The Relationship of Jewish Community Contexts and Jewish Identity: A 22-Community Study

Abstract

This paper explores the manner in which Jewish community contexts relate to Jewish identity. We employ the Decade 2000 Data Set that contains almost 20,000 randomly selected Jewish households from 22 American Jewish communities interviewed from 2000 to 2010. Because of the large sample size, and its incorporation of community infrastructure data, this research also is able to examine various influences on Jewish identity that have not been definitively addressed in previous research, including the manner in which characteristics of Jewish community infrastructure are related to individuals’ Jewish identity. The Decade 2000 Data Set used for the analysis is described and some of the methodological considerations involved in its use are presented. Jewish identity is conceptualized as multidimensional, and a factor analysis results in four Jewish identity factors: a communal religious factor, a private religious factor, a broader ethnic factor, and a local ethnic factor. Multiple regressions for each of the Jewish identity factors are related to Jewish community characteristics; more commonly researched individual-level variables (Jewish background and connections, family status, socioeconomic status, demographic/geographic characteristics); and survey-level variables (such as size of sample and year of study) are also controlled. Surprisingly, except for the local ethnic factor, Jewish community characteristics have little relationship to individual Jewish identity. The contributions to a “sociology of Jewish place” and suggestions for further research are also discussed.

Introduction

The purpose of this paper is to explore the extent to which Jewish community contexts are related to Jewish identity. We employ the Decade 2000 Data Set, which contains almost 20,000 randomly selected Jewish households from 22 American Jewish communities interviewed from 2000 to 2010. Because of the large sample size, this research also is able to examine various influences on Jewish identity that have not been definitively addressed in previous research. Because we are introducing indicators of community context, we are able to test the extent to which individual-level characteristics retain their previously found relationships with the various Jewish identity factors, even when community context is controlled. And because we control for individual-level indicators, we are able to discern the net effect of community-level characteristics on individual Jewish identity.

The bulk of the research on Jewish identity has focused on the relationship between Jewish identity and individual-level variables, and Cohen and Eisen (2000, pp. 183–184), in The Jew Within: Self, Family, and Community in America conclude:

More and more, the meaning of Judaism in America transpires within the self. American Jews have drawn the activity and significance of their group identity into the subjectivity of the individual, the activities of the family, and the few institutions (primarily the synagogue) which are seen as extensions of this intimate sphere.

They suggest that American Jews see themselves as autonomous individual choosers from a vast array of Jewish expressions of identity, few of which are dictated by their community of residence, as was the case for generations in Europe. Highly educated as a population, their virtual community has undoubtedly expanded even more since Cohen and Eisen’s (2000) study, increasing the options ever more beyond the scope of the individual community. More recently, Rebhun (2011) found that the Jewish identity of migrant individuals was only minimally impacted. He suggests that this reflects both the greater dispersion of American Jews across the country, as well as an increasing role of the virtual community in promoting integration of Jews into their Jewish communities.

However, little systematic research has been completed on the manner in which community context might be related to individual Jewish identity. The limited previous research suggests that community context may not only impact Jewish identity directly but also indirectly by modifying the relationships of individual-level characteristics with Jewish identity. We therefore believe it important to consider both communal context and individual-level characteristics in their relationship to individual Jewish identity. One of the questions addressed is whether the communal context reduces the impact of individual-level characteristics on Jewish identity, and vice versa.

Community contexts comprise settings for social relationships, services, and policies which are expected to be related to both religious and ethnic identity. Regnerus et al. (2004) refined the concept of contextual effects on adolescent religiosity as encompassing both microcontexts (e.g., friends, family, schoolmates) and ecological influences. Benson et al. (2003) tested the influence of one such ecological characteristic, the proportion of religious adherents in a community, but found that it has minimal impact on adolescent spirituality. Roehlhepartan et al. (2011) reviewed research that shows that interpersonal contexts (family, peers) and social-structural contexts (institutions such as schools, youth organizations, and religious facilities; culture; and place) as well as life-course events, are all related to the development of spirituality. By extension, one would expect contextual effects on Jewish identity development to include all these kinds of influences as well. While the role of microcontexts from the perspective of the individual has received considerable empirical attention, the role of social-structural contexts has been the focus of much less research.

Previous research has established differences in Jewish infrastructure across communities (e.g., Dollinger 2000; Sheskin 2001), as well as variations in religious identity and practices (both Jewish and non-Jewish) across regions or communities (e.g., Graham 1983; Kosmin and Keysar 2006; Phillips 1993; Sheskin 2001, 2005a; Silk and Walsh 2008; Smith et al. 1998; Stump 2008). One of the relevant sources of variation is size of the Jewish population. Christaller (1933) postulated that a critical mass (or threshold size of a population) was required to establish various communal structures. Extending this to a Jewish context implies that a minimum Jewish population is necessary to support institutions such as synagogues, Jewish community centers, Jewish schools, kosher shopping facilities, and other community organizations, all of which are expected to facilitate practices which reinforce Jewish identity. Cohen (1983, p. 108) suggested that Jewish communities could be characterized by: (1) residents’ aggregate characteristics (such as age, social class, and family life cycle); (2) maturity of their Jewish institutions; (3) density of their Jewish populations; and (4) proximity to major Jewish communities and central institutions. Family-oriented communities, established Jewish institutions, dense Jewish populations, and proximity to Jewish institutions were expected to facilitate expressions of Jewish identity.

In relating communal characteristics to Jewish identity, Rabinowitz et al. (1992) found that Jewish community population size was related negatively to attendance at religious services, but positively to ethnic Jewish identity. They hypothesized that a critical mass was needed for religious involvement to decline and ethnic engagement to increase. The intimacy, solidarity, visibility, and coherence of small Jewish communities (Weissbach 2005) are unlikely to support many major Jewish institutions or provide a large enough local Jewish marriage market to preclude significant intermarriage (Cohen 1983, pp. 108–109), which might undermine expressions of Jewish identity.¹ Cohen (1983, Chap. 5) found weaker communal affiliation in communities with a high Jewish population turnover, while Goldstein and Goldstein (1996) found that communities with high percentages of recent migrants are likely to have a youthful population in the early stages of family development, implying a ritually less observant population.

Some of the community context involves the relationship between Jews and non-Jews in the population, as Horowitz (1999) noted (e.g., the proportion of Jews in the population, not just their size; the proportion of intermarried). Studying Christian population practices across national settings, Kelley and De Graaf (1997) found that a higher proportion of religious as opposed to secular population augments the level of orthodox Christian beliefs in a country. Lazerwitz and Tabory (2002) showed how being a religious majority, as opposed to a minority in terms of religion, enhances Jewish religiosity in a country, even among the non-religious. Presumably such effects on Jewish identity would be even stronger when considered on a regional or more local level. Although in almost all American communities (and in none of the communities in the Decade 2000 Data Set), Jews are not the majority, their proportion does vary, as will be shown below.

As mentioned above, previous research has established that individual religio-ethnic identity varies by demographic (such as age, gender, household structure, and immigration status) and socioeconomic characteristics (such as education, income, and occupation) generally (e.g., Christiano et al. 2007; Demerath 1965; McCloud 2007) and specifically among Jews (e.g., Ament 2004; Hartman and Hartman 2009; Rieger 2004). Certainly intersections exist between the impact of geography and community context on the one hand and demographics and socioeconomic status on the other. After all, communities vary in their socioeconomic status. Much of the research on geography and community context has not sufficiently controlled for individual-level characteristics to determine whether an effect of the communal infrastructure exists beyond the demographic and socioeconomic characteristics of a community’s population. At the same time, the research on individual-level characteristics rarely considers the effect of “place.” The work of the “Religion by Region” project at Trinity College suggests that “place” indeed affects the manner in which religion is expressed, as well as its salience for public (and conceivably for private) life, and touches on some of the different expressions of public Jewish life in various regions of the United States (this project is summarized in Silk and Walsh 2008); however the comparisons of Jewish communities are broad and not detailed to the level of specific communities.

Earlier studies of contextual effects on Jewish identity suffer from smaller samples than ours, often in one particular community (e.g., Cohen’s 1983 Boston research). At the time of his study, Cohen suggested that “only a prodigious research effort with detailed historical and contemporary data on dozens of Jewish communities across the United States could even hope to tackle the task properly” (Cohen 1983, p. 109). Our data set enables us to approach Cohen’s vision by using local Jewish community studies containing much of this information.

We next present our three hypotheses (and seven sub-hypotheses) regarding the impact of community context on Jewish identity, describe the data set and methods employed, discuss the results of our analysis in relation to the hypotheses, and make suggestions for further research. Note that in our hypotheses we at times distinguish between four types of Jewish identity: communal religious identity, private religious identity, broader ethnic identity, and local community integration. These are derived from our factor analysis of the indicators of Jewish identity common to the 22 community studies in our analysis. The method of deriving these four factors and their meaning will be described in greater detail below. Note also that while most of our hypotheses are written in terms of the community context impacting individual identity, we do not really have the data to test causal relationships. The hypotheses are written in terms of the direction of impact that makes the most sense to us.

