The Politics of American Jews: Cohesion, Division, and Representation at the Institutional Level

Jewish Political Studies Review

Visit our new website: Institute for Global Jewish Affairs

Jewish Political Studies Review 12:1-2 (Spring 2000)

The Politics of American Jews: Cohesion, Division, and Representation at the Institutional Level

Laurence A. Kotler-Berkowitz and Lawrence Sternberg

This article examines political cohesion and division in the American Jewish community's central network of political and fundraising institutions. Employing data from the 1997 National Jewish Community Public Affairs Survey, the authors show that political activists in the community's Jewish Community Relations Councils are routinely more liberal in their political preferences than Federation donors, including the synagogue members among them. However, the authors argue, the political division between the activists and their main constituents is usually modest and does not warrant concerns about a lack of effective political representation in the organized Federation-JCRC system.

American Jews, like all groups defined by objective location in the social structure, display patterns of both political cohesion and political division. Many American Jews share a liberal orientation to politics and policies, as well as Democratic partisanship and electoral support, and many of the organizations which claim to represent the political interests of distinctive Jewish constituencies share liberal policy preferences. Moreover, the group's liberal political profile distinguishes it from Americans generally, meaning that Jews concentrate in a particular niche in the political structure. Together, these factors contribute to the group's political cohesion (Kotler-Berkowitz 1997; Zuckerman 1990).

At the same time, a major line of political division slices through the group. Orthodox and more religiously observant Jews have become increasingly conservative and increasingly likely to support Republican candidates, separating them politically and electorally from non-Orthodox and less religiously observant Jews (Cohen 1983a, 1983b, 1989; Cohen and Liebman 1997; Kotler- Berkowitz 1995, 1997; Liebman 1973; Lipset and Raab 1984, 1995; Sigelman 1991; Zuckerman 1990). At the institutional level, the major Orthodox political organization (the Orthodox Union) has increasingly voiced dissenting policy preferences from other Jewish organizations, particularly with regard to church- state separation, abortion, homosexuality, and state recognition of non-Orthodox forms of Judaism in Israel (Kotler-Berkowitz 1997). The emergence of a religiously-based division in American politics, in which doctrinal orthodoxy and religious commitment promote Republican partisanship and voting (Layman 1997), reinforces the political division within the American Jewish community.

Weber's (1946) response to Marx is the classic source for the theoretical position that groups defined by objective location in the social structure vary in the extent to which they are politically cohesive and politically divided. While Weber's retort addressed social class categories, his theoretical position applies generally to social structural groups, whether defined by class, ethnicity, religion, or race (Zuckerman 1989). Here we apply it to American Jews. Jews possess elements of both an ethnic and religious group, and we do not engage in the debate over whether they are best conceptualized as part of the American ethnic or religious structure. Choosing not to address that debate, however, does not bear on our approach to them as persons who share membership in an objective social category.

In this study, we examine institutional bases of political cohesion and division in the American Jewish community. The specific institutional participants we investigate are those involved in the community's central network of fundraising and social planning agencies, the synagogue members among them, and the leaders of a network of political organizations that parallel the fundraising and social planning agencies. We analyze cohesion and division through the use of survey data, employing indicators of political ideology, partisanship, and a series of political attitudes and policy preferences.¹

We concentrated on institutional participants for three reasons. The first reason has to do with addressing a relative lack of knowledge about the politics of institutional participants in the American Jewish community. Many studies have examined, either implicitly or explicitly, patterns of political cohesion and division among Jews at the mass level, and most of those have identified the political differences between Orthodox and non-Orthodox Jews as the major line of division in the group. However, the number of studies examining the political preferences of Jewish institutional participants is much more limited, even as Cohen and Bubis (1990) rightly note that investigating the politics of institutional participants and leaders helps complete our understanding of the community as a whole. Indeed, institutional development and affiliation are defining characteristics of American Jews and their community (Elazar 1995), 1989; Goren 1999; Gordon 1964), and institutions provide crucial bases of communal interactions, strength, and sustainability for American Jews, particularly in an era characterized by high levels of residential and geographic dispersion (Elazar 1989, chapter 10; Goldstein and Goldstein 1996). Because institutions are vital to the American Jewish community, understanding patterns of political cohesion and division among their participants is inherently beneficial.

