by Jewish married women age 15-44 to be only 2.16 for those with some college education, that is, just at replacement leve, and lower than that of less educated Jews and of college educated Protestants and Catholics.

2. In the same set of sessions at which this paper was presented, one presented by Calvin and Frances Goldscheider (1985) assessed the birth expectations of a cohort of young men and women, the High School Class of 1972. Using longitudinal data from interviews with this group in 1973 and 1979, the evidence lends support to the argument that American Jews will achieve fertility levels in the next several decades average close to two children per family, sufficient for population replacement." In 1973, the 197 Jewish women expected an average of 2.34 children. By 1979, the 167 who were followed-up expected 1.96. These averages assume that those who reported expecting to have 4 or more children will average 5.0 For the total white population (exclusing Jews) the corresponding averages were 2.40 and 2.06.

While these data in themselves suggest that this particular Jewish cohort will, if their expectations are fully realized, average close to 2 children per women, several questions must be raised about the likelihood of such an outcome. Given that between 1973 and 1979, the expected average declined by 0.4 children, or 16 percent, how can we assume there will be no further reductions? Between 1973 and 1979, 23 percent of the ever married and 31 percent of the never married had already reduced their expectations, whereas only 13 and 18 percent of these respective groups had increased the number of children expected. How many will, in fact, not marry at all? How many will face fertility problems? How many will experience divorce and how will this affect fertility? Beyond this are even more basic questions related to the nature of the sample itself. Religious identification was based on the question "What religion were you brought up in?" It does not tell us the religion at the time of the survey nor anything about religious identification after marriage. How this identification is or will be affected by marriage and how many respondents are or will remain Jewishly identified remains to be determined.

Assessment of the results of another survey of high school seniors in 1980 showed only 5 percent of the Jewish males and 2 percent of the Jewish females expect not to marry, the lowest percentages of any religious group, but a much larger proportion planned to marry late. The authors themselves recognize the limitations of their data. As they state, "as in the case of birth expectations, we do not know the predictive power of these marriage expectations" (Goldscheider and Goldscheider, 1985:19).

Taking all of these concerns into account, these data can only be suggestive of future patterns. In themselves they certainly do not point to above-replacement levels of fertility and, given the concerns expressed, one could easily build a case that they point to fertility levels below replacement.

3. Among the questions that would be appropriate for inclusion in surveys are the following: 1) What is your current marital status? 2) Have you been married more than once? 3) What was the date of your first marriage? 4) How was the first marriage terminated? 5) How many babies has the woman ever had, not counting stillbirths? 6) How many children in all do you expect to have?

Elazar Papers Index / JCPA Home Page / Top of Page