Hypotheses Related to Jewish Community Infrastructure/Context

Hypothesis 1: Individual Jewish identity is related to the nature of the Jewish infrastructure/context and population where an individual resides. More specifically:

Hypothesis 1a: The size of the Jewish community will impact Jewish identity; larger Jewish communities will act to strengthen ethnic identity, while smaller Jewish communities will exhibit stronger religious identity. Because larger Jewish populations can support more Jewish infrastructure and more Jewish programming (e.g., synagogues, kosher food outlets, Jewish agencies, and Jewish educational and cultural programs) and can allow more of a “community feeling” to develop, we expect that in larger Jewish communities, more interaction exists among Jews, more opportunities exist to express local ethnic identity, and stronger Jewish identification is manifested. This is both because the broader community sees a larger number of Jews and identifies them as such and because internally more Jews exist with whom to interact. This is supported by previous research from Rabinowitz et al. (1992).

On the other hand, Rabinowitz et al. (1992) found that smaller Jewish communities exhibited stronger religious identity than larger ones. Smith (2003) suggests that social marginalization (as would happen with a “threatened” minority) results in greater separation from the dominant (non-Jewish) establishment, which would thus strengthen individual Jewish identity. A further possibility is that in large Jewish communities, many people see that Jewish institutions are operating without their help. In smaller Jewish communities, those same individuals might come to the fore, believing that if they do not step forward, no one else will. Such an argument has been validated with regard to smaller congregational groups in religious congregations more generally (Dougherty and Whitehead 2011). It should be noted, however, that using data from NJPS 1990, Sheskin (1991) found that Jewish religious and ethnic identity was strongest in medium-size Jewish communities.

Hypothesis 1b: A higher percent Jewish in a community will be related to stronger Jewish identity. As with number of Jews, it is expected that a higher percentage of Jews in a community results in greater visibility and more identification as such by others, and hence strengthens Jewish identity. Note that the influence of this variable may interact with the religious or ethnic character of the broader community (discussed below).

Hypothesis 1c: Stable Jewish communities will be characterized by stronger Jewish identity. That stability is related to Jewish identity is supported by the work of Goldstein and Goldstein (1996) on the migration patterns of American Jews. They found what they termed “traditional holding power”: more stable communities tended to be more traditional and more observant (or it could be that more observant communities tended to be more stable). Earlier work by Goldstein (1981, 1990, 1991) found that: (1) high levels of geographic mobility may break down the ties of individuals to a community and its local institutions, thus undermining Jewish identification; and (2) participation of immigrants may never reach the level of the locally born population. Cohen’s (1983) work also found that communities with a high Jewish population turnover tended to have fewer Jewish communal affiliations. Goldscheider (1986) contradicted Goldstein and Cohen, suggesting that new migrants may expand existing networks and even create new ones in a relatively short time and may create networks that substitute for the formal Jewish community. More recently, Rebhun (2011) suggested that the effect of Jews’ migration patterns on Jewish identity and group commitment has diminished, as the extent and scope of their mobility has dispersed Jews throughout the United States, and the Internet has afforded quicker anticipatory socialization and adaptation to the new environment. We expected that stable or increasing synagogue membership would be associated with stronger Jewish identity, especially religious identity. Decreasing synagogue membership might reflect more turnover of the population and would imply waning interest in Jewish communal associations, so we anticipated it to have a negative relationship with strength of Jewish identity.

Hypothesis 1d: Jewish populations that are clustered in one part of a metropolitan area, rather than being geographically dispersed throughout that metropolitan area, will exhibit stronger Jewish identity. In addition to the reasoning presented under Hypothesis 2 below as to why stronger levels of Jewish identity exist in areas of greater Jewish population density, geographic clustering also results in a Jewish community that is more easily served by Jewish institutions. If a Jewish community is concentrated in one location it is easier for communal institutions such as synagogues, Jewish day schools, and JCCs to provide services. A Jewish community is more likely to be able to support Jewish retail activity, such as kosher restaurants and Judaica stores, if a large percentage of the community lives within a short distance of such establishments. Residentially based social services, such as home-delivered meals, are also more efficiently operated within a community that is geographically concentrated.

Hypothesis 1e: Jewish communities with a higher percentage of married couples who are intermarried will be characterized by individuals with weaker Jewish identity. Intermarried Jews typically have weaker Jewish identities (Cohen 2006), either as precedent to their intermarriage or as a result of investment in social capital bridging between Jews and non-Jews, rather than social capital bonding within the Jewish group (Phillips and Fishman 2006). We expect this to have an impact on the interpersonal climate in the community, especially where the percentage of intermarried is higher.

Hypothesis 2: The more developed the Jewish institutions in a community, the stronger the individual Jewish identity. More specifically:

Hypothesis 2a: The total number of synagogues in a community will act to strengthen Jewish identity, and specifically, the greater the number of Orthodox synagogues in a community, the stronger the religious identity. A greater number of synagogues in a community indicates that a variety of worship and congregational options exist in the community, and that households probably, on average, live geographically closer to a synagogue. Because most synagogues provide both religious and secular (or ethnic) communal activities in addition to religious services, a greater number of synagogues provide more opportunities to participate in Jewish-sponsored events.

In addition to offering a wider variety of places of worship, particularly for the Orthodox, the number of Orthodox synagogues is likely to be indicative of the existence of other institutions catering to an Orthodox lifestyle, such as kosher food establishments, mikva’ot (ritual baths), and eruvim (ritual enclosures). Orthodox synagogues are investments in social and cultural “bonding” capital which integrates the Orthodox community. A strong Orthodox infrastructure supports the lifestyle necessary for the indicators forming our private religious factor (keeping kosher inside and outside the home, lighting Friday night candles, attending synagogue services) (Table 3).

Hypothesis 2b: A larger Jewish Federation annual campaign per household will act to strengthen Jewish identity, especially on the local level. The size of the Jewish Federation annual campaign per Jewish household is reflective of the existence of a significant local Jewish infrastructure, even though in most communities a substantial portion (30–40 %) of the Jewish Federation annual campaign is used for needs nationally, in Israel, and around the world rather than for local programming and infrastructure.² The success of the annual campaign is indicative of the sophistication and organization of a Jewish community as well as its level of affluence. In addition, a more successful annual campaign generally implies the existence of stronger Jewish agencies that can provide more services and programs for the Jewish community, thereby enhancing Jewish identity further.

Hypothesis 3: The strength of the relationship between individual-level variables and individual Jewish identity will be impacted by community context. Stump (2008) found that strength of belief has the greatest influence on religious participation in the Pacific and New England regions, and the least effect in the East, South Central, and Mountain regions. He found fewer regional differences for the influence of socioeconomic status and demographic variables on religious service attendance. He did, however, find a significant effect of socioeconomic status and demographic variables on Protestant attendance in the South, reflecting the general importance of religion in that region, and possibly supporting the expectation that higher-status individuals actively support and participate in religious institutions at higher rates. More recently, Putnam and Campbell (2010) found that income did not have an independent effect on religiosity, once gender, race, ethnicity, size of place, and region were controlled. Controlling for region, therefore, may also clarify the relationship between the individual-level variables we use (described below) and Jewish identity.

Data

The 1971, 1990, and 2000–2001 National Jewish Population Surveys (NJPS 1971, NJPS 1990, NJPS 2000–2001) provide the largest national samples of Jews in single surveys (Kosmin et al. 1991; Kotler-Berkowitz et al. 2003; Massarik and Chenkin 1973). However, these studies are quite limited for the purpose of studying variations among local Jewish communities (Sheskin 2005b). None of the three NJPSs was designed to produce data at the local Jewish community level, and NJPS 2000–2001 was designed only to produce accurate data for the four major census regions. Further, the latest NJPS was conducted over a decade ago, and while much can still be gleaned from it, Jewish identity in its various expressions may have changed over the past decade. The North American Jewish Data Bank (www.jewishdatabank.org) has collected more than 200 local Jewish Federation-sponsored Jewish community studies, which offer greater potential for studying the manner in which Jewish identity varies for different types of individuals by community context.

There are two methods that may be used to facilitate comparisons of community studies. One is to analyze each community separately and to compile the results in an aggregated “meta-analysis” (Cooper and Patall 2009), an approach that allows for determining whether the same results (e.g., the relationship between age and Jewish identity or the relationship between education and Jewish identity) are found in different communal contexts. It works best when relatively simple findings are compared and the data (e.g., question wording, response categories, indicators, sampling methods, and data collection methodology) have been standardized for comparability.³ One disadvantage of meta-analysis of this type is that it does not enable analysis of population subgroups for which the sample size in any given community sample is too small.

An alternative and often preferable approach, when the data and resources allow, is to aggregate the individual data sets into a single data set, a technique variously termed “individual participant data” (Cooper and Patall 2009; Riley et al. 2010) or “integrative data analysis” (Curran and Hussong 2009), and conduct analyses on the aggregated individual level, while controlling for multiple levels of variation (e.g., community level, individual level, and survey level). Curran et al. (2008, p. 365) suggest that:

The strategy of pooling data drawn from separate investigations holds many benefits, including increased statistical power, greater sample heterogeneity in important subject demographics, the broader psychometric assessment of constructs, and the ability to estimate a variety of models that would not be possible within any single data set.

However, integrative data analysis also has drawbacks, including the challenge of standardizing measurement for construct development and the effort needed to standardize the questions asked and the response categories. For reasons elaborated below, we were able to use integrative data analysis as the primary approach for the current study.