The second reason is related to the specific institutions we investigated. The fundraising and political institutions we examined are the closest to centralized institutions that exist in an organizationally diverse American Jewish community. Jewish Federations are the primary fundraising and social planning agencies in regionally-specified Jewish communities across America. Jewish Community Relations Councils or Committees (JCRCs) are corresponding political organizations, claiming to represent the political interests of Jews who are organized within the Federation system.² Among their other functions, JCRCs lobby state and local governments, promote relations with other social groups and organizations, fight anti-Semitism, and conduct media campaigns to protect and promote Jewish interests. As the closest to centralized institutions that exist in the community, these organizations are particularly important in shaping patterns of political cohesion and division among the group's institutional participants and within the community as a whole.

As noted above, we also investigated synagogue members among the Federation donors. Synagogues are the Jewish group's primary religious institutions, providing a major basis of social cohesion and interaction in the community, and an alternative set of organizations for institutional participants. As such, synagogue memberships add to the complex patterns of institutional affiliations that characterize the Jewish group. Specifying such memberships in the analysis enables us to investigate that complexity more clearly. Specifically, it allows us to determine if synagogue members who contribute to the Federations hold different political views from Federation participants who do not belong to synagogues.

The third reason we investigated cohesion and division within these particular institutions and among their participants is related to an important normative concern: the issue of political representation within the organized Jewish community. The fact that a specific network of Jewish political organizations claims to represent the interests of the Federation contributors, including the synagogue members among them, led us to consider the connection between cohesion, division, and political representation. To what extent does political cohesion between the political and other institutional participants signify accurate political representation, and to what extent do institutional lines of political division compromise political representation? Moreover, by dividing Federation participants into those who are and are not synagogue members, we can test whether the political activists offer greater political representation to one group than the other.

Data

We employed data from the 1997 National Jewish Community Public Affairs Survey (PAS),³ which includes responses from over 6,500 contributors to Jewish Federations and more than 600 leaders of JCRCs, nearly all of whom, 97 percent, also report making a donation to their Federation in the year prior to the survey.

Surveys were mailed in November 1997 to randomly selected individuals from lists of contributors to Jewish Federations in 14 communities across the United States: Atlanta, Bergen County (NJ), Cleveland, Columbus, Dallas, Detroit, Houston, Indianapolis, Los Angeles, Milwaukee, Philadelphia, San Francisco and the East Bay, and Seattle. A separate mailing was made to the entire JCRC leadership in these communities, with the exception of Columbus and the East Bay. Because we did not have a parallel JCRC sample for the Columbus Federation sample, we removed all Columbus respondents from our analysis. However, because the San Francisco JCRC serves the political interests of the East Bay as well as San Francisco itself, we have retained East Bay respondents in our study. The merged data set which results from combining the random sample of Federation contributors with the inclusive sampling of corresponding JCRC leaders permits an investigation of the relationship between the politics of the political leadership elite and the core constituency they serve.

We are fully aware of the potential limits of these data and the methods used to collect them. Despite the large absolute number of respondents, the low response rates - approximately 25 percent of the more than 20,000 randomly sampled Federation donors and 50 percent of the 1,258 sampled JCRC leaders - raise questions of non-response bias and therefore of the generalizability of our findings to the target population as a whole. Nonetheless, we have reason to believe that problems of non-response are small and do not significantly affect our ability to generalize about the participants in these particular Jewish philanthropic and political organizations.

To address the question of non-response bias, we compared the respondents in the 1997 Public Affairs Survey with respondents in the 1990 National Jewish Population Survey (NJPS).⁴ The 1990 NJPS was devised to create a national demographic profile of all American Jews and the households in which they live, including the non-Jewish members of those households. The survey interviewed a random selection of Jews across the United States and involved 2,441 respondents. For purposes of comparison, we first used a subset of the NJPS population, those who identified their current religion as Jewish (N=1,734); (see Kosmin et al. 1991). From among these "Jews by Religion," we then selected respondents who reported making a contribution to the Jewish Federation in their community in the previous year (N=726). This group of respondents from the NJPS is, as best as we can construct, a similar group to that which the PAS polled.

Table 1 compares respondents from the Public Affairs Survey and the NJPS reference group along a series of variables. Demographically, the PAS respondents are slightly older, are more likely to be male, are somewhat more likely to be married, and have higher levels of household income than their NJPS counterparts. With respect to religious identity and affiliation, the PAS respondent group contains fewer Conservative Jews and more secular Jews, and PAS respondents are somewhat more likely to be synagogue members. Finally, PAS respondents are more politically moderate and less liberal and conservative than their NJPS counterparts. None of these differences, while clearly extant, are so severe as to raise doubts that PAS respondents are a decidedly unrepresentative sample of Jews connected to the community's central philanthropic and political agencies.