Of the 36 local community studies completed from 2000 to 2010, all but six were conducted by one of two researchers (Ira M. Sheskin; Jacob B. Ukeles/Ron Miller). This analysis uses the 22 data sets collected by Ira M. Sheskin as the principal investigator since the completion of NJPS 2000–2001, which comprise the Decade 2000 Data Set. This limitation had a number of significant advantages. First, the questionnaire used in each of these local Jewish community studies was basically the same, with minimal variation from community to community in almost all basic measures of Jewish identity.⁴ The survey research literature indicates that even small changes in question wording or in the sequence in which questions are asked in a survey can have a significant impact upon survey results (Bradburn et al. 2004), so this was an important advantage. Second, and of major import, Sheskin had already compiled all 22 studies into a single meta-data file, having performed the preliminary comparisons of the questionnaires and eliminating (for the most part) variation by standardizing response categories. It should be noted that this preparation is extremely time-consuming and is mentioned as a major drawback for doing this kind of integrative data analysis (Cooper and Patall 2009; Curran et al. 2008). Third, numerous comparisons of community contexts were already available. In the latest community study (Middlesex) included in the Decade 2000 Data Set, comparison tables for almost 200 questions were included in the study report (Sheskin 2009). This greatly facilitated an understanding of the range of variation involved and types of communal comparisons possible with the Decade 2000 Data Set. Fourth, Sheskin used the same basic methodology for determining the survey sample (usually a combination of RDD [random digit dialing] and DJN [Distinctive Jewish Names] techniques) for each study. Note, however, the variation in proportion of the sample determined by RDD and DJN methods (Table 1).⁵ Fifth, the same procedure was used to select a respondent from the household to interview (any cooperative adult, Jewish or not, who answered the telephone in a Jewish household). In each study respondents were pursued intensively⁶ until a high cooperation rate was achieved. Sixth, all 22 community studies used the same definition of a Jewish person. A Jewish person was defined as any person who currently considers himself/herself Jewish (or who is identified as such by the respondent) or who was born Jewish or raised Jewish and has not formally converted to another religion and does not regularly attend religious services of another religion (irrespective of formal conversion). A Jewish household was defined as any household containing a Jewish person.

Table 1 Attributes of community studies in the Decade 2000 Data Set

Nevertheless, variability exists among the studies. Some questions had to be eliminated from our analysis because they were not asked in all communities or had been altered significantly from study to study. In terms of the samples, while most were derived from a combination of RDD and DJN techniques, they varied in the percentage of the sample that each sampling technique provided, and in Jacksonville, in addition to RDD and DJN samples, a sample was drawn from the Jewish Federation mailing list. Further, the studies varied in their cooperation rates, ranging from 49 to 97 % for the screener (which determined eligibility for the survey) and from 64 to 99 % for the survey itself. Further, the studies were conducted over 10 years, which may affect the results.

Table 1 presents the communities included in the Decade 2000 Data Set, the year of each study, the sample size, the percentage of the sample contributed by RDD and DJN techniques, and the screener and survey cooperation rates. Some of the variables in Table 1 were added as a final level of analysis (as survey-level characteristics) in the multiple regression analyses below, to see to what extent, if any, they contribute to the variation in Jewish identity once other sources of variation have been eliminated. (Because of multicollinearity issues, only a selection of these variables could be included in the multiple regressions.) For more detail on the methodology of each local Jewish community study included in the Decade 2000 Data Set, see Chap. 2 of the Main Report for each study. All Main Reports are available at www.jewishdatabank.org.

Finally, note that while the 22 community studies included do not form a national probability sample, the 19,800 interviews do randomly represent almost 547,000 Jewish households containing 1,247,000 persons, of whom about 1,081,000 are Jewish.

In summary, while challenges do exist in comparing the results among the Jewish communities and in combining the 22 separate studies into one, we have every confidence that the Decade 2000 Data Set represents a significant resource for the social scientific study of American Jews. The lack of such a data set for comparative analysis has meant that, until now, only a very limited number of studies have utilized local Jewish community studies for the types of analysis undertaken here (e.g., Phillips 1993; Rebhun 1995; Sheskin 2001, 2010a, b).

Methods

We chose to use multivariate analysis for most of this study because of the large number of variables to be analyzed and their multiple inter-correlations. The large sample size (n = 19,800) also made this possible and desirable.

First, we begin with a factor analysis of Jewish identity indicators. Factor analysis allows us to combine multiple indicators of Jewish identity that are highly inter-correlated with one another into a smaller number (in this case, four) of factors (or composite variables). Such exploratory factor analysis does not presuppose a structure for the dimensions of Jewish identity (that is, it does not “expect” a religious or an ethnic dimension and select variables for them accordingly), but rather allows the user to interpret the structure that the relationships between the variables themselves present. That the results correspond to dimensions found in prior studies, using different samples (as seen below), validates both prior theoretical assumptions and the factor analysis itself.

Second, we employ four multiple regression models to explain variations in the four Jewish identity factors resulting from the factor analysis. Our main contribution is the inclusion of community-level characteristics related to the Jewish community’s infrastructure for which our hypotheses are presented above. We also controlled for individual-level variables including Jewish background/connection variables, family status, socioeconomic status, demographic/geographic variables, and survey-level characteristics, to determine whether differences in survey methodology and the year of the survey can explain differences in the findings. The use of survey-level characteristics also allows us to determine the net effect of each of the other levels of variables when survey characteristics are controlled.

Note that our selection of independent variables for the regression analyses was guided by three considerations: (1) variables measured in all of the 22 community studies; (2) findings or discussion in previous research that indicated a potential relationship with the dependent variables; (3) collinearity statistics, which limited the inclusion of variables that were highly inter-correlated with each other.⁷

Note also that while we have made the Jewish identity factors the dependent variables, we do not necessarily posit causality—people with certain types of Jewish identity may seek communities with certain infrastructures, and/or a community’s infrastructure may influence certain aspects of Jewish identity. The present data do not enable us to disentangle these possibilities.

Indicators

Independent Variables

Community-Level Indicators

The indicators of community context are a mixture of aggregate individual-level data (e.g., number of intermarried couples), institutional-level data (e.g., number of synagogues and size of Jewish Federation annual campaign per household) and broader population data (e.g., size of the broader population, proportion of Jews in the broader population, and proportion in the broader population with no professed religion). Indicators of a Jewish community’s infrastructure included in this analysis derive from both the telephone survey and the Jewish Institutions Survey conducted in each community and reported in the Main Report for each community (available at www.jewishdatabank.org).⁸ In addition, Laurence Kotler-Berkowitz of the Research Department of the Jewish Federations of North America provided recent annual campaign data for each community (also reproduced in Sheskin 2011). Table 2 presents the variables that we use to indicate Jewish community characteristics. While other variables are available for describing Jewish community characteristics, not all could be included because of multicollinearity in the multiple regression analysis. More details on these variables can be found in Hartman and Sheskin (2011).

Table 2 Jewish community infrastructure by community of study

Size of Jewish Community The number of Jews and the number of Jewish households in a community can both be used to measure Jewish community size. These numbers were generally estimated from the first 10,000 RDD telephone numbers used in each survey, using a standard methodology applied in many Jewish community studies. If, for example, the 10,000 RDD telephone numbers yield responses from 4,000 households and 400 of these households are Jewish, 10 % (400/4,000) of households can be estimated to be Jewish.

The number of Jews for the 22 Jewish communities included in this analysis ranges from about 8,000 Jews in Lehigh Valley to over 215,000 Jews in Washington, DC. The number of Jewish households varies from 4,000 households in the Lehigh Valley to 110,000 households in Washington, DC. Number of Jewish households (rather than population size or the log of population size) was selected for use in the four multiple regression models.

The percent Jewish for the 22 Jewish communities was calculated using the above estimate for size of Jewish community, divided by the estimated number of households in the geographic area provided by the US Census (www.census.gov). For the 22 Jewish communities included in this analysis, percent Jewish varies from less than 1 % in San Antonio to nearly 40 % in South Palm Beach, although most communities have values below 10 %.

Stability of Jewish Population Two variables were considered as indicators of the stability of a Jewish community⁹: whether synagogue membership had (1) increased or (2) decreased by more than 20 % over the past 5–10 years, based on data from each study’s Jewish Institutions Survey. (In most communities, the data were available for the past 10 years, but in some communities only a five-year or seven-year change was available.) Of the 22 communities, nine show increased synagogue membership, six show decreased membership, and six show no significant change.

Density of Jewish Population Each respondent was asked their zip code. In each community, the percentage of all Jewish households who live in the three zip code areas containing the greatest percentage of Jewish households was calculated. The percentage of households in the top three zip codes for the 22 Jewish communities included in this analysis varies from 13 % in Washington DC to 66 % in Middlesex and Westport.

Number of Orthodox Synagogues The Jewish Institutions Survey in each community provided these data. The number of Orthodox synagogues for the 22 Jewish communities included in this analysis ranges from one in Sarasota to 25 in Detroit. Note that we also analyzed the impact of the total number of synagogues in the area, but it was too highly correlated with the size of Jewish population and the size of Jewish Federation annual campaign to include in the four regression analyses.

Size of Jewish Federation Annual Campaign We have employed the amounts from the latest (2009) Jewish Federation annual campaign data¹⁰ for each community (also reproduced in Sheskin 2011), rather than the amount of the annual campaign from the year of each community study, so as to minimize fluctuations due to variations in the general economic climate over the decade. By dividing by the number of Jewish households, we have in effect standardized the comparison, so that the size of the annual campaign is not simply reflective of the size of the community but rather the mobilization of resources from each Jewish household (on average). The size of the 2009 Jewish Federation annual campaign for the 22 Jewish communities varies from $436,000 in Portland (ME) to $31,671,000 in Detroit; per household the amount varies from $79 in Atlantic County to $1,066 in Detroit.

The percentage of married couples who are intermarried for the 22 Jewish communities varies from 9 % in South Palm Beach to 61 % in Portland (ME). An intermarriage was defined as a marriage in which one spouse was born Jewish and currently considers himself/herself Jewish and the other spouse was not born Jewish and does not currently consider himself/herself Jewish.

Dependent Variables: Jewish Identity Indicators

Ethnic or religious identity is usually considered part of an individual’s self-concept in terms of group belonging and the value and emotional significance attached to it (Rebhun 2004). Local Jewish community studies do not usually query an individual’s self-concept, but rather ask about several behavioral expressions of that identity. Therefore, our indicators of Jewish identity primarily represent behavioral practices or inclinations and expressions of Jewish identity in various ways, rather than notions of Jewish identity itself. Since many studies of Jewish identity conflate the self-concept and religious/ethnic practices or their value, using such indicators to indicate Jewish identity is not unusual. Starting from the premise that Jewish identity is multidimensional, and desiring to determine what dimensions of Jewish identity were expressed in this population (rather than a priori assuming what the construct of Jewish identity is and selecting indicators to reflect that construct), we performed a principal components factor analysis with Varimax rotation of the 17 variables expressing Jewish identity that were common to all 22 communities. The first column of Table 3 presents the means and standard deviations of all indicators used in the factor analysis.

Table 3 Jewish identity factors: Loadings from principal components analysis with Varimax rotation

Four factors emerged from the factor analysis of the 17 Jewish identity variables for all 22 communities combined (Table 3). Note that factor loadings of 0.45 or higher were considered when naming each factor. The four factors combined account for 57 % of the variance in the original 17 variables. In the following description of Table 3, the numbers in parentheses are factor loadings.

Factor 1 (Communal Religious Factor or “Ceremony”) reflects the religious identity of American Jews and includes the more common religious practices observed by many American Jews: lighting Chanukah candles (.828), participating in a Passover seder (.782), mezuzah on the front door (.615), attending synagogue services (.523), and synagogue membership (.496). This factor accounts for 15 % of the variance in the original 17 variables.

For the most part, the variables loading highly on Factor 1 are those which Alexander (1987, p. 124) referred to as “ceremony,” and “affirming membership in the social and cosmological order.” Sklare (1971) noted that these practices are generally compatible with American society in that they can be redefined in modern non-supra-mundane terms, do not demand social isolation or a unique lifestyle, provide a “Jewish” alternative when such is needed in the broader American religious scene, and are performed annually or infrequently. Hartman and Hartman (1996a, b, 2009) found a similar factor in their analyses of NJPS 1990, NJPS 2000–2001, and the 1991 New York Jewish Population Study.

This factor represents Jewish religious capital. Putnam (2000) distinguishes between bonding social capital, which is an investment in social networks of homogeneous groups of people, and bridging social capital, which is an investment in social networks of heterogeneous groups. Some of the variables that load highly on it may represent participation in activities with heterogeneous groups of Jews and even non-Jews (e.g., participating in a Passover seder and attending synagogue services), so this Communal Religious Factor may represent some degree of bridging capital as opposed to the more inward directed bonding capital which characterizes the Private Religious Factor (“Ritual”) (Factor 4 below). See Beyerlein and Hipp (2006) on the bridging effects of congregational involvement.

Factor 2 (Broader Ethnic Factor or “Klal Yisrael”) reflects the ethnic identity of American Jews: visits to Israel (.683), Jewish organization membership (.657), donation to the local Jewish Federation in the past year (.587), emotional attachment to Israel (.583), and donations to Jewish charities other than the local Jewish Federation in the past year (.572). This factor accounts for 14 % of the variance in the original 17 variables.

These variables include some of those more commonly associated with ethnic identity identified by Cohen (1983): attachment to Israel and attachment to non-synagogue Jewish institutions. An additional variable, namely donations to Jewish charities other than the Jewish Federation in the past year, which Cohen did not include in his ethnic dimension, loaded highly on this factor. Such donations do indicate an investment in Jewish social and cultural capital and, thus, can be viewed as part of ethnic identity.

Factor 3 (Local Ethnic Factor) reflects the ethnic identity of American Jews in terms of integration into the local community as regards non-religious or non-synagogue Jewish institutions: familiarity with the local Jewish Federation (.780), familiarity with the local Jewish Family Service (.772), being at least somewhat familiar with at least one local Jewish agency (.754), and participating in or attending any program at, or sponsored by, the local Jewish Community Center in the past year (.489). Note that only Jewish institutions common to all communities were included in this factor. This factor accounts for 14 % of the variance in the original 17 variables.

Factor 4 (Private Religious Factor or “Ritual”) reflects the religious identity of American Jews, including variables related to “personal rituals” that reveal stricter, daily, and personal commitment to ritual: keeping kosher outside the home (.879), keeping kosher inside the home (.865), and lighting Friday night candles (.584). Hartman and Hartman (2009) found a similar factor in their analyses of NJPS 1990 and NJPS 2000–2001 and the 1991 New York Jewish Population Survey. Note that it represents religious capital that would bond together similarly observant Jews (“bonding” religious capital).

To validate the structure of Jewish identity found in the Total Factor Analysis (TFA) just described, factor analyses were run separately for each of the 22 communities. The same 17 variables were employed and a four-factor solution requested. The percentage of variation explained by the four-factor solution ranges from 52 to 60 %, meaning this factor analysis is acceptable in each community and the percentage of variance explained in each of the separate factor analyses is quite similar to the total (57 %). Some of the variation by community included:

• (a) The order of the dimensions, indicating that certain types of Jewish identity are more central in some communities than others. For example, in Bergen County, Detroit, Miami, and Middlesex, the most prominent factor (i.e., the factor explaining the most variation in these variables) was the private religious factor rather than the communal religious factor, reflecting the higher percentage of Orthodox in these communities. However, since the four factors in the TFA contribute roughly equal portions to the 57 % total variance explained, these variations in order are of interest but do not invalidate the use of the TFA.

• (b) In the TFA, synagogue membership (which can function both religiously and ethnically) loads more highly on the communal religious factor (.496) than on the broader ethnic factor (.352). The same pattern is found in most of the communities. However, in a few, synagogue membership loads more highly on the broader ethnic factor; in some instances, it also loads highly on the private religious factor or the local ethnic factor. Likewise, in the TFA, attending synagogue services (which is important for ritual observance) loads more highly on the communal religious factor (.523) than the private religious factor (.450). These variations reflect, of course, the multifaceted functions of synagogues (Wertheimer 2005), as well as the extent to which synagogues can vary across communities. In communities in which the communal religious factor is dominant, synagogue service attendance loads higher on this factor, while in communities where the private religious factor is dominant, synagogue service attendance loads more highly on that factor.

• (c) In the TFA, lighting Friday night candles loads more highly on the private religious factor (.584) than on the communal religious factor (.437). However, in some communities, it loads more highly on the communal religious factor than on the private religious factor. We believe this finding suggests that lighting Friday night candles may be more normative in some communities and less of a private ritual than in other communities.

• (d) In the TFA, donated to the local Jewish Federation in the past year loads more highly on the broader ethnic factor (.587) than on the local ethnic factor (.313). In some communities (e.g., Lehigh Valley, Minneapolis, St. Paul, and San Antonio), it loads more highly on the local ethnic factor than on the broader ethnic factor. In these communities, donations may be directed more to the local community than nationally and to Israel, or may reflect more about integration into the local community than commitment to the broader Jewish peoplehood.

Further analyses of these geographic variations are reserved for future research. We concluded that the similarities in the structure between the individual communities and the total sample were strong enough to warrant using the TFA for the total sample.

Control Variables

The means and standard deviations of all independent variables are presented in Table 4.

Table 4 Means and standard deviations of independent variables used in the analysis

Individual-Level Variables

We controlled for four types of individual-level variables that were found to be related to Jewish identity in previous studies. In some cases (e.g., length of residence vs. population stability), they enable us to determine whether it is the individual-level variable or the community characteristic that influences individual Jewish identity. Variables related to the individual are examined in four different groups: (1) Jewish background and connections, (2) family status, (3) socioeconomic status, and (4) demographic/geographic variables.¹¹

Jewish Background and Connections Variables

Denominational groups (whether self-identified or affiliated) differ in: (1) the extent to which adherents see religion as a dominant influence on daily behaviors, (2) their emphasis on religious and/or ethnic identity, and (3) the extent to which they value exposure and involvement in the broader (non-exclusively) Jewish world or universalism (Hartman and Hartman 2001). Thus, including denominational self-identification in the multiple regression models controls for this source of considerable variation in Jewish identity, so that findings related to other variables are net of this effect.