Table 1: Comparison Of 1997 PAS Respondents and 1990 NJPS Reference Group

1997 PAS Respondents 1990 NJPS Reference Group

Age

Under 35 7 18

35-44 18 23

45-54 23 15

55-64 16 16

65-74 20 18

75+ 16 11

Gender

Male 58 46

Female 42 54

Marital Status

Married 77 67

Not Married 23 34

Intermarriage

Spouse Jewish 93 89

Spouse not Jewish 7 11

Household Income

Under $30,000 10 20

$30,000-$49,999 15 27

$50,000-74,999 18

$50,000-$79,999 24

$75,000-$99,999 16

$80,000-$124,999 16

$100,000-$199,999 25

$125,000-$199,999 7

$200,000 and more 15 5

Denomination

Orthodox 5 7

Conservative 36 48

Reform 38 37

Reconstructionist 3 2

Just Jewish 15 3

Other 3 3

Synagogue Membership

Yes 77 64

No 24 36

Religious ServiceAttendance

Not at all 11 7

Several times/year 49 54

At least 1/month 22 24

At least 1/week 18 15

Jewish Day School
Yes 5 11

No 95 89

Political Ideology
Liberal 33 43

Moderate 55 37

Conservative 13 20

Total N 6523 726

Entries are percentages except those for Total N.

On other significant measures, the PAS respondents are very similar to the NJPS respondents, adding to our belief that the PAS data are valid. Outside the Conservative and secular categories, equal percentages of respondents categorized themselves as Orthodox, Reform, Reconstructionist, and other. PAS respondents do not consistently attend religious services more or less frequently than the NJPS respondents. Among married respondents in the two data sets, rates of marriage to non-Jews are very similar. Moreover, while NJPS respondents reported Jewish day school education at twice the rate of PAS respondents, both percentages are so low that the NJPS doubling is of little substantive significance.

We have two final reasons to be confident that non-response bias is not a problem in the PAS data. First, we created an ordinal measure of embeddedness in Jewish religious contexts, in which respondents received 1 point for synagogue membership, 1 point for going to religious services at least once a month, and 1 point for receiving a Jewish day school education. We then correlated this with a three-point ordinal scale of political ideology, ranging from liberal (-1) to moderate (0) to conservative (+1). The correlation, .11 (p = .01), is consistent with much previous research showing a weak to moderate connection among Jews between frequent location in religious contexts and political conservatism.

Second, we selected Orthodox, Conservative, Reform, and Reconstructionist respondents from both the PAS and NJPS surveys and created identical, ordinal measures of denominational traditionalism. We then correlated those measures with identical scales of political ideology. In the PAS data, the correlation was .19 (p=.01) and in the NJPS it was .14 (p=.05). When we added secular Jews to the end of our ordinal scales of denominational traditionalism, the correlations were .14 for the PAS respondents (p=.01) and .13 for the NJPS respondents (p=.05).⁵ Quite clearly, measures of Jewish denomination and political ideology among PAS respondents are correlated at essentially the same strength as identical measures among the NJPS reference group.

These four factors - the similar correlations between matching measures of denominational traditionalism and political ideology in the PAS and NJPS data sets; the connection between Jewish religious contexts and political conservatism among PAS respondents; the several demographic variables on which PAS and NJPS respondents are nearly the same; and the relatively small differences on other demographic variables - lend substantial evidence to our claim that the PAS data do not suffer from non-response bias. Those who provided answers to the PAS survey do not appear to be significantly different - either demographically or with regard to several key relationships - from a parallel reference group in the NJPS data, which we know were collected to provide a representative sample of American Jews. The PAS data, we believe, are valid and reliable, and can be employed to make generalizations about Jews who are involved in the community's major network of philanthropic and political organizations (though of course we make no claim that they represent the general population of American Jews).

Measures

In this research, we investigate how institutional affiliations structure political cohesion and division among American Jews. Are memberships in alternative Jewish institutions - political, philanthropic, and religious - tied to differences in partisanship, ideology, policy preferences, and other types of political and social attitudes? Consequently, are political cohesion and division among American Jewish institutional participants related to specific types of institutional affiliations within the organized community?

We divided respondents into three institutional categories. One group, whom we frequently refer to as political activists or political leaders, comprises the separately-sampled JCRC leadership. We then separated the remaining respondents, randomly sampled from the Federation donor lists, into those who do and do not belong to synagogues. The former group we call synagogue members, while the latter we refer to as donors (or occasionally "just donors").