Jewish denominational self-identification is measured by responses to the question, “Do you consider yourself Orthodox, Conservative, Reconstructionist, Reform, or Just Jewish?” Because of the small number of Reconstructionist individuals (about 1 % of the sample), they were combined with Reform for the purposes of the analysis. In our sample, about 5 % self-identified as Orthodox, 31 % as Conservative, 33 % as Reform or Reconstructionist, and 32 % as Just Jewish. In contrast, in NJPS 2000–2001, 8 % self-identified as Orthodox, 25 % as Conservative, 37 % as Reform or Reconstructionist, and 30 % as Just Jewish (Sheskin 2011, Chap. 6). Although our sample somewhat under-represents the Orthodox compared to NJPS 2000–2001 (mostly because New York is not one of the 22 communities), the sample size of Orthodox Jews is still about 1,000.

We had several indicators of Jewish experiences as a child or young adult, including attendance at a Jewish day school, attendance at a Jewish supplementary school, attendance at a Jewish overnight camp, regular participation in a Jewish youth group as a teenager, and regular participation in Hillel while in college. The relationships between Jewish education, both formal and informal, and adult Jewish behaviors such as ritual observance, synagogue membership, involvement in Jewish organizations, in-marriage, and opposition to children’s intermarriage are well established (e.g., Baker and Ukeles 1994; Bock 1977; Bubis and Marks 1975; Cnaan 1993; Cohen 1995, 2000, 2007; Cohen and Ganapol 1998; Cohen and Kotler-Berkowitz 2004; Cohen et al. 2011; Dashefsky and Lebson 2002; Fishman and Goldstein 1993; Goldstein 1993; Hartman and Hartman 2003; Himmelfarb 1974; Horowitz 2001; Keysar and Kosmin 2001, 2005; Schiff and Schneider 1994; Sheskin 1997, 2009, 2010c; Waxman 2003).

The high percentage of intermarried couples (about 30 %) and respondents who identify as “Just Jewish” (about 30 %; not necessarily the same 30 %) raises the question of whether their children are being raised Jewish. Respondents were therefore explicitly asked whether the children in their household were being raised Jewish. Including this variable allows us to determine whether raising children Jewish has an effect on Jewish identity (or reflects Jewish identity) over and above whether the respondent is intermarried or identifies with a particular denomination. Because having Jewish children often brings families into contact with other Jewish families, whether for Jewish education or participation in synagogues or the Jewish Community Center, we expect that having Jewish children will be associated with stronger Jewish identity. (Of course, it could also be that those with stronger Jewish identity are more likely to raise their children Jewish. With the current data set, this cannot be deconstructed.)

As mentioned above, denser concentrations of Jews are usually associated with a more developed Jewish infrastructure. However, not everyone in a community lives in the areas of densest Jewish population. We therefore coded each respondent as to whether they lived in one of the three zip code areas with the greatest number of Jewish households in the community. We expect that those living in the densest Jewish areas have heightened opportunities to interact with the Jewish community, which will reinforce commitment to the community and reflect or result in stronger Jewish identity.

Also as previously mentioned, intermarriage to a non-Jew presupposes at least some investment of time and energy in “bridging” social and ethnic capital between Jews and non-Jews (as opposed to “bonding” capital within the Jewish community). Cohen (2006) has shown the divergence in Jewish identity and practice between the intermarried and the in-married. As a result, we expect intermarriage to be associated with weaker Jewish identity in all respects, especially when other variables are held constant, and it was therefore introduced as a control variable.

Family Status Variables

Marital status and household structure have been related to the strength of Jewish identity. Because Judaism is constructed around family ritual, and Jewish communal events are often organized around family participation, especially inclusive of school-age (or younger) children (Cohen and Ritterband 1988), being married reinforces Jewish identity, particularly in its communal expressions. Previous research has found that married couples often are more affiliated with Jewish organizations and synagogues, especially in relation to activities for children. They tend to be more active if they have children enrolled in Jewish education (Cohen and Ritterband 1988 as cited in Liebman and Cohen 1990). Having more children age 12 or younger is therefore also related to greater Jewish involvement in the community. At the same time, we recognize that having young children increases the domestic and financial responsibilities of adults, which may actually hinder communal participation. Larger household size similarly results in increased domestic roles, which may hinder participation in activities related to the communal religious and local ethnic factors, particularly the latter. This would be in addition to the greater financial burden, which would be controlled by the income variable (discussed below).

On the other hand, singles (especially if they are older) sometimes feel alienated by the Jewish community because of this familial orientation (Diamant 1989; Fishman 1993; Hartman and Hartman 2009; Schneider 1984). Divorce is not generally a stigma in the Jewish community. However, it may broaden the “bridging” social capital of the individual, and in that sense weaken the “bonding” Jewish capital, and with it, Jewish identity. Divorced individuals may have more difficulty finding their place in the Jewish community and may also have broadened social contacts in their search for second (or higher) mates. The measure used is whether the respondent has ever divorced.

Socioeconomic Status Variables

Many sociological studies of religion find that higher levels of secular education are associated with lower levels of religiosity (Beyerlein 2004; Darnell and Sherkat 1997; Massengill 2008). Past studies of American Jews, however, have shown an overall positive relationship between secular education and Jewish identity, which has been extensively studied (e.g., Hartman and Hartman 1996a, b, 2009, 2011; Lehrer 2009). Higher education, of course, generally leads to higher income and given the high cost of living Jewishly (Chiswick and Chiswick 2000; Wertheimer 2010), one would expect stronger Jewish identity among respondents with higher levels of education and income. Sheskin (2004, Table 7-1) shows a strong relationship between household income and synagogue membership, for example, in almost all 22 communities in our sample.

However, too much investment in particularistic Jewish cultural and social capital appears to have a negative impact on secular achievement (Chiswick and Huang 2008; Hurst and Mott 2006), resulting in the Orthodox having a somewhat lower level of education than the other denominations. Hence, an inverse relationship between higher education and the private religious factor has been found in the past (Hartman and Hartman 2009) and is also consistent with the work of Schieman (2010), who found that socioeconomic status is negatively related to beliefs in divine involvement and control.

Structural location theory (de Vaus and McAllister 1987) hypothesizes that the greater the investment in the secular infrastructure (e.g., participating in the labor force), the lower the investment in the religious infrastructure and hence lower levels of religiosity should be evident. Investment in the labor force may also be seen as an investment in “bridging” social capital, which may be opposed to particularistic religious or ethnic identity. However, as regards Jewish women, religious and ethnic identity have been found to be related positively to labor force involvement, especially once family status and number of children have been controlled (Hartman and Hartman 2011). By controlling for this variable, we control for the sources of variation it introduces.

Education is measured in this study by highest degree attained. Labor force participation is measured by whether respondents are currently employed. We have two indicators of socioeconomic status: household income and housing value.

Demographic/Geographic Variables

Gender: Women have been found to be more religiously and ethnically identified, both among Jews (Hartman and Hartman 2009) and more generally (e.g., Collett and Lizardo 2009; Hertel 1995; Mueller and Johnson 1975; Putnam and Campbell 2010; Rayburn 2004; Stark 2002; Walter and Davie 1998; Weber 1963 [1922]; Woolever et al. 2006). The reasons for this have been explained in some detail in the literature cited. For our purposes, we include the variable to reinforce previous findings as well as to control for any variation in Jewish identity that gender may introduce.

Age: Putnam and Campbell (2010) find a clear positive relationship between age and religiosity. However, the relationship between age and Jewish identity has been found to be more complex. While older Jews tend to be more involved in activities related to the communal religious factor and local ethnic factor, younger Jews tend to be more involved in private expressions of identity (Hartman and Hartman 2009). On the other hand, Sered (1992) found personal religiosity among Jews to increase with age (at least among women). In other respects, the Baby Boom generation is less (or more) Jewishly identified than younger or older cohorts (Hartman and Hartman 2009, pp. 138–140; Waxman 2001). Age is therefore introduced as a control variable, and it reflects cohort (year of birth) as well as life cycle stage.

Immigrant Status: While one might expect international immigrants to be less integrated into the local community, Goldstein and Goldstein (1996) found that they were in fact often more integrated because of their heightened need for local services. In particular, Jews from the former Soviet Union received significant assistance from Jewish Family Services throughout the country, and, thus, became aware of the local Jewish community.

Length of Residence in the Community: Rebhun (1995) examined whether in-migrants to a community had stronger or weaker Jewish identity and found that mobility had a small negative effect on Jewish identity after controlling for major socio-demographic variables. This supported the hypothesis that migration disrupts, rather than enhances, Jewish identity. However, Rebhun’s indicator of Jewish identity did not distinguish the different dimensions of Jewish identity (rather, it conflated them into one indicator). Further, Rebhun’s analysis only reflected the three large communities of Boston, Los Angeles, and Philadelphia; whether this holds true for other communities will require the empirical testing we do here.

Goldstein and Goldstein (1996, p. 206), analyzing NJPS 1990, found a “holding power of traditionalism,” that is, the most traditional Jews (including Orthodox) were less likely to be mobile. The Goldsteins also found that Jews new to a community may be less affiliated with Jewish institutions (unless they are international immigrants in need of services the institutions offer). Sheskin shows a positive relationship between length of residence and affiliation with Jewish institutions (e.g., Sheskin 2009, Table 7-1). Rebhun’s (2011) latest analysis, using NJPS 1990 and NJPS 2000–2001, found that migration had little net effect on Jewish identity, and he predicted that the effect of migration would continue to weaken.

Controlling for length of residence in the community therefore controls for these various sources of influence on Jewish identity.

Intentions to Move: We also control for intentions to move from the community, as respondents who expect to leave a community are likely to be pulling up roots, and therefore may be less involved in their local community.

Survey-Level Variables

Three survey-level indicators were selected to examine whether the manner in which the data were collected had a significant effect on the findings: the extent of cooperation with the survey once it was determined that a household was Jewish (the survey cooperation rate), the sample size, and the year of the field work.

We expected that the relative proportions of the sample generated by RDD and DJN sampling techniques might be related to Jewish identity (a higher percent of DJN being associated with stronger Jewish identity, per the rationale presented in footnote 5). The variation among the communities is relatively small. Most of the communities have over 60 % DJN and four of the communities have 100 % RDD samples (Table 1), three of which are in Florida. We mention this as a source of variation in case additional community studies are added in the future when the influence of this variable can be tested more adequately.

Some recent literature suggests that some aspects of Jewish identity, such as informal social ties, have been weakening over time although, in other respects, Jewish identity, such as Jews’ ties to communal institutions and activities, may be strengthening (Cohen 2003; Rebhun 2011). Thus, we have controlled for the year of the field work, although we did not expect that the ten-year span would make a significant difference in Jewish identity.

Multiple Regression Results

Four multiple regression models were developed, one for each Jewish identity factor. The dependent variables were the factor scores on each of the four factors defined in Table 3. The variables were entered into the model in groups to identify the respective contributions of these groups. First we entered each group of variables separately to identify their respective gross contributions to Jewish identity. Table 5 presents the R ² for each of these separate regression analyses. Here we see some interesting differences between the indicators of religious identity and those of ethnic identity.

Table 5 Regression analyses for the four Jewish identity factors

For the two factors of religious identity, individual-level characteristics of Jewish background/connection at 27 and 33 % were, by far, more important than any other group of variables. Other individual-level characteristics explained much less of the variance in the religious identity factors, but for the most part explained more than the community-level characteristics.

For the two ethnic factors, a similar pattern was observed, but the individual characteristics not specifically related to Jewish background/connection were relatively more important than they had been for the religious identity factors. For example, demographic/geographic variables explained less than 3 % of each of the two religious identity factors, but a gross 7.0 % of the broader ethnic factor, and 12.2 % of the local ethnic factor.

The community-level characteristics explained more variance in the ethnic identity factors than in the religious identity factors: the Jewish community infrastructure/context variables explained 12.8 % of the variance in the local ethnic factor and 10.9 % of the variance in the broader ethnic factor. Only for the local ethnic factor did the Jewish community characteristics explain more of the variance than any group of individual-level variables.

Finally, survey-level variables explained only 3 % or less of the variance in any of the identity factors, reinforcing the validity of the meta-file.

One of the purposes of these separate regressions for each group of variables was to eliminate variables that had no relationship with individual Jewish identity. However, in each group of variables, each of the independent variables was significant at p < .05 for at least one factor of Jewish identity, so all were retained for the next stage of the analysis.

The regressions above enabled gross comparisons. That is, we have examined each group of variables individually, without holding the other groups of variables constant. When we combined all variables in stepwise multiple regressions, the analysis was clarified somewhat by using three models.

In Model 1 (Table 6), we entered the Jewish community/infrastructure context variables only to discern the extent to which various aspects of Jewish identity are related to Jewish community infrastructure/context.

Table 6 Multiple regression analysis of Jewish identity factors showing standardized regression coefficients, βs (unstandardized regression coefficients, Bs, in parentheses)*

In Model 2, we added the various individual-level variables, including Jewish background/connection, family status, socioeconomic status, and demographic/geographic variables.

In Model 3, we added survey-level variables, both to control for variation in survey administration and to validate the idea of aggregating the various surveys into a meta-data file.

At the bottom of Table 6, the cumulative R ² at each stage of analysis is presented. Because all variables have been entered into the multiple regressions at the final stage, the regression coefficients in Model 3 reflect the net effects of each of the variables once all other variables have been controlled. When considering the strength of the relationship of a particular variable across factors, we examine the unstandardized coefficients (B); when comparing the relative importance of a variable’s relationship with a particular factor, we look at the standardized regression coefficients (β). For clarity, only coefficients with a statistical significance of at least p < .05 are presented in Table 6.¹²

Looking first at the regression coefficients for the community-level variables in Model 1, we compare which community characteristics are related to each of the four expressions of Jewish identity, and the relative strengths of the relationships (using the unstandardized coefficients—Bs—in parentheses). While four and five community-level characteristics are related to the religious Jewish identity factors, six characteristics are related significantly to broader ethnic identity and seven characteristics to local ethnic identity. None of the regression coefficients are as large for the religious identity factors as they are for the two ethnic identity factors. This is reflected in the R ² for Model 1, which indicates that only 3 % of the variance (or less) is explained by these community-level characteristics in the two religious identity factors, while more than 10 % of the variance is explained by these characteristics in the two ethnic identity factors (10.9 and 12.8 %, respectively). In particular, the percent Jewish, Jewish density (% of Jewish households in the top three zips), and community stability (increasing or decreasing synagogue membership) are most strongly related to the ethnic factors. Note that community instability (increasing or decreasing synagogue membership) is negatively related to local community integration, with β = −.104 and −.108, respectively, suggesting that stability promotes greater local ethnic identity.

Once individual-level variables and survey-level variables are introduced into the regression models (Model 3), most of the community-level variables lose the statistical significance of their relationship with religious identity and broader ethnic identity. However, seven characteristics retain statistically significant relationships with local ethnic identity even in Model 3, and a comparison of unstandardized regression coefficients over Models 1–3 for local ethnic identity shows that most of the relationships retain at least the same strength as they had independently in Model 1. In addition to percent Jewish, Jewish density, and community stability, smaller Jewish population, larger annual campaign per household, and fewer Orthodox synagogues are all related to stronger local ethnic identity. Smaller Jewish populations, less dense Jewish communities, and stronger Federation campaigns (per household) are all related more strongly to local ethnic identity than any of the other variables except length of residence in the community.

Model 3 shows which of the individual-level indicators are related to each of the Jewish identity factors, even when community-level and survey-level variables are held constant.

For the communal religious factor (Ceremony), being Conservative or Reform/Reconstructionist have the strongest relationships (β = .379 and .305, respectively), followed by being Orthodox (β = .119); that is, denominational preference (of any kind) has the strongest relationship with expressing Jewish identity in a communal religious way, possibly because some of the communal celebrations take place in synagogues or with other congregants. Three other individual-level variables have particularly strong relationships with this expression of Jewish identity: being intermarried is negatively related (β = −.197), as is being single, never married (β = −.132), while having a larger household is related positively to communal religious identity. Note that being intermarried has a stronger negative relationship with communal religious identity than does the proportion intermarried in the community (β = −.098).

For the private religious factor (Ritual), being Orthodox has a β of .479, nearly three times the strength of any other coefficient, and being Conservative has the next strongest relationship (β = .163). Being Reform/Reconstructionist has a negative relationship (β = −.062) with the private religious factor. Given the correlation of this factor with Orthodox identification, it is not surprising that we see a significant relationship with having attended a Jewish day school as a child (β = .072) and household size (β = .057). (Typically, Orthodox Jews have larger households than other Jewish households.)

For the broader ethnic factor (“Klal Yisrael”), the strongest relationships to Jewish identity are being Conservative (β = .129), not being intermarried (β = −.157), and being older (β = .258). Jewish community context has minimal relationship to this type of identity, once individual-level and survey-level indicators have been controlled. Note that the communal percentage of intermarried couples retains a statistically significant (small, positive) effect (β = −.049) in Model 3.

For the local ethnic factor, we see the strongest relationships with Jewish community context, as noted above. The strongest individual-level relationship is with length of residence in the community (β = .267); it is noteworthy that the individual variable of length of residence is associated more strongly with local ethnic identity than the community-level characteristic of community stability. Higher education is also related to stronger local ethnic integration (β = .116), as is self-identifying as Conservative (β = .119).

As indicated above, the survey-level variables contribute little explanation to the four Jewish identity factors as a group (Table 5 shows R ² less than 3 % when they are entered into the regression equations by themselves, for each of the four identity factors), which reinforces the validity of using a meta-data file for this type of analysis. Adding in the survey-level indicators to the stepwise regression model in Table 6 (comparing Models 2 and 3) makes little difference for the two religious identity factors, although the survey indicators weaken the effect of the community-level indicators in both cases. While some of the survey-level variables retain statistically significant relationships with religious identity even in Model 3, they are relatively weak relationships (beta coefficients <.080). The survey-level indicators make little difference to local ethnic identity, either.

However, the survey-level indicators add nearly 6 % to the explanation of variance in the broader ethnic factor. In particular, in communities where the survey cooperation rate is higher (β = .360) and the sample size larger (β = .122), broader ethnic identity is stronger. We suggest that these are indicators of stronger communal integration and therefore replace the more indirect measures of number of Jewish households and their density. Annual campaign per household remains statistically significant in Model 3, as does increasing synagogue membership, both signs of a vibrant Jewish community. In other words, the survey-level characteristics seem to be acting more as community indicators than methodological controls.

Given the idiosyncratic nature of some of the community studies (as noted in the Main Reports of the community studies), some of the variation in survey methodology is intricately related to particular communities and therefore may indeed be less indicative of the survey methodology than of community characteristics. For example, the fieldwork for the Detroit, Washington, DC, and Las Vegas studies was accomplished using a professional market research firm whereas the 19 other studies were completed using (paid) members of the local Jewish community. Could the Las Vegas results be different because of this methodological difference? Perhaps. But, the Las Vegas results are probably different from the other communities because Las Vegas is simply a very different place, Jewishly and otherwise. Using the recently released New York community study (Cohen et al. 2012), Sheskin (2012) notes the similarities and differences of New York City from other communities (no other community being as large, diverse, or poor) and concludes that Jewish communities do vary. Thus, when we control for survey-level variables it is not clear, exactly, what we are controlling. The fact that the survey-level variables add little to the overall explanation in most of the identity factors is the more important result because it validates the concept of combining and analyzing the 22 community studies together.

Next, we consider each of our hypotheses, on the basis of Model 3, which shows the net effect of each of the variables when all variables have been controlled.

Hypothesis 1: (Jewish identity is impacted by the nature of the Jewish infrastructure/context and population where an individual resides) is only very partially supported, and, in fact, at times is related to Jewish identity in ways contrary to our expectations.

Hypothesis 1a: (The size of the Jewish community will impact Jewish identity: larger Jewish communities will act to strengthen ethnic identity, while smaller Jewish communities will exhibit stronger religious identity) is not supported. The number of Jewish households in the community has a significant relationship only with the local ethnic factor, but not in the hypothesized direction. That is, individuals in smaller Jewish communities tend to be more integrated into the local Jewish community, as indicated by the local ethnic factor, than are individuals in larger Jewish communities. Number of Jewish households is not related significantly to the other Jewish identity factors.

Hypothesis 1b: (A higher percent Jewish in a community will act to strengthen Jewish identity) is not supported. A higher percentage Jewish is negatively associated with private religious identity and broader ethnic identity and is not related significantly to the other two expressions of Jewish identity.

Hypothesis 1c: (Stable Jewish communities will be characterized by stronger Jewish identity) is supported with regard to the local ethnic factor, which shows a negative relationship with both increasing and decreasing synagogue membership. Broader ethnic identity, on the other hand, is positively associated with increasing synagogue membership, a sign of a vibrant Jewish community.

Hypothesis 1d: (Jewish populations that are clustered in one part of a metropolitan area, rather than being geographically dispersed throughout that metropolitan area, will exhibit stronger Jewish identity) is not supported. In fact, density is related negatively to the local ethnic factor, and is not related significantly to any of the other identity factors.

Hypothesis 1e: (Jewish communities with a higher percentage of married couples who are intermarried will be characterized by individuals with weaker Jewish identity) is partially supported by a significant negative relationship with communal religious identity. Of more importance, however, is whether the individual is intermarried. Surprisingly, the percentage intermarried in a community has a weak positive relationship with broader ethnic identity when other variables are controlled.

Hypothesis 2: (The more developed the Jewish institutions in a community, the stronger the individual Jewish identity) is partially supported.

Hypothesis 2a: (The greater the number of Orthodox synagogues in a community, the stronger the religious identity) is not supported. The number of Orthodox synagogues in a community has only one statistically significant relationship with the Jewish identity factors, and it is a negative relationship with local ethnic identity. The Orthodox community may be more identified with the local Orthodox community than the broader Jewish communal services and institutions.

Hypothesis 2b: (A larger Jewish Federation annual campaign per household will act to strengthen Jewish identity, especially the local ethnic factor) is partially supported. The size of the Jewish Federation annual campaign per household is related positively to the two ethnic factors, as expected. However, it is related negatively to the two religious identity factors. We considered that this might be related to Orthodox households contributing less to the local Jewish Federation. The Decade 2000 data show that Orthodox households are more likely to contribute to Jewish Federation annual campaigns than non-Orthodox households, but they tend to contribute smaller amounts due to lower incomes and their tendency to contribute to Orthodox-specific charities, such as Orthodox Jewish day schools.

In sum, most of our hypotheses about the relationships between Jewish community characteristics and Jewish identity were only partially supported or not confirmed, and in several cases were reverse significant. In fact, not for even one Jewish community-level variable are our hypotheses supported across all four factors. In some cases, controlling for the individual-level variables eliminated the significance of the community-level variable, suggesting that certain types of communities may attract households with characteristics that are related to different types of Jewish identity, and this may influence the nature of Jewish identity in that community, but it is not the community-level characteristics themselves that do so. Community-level variables retain their relationship with Jewish identity mainly on the local ethnic level.

Hypothesis 3: (The strength of the relationship between individual-level variables and individual Jewish identity will be impacted by community context) was supported for local ethnic identity, but not for the other three factors of Jewish identity. Table 7 shows the unstandardized regression coefficients of the individual-level variables when community-level variables are excluded from (Model A) and included in the regression equations (Model B). For the first three factors (the two religious identity factors and the broader ethnic identity factor) controlling for community-level characteristics hardly affects the importance of the individual-level variables and Jewish identity, be they denominational self-identification, Jewish background, family status, socioeconomic status, or demographic/geographic characteristics. Although for some of the variables the coefficients are smaller when community-level variables are controlled (Model A > Model B), the differences are small. In four cases, the unstandardized regression coefficients become slightly larger when community context is controlled, suggesting a very small effect of clarifying the effect of the individual-level variable (the negative effect of living in a household with Jewish children on broad ethnic identity, the positive effect of household income on broad ethnic identity, and the negative relationship of length of residence in a community on private religious identity and broad ethnic identity).

Table 7 Unstandardized regression coefficients for individual-level variables in two multiple regression models of Jewish identity factors* (Model A does not include Jewish community-level variables; Model B does)**

There are more changes in the relationship of individual-level variables to local ethnic identity when the community-level variables are controlled. In 11 of the 25 cases, the unstandardized regression coefficients of the individual-level variables are smaller when community context is controlled. Thus, the relationship of being Orthodox, Conservative, or Reform, having gone to a Jewish day school or participated in a Jewish youth group, currently being employed, and length of residence in the community all are reduced when community-level indicators enter the regression equation, suggesting that part of their effect is the kind of communities these people (choose to) live in. On the other hand, in six of the cases, the relationship of the individual-level variable to local community integration is strengthened when the community context is controlled. The relationship of local community integration to three variables becomes statistically significant, once community context is controlled: household size, housing value, and intent to move. Some of the relationship with the local ethnic factor was thus camouflaged when community-level indicators were not held constant. While none of the changes are extremely large, this pattern reinforces what we found above, that community context is especially important in relation to local ethnic identity, as compared to its importance in relation to the other identity factors.

Conclusions

While the importance of place in both religious and ethnic identity has long been recognized by geographers (Frazier and Tettey-Fio 2010; Stump 2008), it has not been researched systematically regarding Jewish identity. Our research may be viewed as the beginning of a systematic “sociology of Jewish place” (Horowitz 1999). In this concluding discussion, we will focus on what we have learned about the structure of Jewish identity and how it varies across communities; the relationship of community characteristics to various types of Jewish identity; implications for community leaders; and implications for future research on Jewish identity.

The Structure of Jewish Identity

The first important result is that we are able to identify the same structure of Jewish identity across all 22 communities. These factors reinforce previous multidimensional studies of Jewish identity (e.g., Cohen 2009; Glenn and Sokoloff 2010; Hartman and Hartman 1996a, 2001, 2009; Phillips 1991; Rebhun 2004; Sharot 2011). The expression of religious identity is shown to be separate from the expression of ethnic identity. Religious identity, in turn, can be divided into a communal religious factor, which includes the more commonly observed ritual expressions of Jewish identity, and the private religious factor, which includes the less commonly observed expressions that entail a private commitment to daily ritual. Likewise, ethnic identity can be divided into a broader ethnic factor, which reflects Jewish “secular” culture and peoplehood (“klal yisrael”), and a factor of local ethnicity, which includes familiarity with and participation in the local Jewish Federation and its agencies. By replicating the four-factor structure of Jewish identity for each of the 22 communities individually, we reinforced the validity of the meta-analysis.

Along with this predominant similarity between communities, our analysis also highlighted variation in this structure across communities, an area that could benefit from further research. For example, in communities where synagogue membership loads more highly on the broader ethnic factor rather than the communal religious factor, is individual religious identity stronger on average? In communities where contributions to the Jewish Federation load more highly on the local ethnic factor rather than the broader ethnic factor, is broader ethnic identity weaker on average? This might be another avenue for investigating the impact of community context on individual Jewish identity.

Previous studies of Jewish identity have identified public and private dimensions of both religious and ethnic identity. Because there were few if any indicators in the 22 questionnaires concerning private ethnic identity, no separate factor of private ethnicity emerged. This finding reinforces the desirability of standardizing local Jewish community study questionnaires and including questions based upon this theoretical framework, even if it is not in the immediate interest of a given community. Only by some centralized subsidization of the cost of adding such “non-practical” questions to local community studies, could individual communities perhaps be enticed to cooperate with such a venture. It certainly is worth exploring, although most local community studies already have lengthy questionnaires and adding such material might mean deleting questions that local Jewish Federations would consider more important. These 22 local Jewish community studies benefitted from the fact that all 22 questionnaires were designed by the same researcher and only minor wording changes occurred on a few questions. More standardization between researchers would make analyses such as the one presented here much more possible and worthwhile.

The Relationship of Community Characteristics to Individual Jewish Identity

Besides increasing sample size (in this case, to almost 20,000) and achieving significant representation across various Jewish settings, an important innovation in using these 22 community studies in the Decade 2000 Data Set is that we have much more detail about the Jewish infrastructure in the community of each respondent than has been the case in any analysis of local or National Jewish Population Surveys in the past. Almost all of the infrastructure variables in Table 2 were collected as part of the original studies and were included in the study reports. This practice facilitated addressing the question of the extent to which Jewish community infrastructure is related to an individual’s expressions of Jewish identity. The answer is that Jewish community infrastructure is related to Jewish identity primarily in terms of local ethnic identity; its contributions to the other three factors of Jewish identity (communal religious identity, private religious identity, and broader ethnic identity) are minor, once we control for individual-level and survey-level characteristics.

One question that arises from this finding is what the role of local community ethnicity is in reinforcing individual Jewish identity of various sorts; in other words, looking at local community ethnicity as a means to an end, rather than an end in itself. An even bigger issue is how the broader infrastructure of a community matters to an individual. Perhaps only in the actual interaction with communal institutions (appearing in the “familiarity with” and “participation in” questions loading highly on the local ethnic factor) is Jewish identity reflected. Nevertheless, certain features of the broader Jewish context would seem to be necessary (a minimal “critical mass” of Jews, having synagogues in proximity, perhaps kosher eating establishments, some Jewish services) to reinforce Jewish identity.

So perhaps the important question is: which of the features in the Jewish communal context are important and how are they important? The minor effect of community context on most of the expressions of individual Jewish identity suggests that the most important characteristics of the communal context may not be included in our analysis, and may even be characteristics for which we do not usually collect data. Similarly, these community studies do not have some of the more familiar attitudinal indicators usually asked to determine strength of Jewish identity (such as “How important is being Jewish in your life?” and extent of agreement with the statement, “Jews in the United States and Jews elsewhere around the world share a common destiny”) and social—but impractical—indicators (such as “How many of the people you consider to be your closest friends are Jewish?” and extent of agreement with the statement, “I have a special responsibility to take care of Jews in need around the world”). Such limitations to the data may minimize the relationship that is found. Again, we emphasize the desirability of standardizing a portion of the community studies’ questionnaires to facilitate comparability between researchers and inclusion of more communities in a meta-data file such as Decade 2000. Particularly when national studies of American Jews are not being funded, the ability to aggregate local Jewish community studies is imperative, and subsidizing a common core of questions for community studies would be far less costly than developing a new national Jewish population study.

Despite their weak contribution to the explanation of three of the factors of Jewish identity, features of Jewish infrastructure are significantly related to integration into the local Jewish community (the local ethnic factor). As mentioned above, with the present data we cannot determine causality—people with certain types of Jewish identity may seek communities with certain infrastructures, and/or a community’s infrastructure may influence certain aspects of Jewish identity. But the relationships themselves are notable. Integration into the local community is more likely in smaller Jewish communities, which are not particularly dense and which tend to have stable Jewish populations. Where there is stronger integration into the local community, individuals in the community are more likely to be mobilized to contribute to the Jewish Federation annual campaign.

It is possible that there are other features of Jewish communities not measured by these surveys (e.g., number of actual cultural activities offered), which further research may uncover. It is also possible that characteristics of virtual communities are related to Jewish identity. The latest New York community study (Cohen 2012) notes that more than a third of New York Jews access Jewish websites, for example, and 16 % claim to belong to online Jewish groups (p. 114). Again, further research is needed to explore this.

Implications for Community Leaders

Local community studies are usually undertaken to provide practical applications for the local community. It is worthwhile to consider whether this aggregate analysis of community studies has any practical application for local community leaders. First, it is important to note that strong Jewish identity is found in many types of communities. Some community features are not related to most expressions of Jewish identity, but are positively related to one another. Clearly, a Jewish community being larger and more developed is not a clear recipe for stronger Jewish identity in all of its aspects, nor is Jewish population density or mobilizing a strong Jewish Federation annual campaign. Strong Jewish identity can be found in stable Jewish communities and not-so-stable Jewish communities. The options seem to be limitless. As Goldscheider (1986) once suggested, Jews will find a way to create the social and institutional networks that work for them, if they do not already exist; they are not limited to a particular recipe for communal structure.

The stronger relationship of smaller, stable Jewish communities to local ethnic identity may, however, be instructive. Even in larger communities, if smaller units of close interaction can be simulated (e.g., in learning havurahs), a similar effect of stronger local ethnic identity may be achieved.

Another application of our research may be derived from the persisting importance of individual-level characteristics for Jewish identity. Controlling for community-level characteristics did not erase the significance of the more commonly researched relationships between individual-level variables and Jewish identity, be they denominational self-identification, Jewish background, family status, socioeconomic status, or demographic characteristics. It does become clear from the results, however, that some types of individual‐level characteristics are related to certain expressions of Jewish identity, while other types have stronger relationships to other expressions; further, some of the relationships between individual‐level characteristics and expressions of Jewish identity are positive with one factor of Jewish identity, and negative with another. For example, Jewish background and denominational preference are much more closely related to religious identity, while the socioeconomic characteristics of education and income are more related to broader ethnic orientation among Jews. Awareness of individual characteristics of residents may be an important variable in tailoring community activities to a particular population’s needs. Younger Jews may be attracted to communal religious activities, while older Jews respond to activities tapping broader ethnic identity.

While family status is only weakly related to most of the expressions of Jewish identity, being single seems to be associated with weaker communal religious identity more than other expressions of Jewish identity. Community leaders might be sensitive to this and attempt to design community religious activities that are inclusive of those in their community who are not married.

Implications for Future Research

Our findings suggest that analyses of religious and ethnic expressions of Jewish identity should not be lumped together, and that some of the contradictory emphases in different research studies of Jewish identity (see Hartman, forthcoming) may be linked to foci on different types of Jewish identity rather than Jewish identity in its multifaceted complexity.

We should note that while the 22 community studies in the Decade 2000 Data Set represent a wide range of Jewish community settings, they are not a “representative” sample of American Jewish communities. We lack representation of the “Pacific Northwest” (using Silk and Walsh’s [2008] regional divisions) and only sparse representation of the “Pacific,” “Mountain West,” and “Southern Crossroads.” The data set does not include several of the communities with the largest concentrations of American Jews (New York, Los Angeles, Chicago, and Boston). We do not know how their inclusion might change the results of our analysis. In the future we hope to incorporate community studies of other researchers and later studies completed by Professor Sheskin to make the meta-data file more representative geographically and to keep the data current. Adding in earlier studies of these same communities may also clarify the suggested trends by using a different methodological approach. Additional studies, incorporating larger Jewish communities, communities in geographic areas not currently represented adequately, and communities with data from a longer time span will almost certainly increase the variation of the variables and enhance the analysis.

Following the approach of Stump (1986), who analyzed regional variations in the determinants of religiosity, further research will also examine the hypothesis that in different regions of the country, types of Jewish identity are related to various features of the Jewish community and perhaps to the broader community characteristics as well, and these may in turn affect how individual-level characteristics are related to Jewish identity. For example, living in the more religious South may result in stronger religious Jewish identity, while living in the less communally oriented West may result in stronger non-traditional or private expressions of Jewish identity.

Another question we hope to consider is whether identifying as Orthodox, Conservative, or Reform implies different expressions of Jewish identity in the various regions of the country. Our findings reinforce previous conclusions of the important relationships between denominational preference and strength of different kinds of Jewish identity. However, the interaction of denomination and place has been the subject of little systematic study.

With regard to survey cooperation rate, it does not surprise us that in communities with stronger Jewish religious identity and local ethnicity, the Jewish population is more ready to volunteer time to participate in a survey about its Jewish community. This relationship reinforces the need to control for the variable to make certain it does not affect other findings. Sample size may also reflect the readiness of the community to participate. We do not think these findings invalidate the results of the meta-analysis; they do, however, speak to the need to control for variance at the survey level. In the future we hope to find procedures to include more about survey variance in our analyses.

In conclusion, we are just at the beginning of this “sociology of Jewish place,” but we are at a stage that we hope will allow us to address questions of interest about the construction of Jewish identity in contemporary American life. We hope that the potential in this analysis will strengthen a commitment to introduce more standardization into portions of the community studies’ questionnaires. We also hope more questions can be included that might result in a more nuanced reflection of contemporary Jewish engagement in communal life.