Some comments about the institutional categories and their names are in order. A majority of the JCRC leaders (86 percent) are also members of synagogues, according to the surveys they completed, but none of them has been placed in the synagogue member category. In other words, the sample of JCRC leaders remains a separate group in our analysis. In addition, all respondents, regardless of which group they are placed in, contribute to Federations;⁵ however, only the group we refer to as donors has this as their sole institutional affiliation in this particular data set. We have, in sum, separated and identified the groups by their distinguishing rather than exclusive characteristic, and we trust the naming scheme will allow us to avoid, as much as possible, awkward references to the specific characteristics of each group in the remainder of this report.⁷

Table 2 provides a breakdown of respondents by institutional affiliation. A clear majority of respondents, nearly 70 percent, are synagogue members, while slightly more than one-fifth are just donors. Despite oversampling of the leadership of JCRCs and higher response rates, fewer than one-tenth of all respondents in the merged sample are political activists, a reflection of the elite nature of affiliation in political institutions.

Table 2: Institutional Affiliations of
1997 Pas Respondents

Institutional Affiliation Percentage N

Synagogue Members 68.4 4460

Donors 2.2 1451

Political Activists 9.4 612

Total 100.0 6523

Our first set of analyses below utilizes these institutional categories as independent variables in a series of bivariate tests. Subsequent analyses are multivariate, in which we remain focused on the institutional bases of the community's political cohesion and division while incorporating into statistical models a variety of other factors that could impact the political patterns. Here, the PAS provides data that allowed us to control for the effects of denominational affiliations (Orthodox, Conservative, Reform, and Reconstructionist), synagogue attendance, Jewish day school education, ethnic solidarity, and a series of demographic variables, including income, age, marriage, intermarriage, and gender. We also employed political ideology as a predictor variable when we examine other dependent variables and partisanship as a predictor when ideology serves as the dependent variable.

The Public Affairs Survey contains extensive data on the political preferences of respondents, and we employed a total of nine indicators of political positions as dependent variables. Measures of partisanship, ideology, and preferences on size and services of government are 3-point ordinal scales (which we treat as interval), ranging from -1 (Democrats and the most liberal positions on ideology and government size/services), through 0 (independents and moderates), and concluding at 1 (Republicans and the most conservative positions). The remaining dependent measures are 7-point ordinal scales (which, again, we treat as interval), and cover such topics as church-state separation, concerns for minorities (immigration, affirmative action, and welfare), cultural liberalism (abortion, school vouchers, and unmarried and same- sex partners), anti-Semitism, the Middle East peace process, and religious pluralism in Israel. Here, -3 represents liberal positions on church-state separation, minorities and cultural issues; low levels of suspicion of anti-Semitism; dovish stands on the Middle East peace process; and support for religious pluralism in Israel. In contrast, +3 on these scales represents the opposite preferences and attitudes: conservative stances on church-state separation, minorities and cultural issues; high levels of suspicion of anti-Semitism; hawkish positions on the Middle East peace process; and support for continued Orthodox control of religious life in Israel. On all the scales, 0 represents a middle or moderate position on the issue. The Appendix provides detailed information on variable construction.

Political Cohesion and Division: Empirical Findings

We begin the empirical analysis by examining political ideology (or general political orientation), because it allows us to make some broad comparisons among American Jews and between Jews and Americans generally. Table 3 presents data on the political ideologies of respondents from the 1997 PAS, the 1990 NJPS,⁸ and the 1990 and 1996 American National Election Studies

Table 3: Political Ideologies of American Jews and Americans Generally

Liberal Moderate Conservative

1997 Public Affairs Survey

All Respondents 33 55 13

Political Activists 45 47 9

Donors 37 53 10

Synagogue Members 29 57 14

1990 National JewishPopulation Survey

Jews by Religion 48 34 19

Reference Group 43 37 20

National Election Studies

1990, All Respondents 25 36 39

1996, All Respondents 26 30 44

Entries are percentages, totaling 100 (or 99 or 101 due to rounding) across rows.

(ANES),⁹ the dates of which correspond to the two surveys of Jews.

While we acknowledge the limitation of comparing across different data sets, as well as the difficulties in interpreting responses to questions of political ideology, we think the data reveal important patterns. As a group, these institutionally-involved American Jews retain some, but certainly not all, of the liberal political profile typical of American Jews. Among PAS respondents, liberals outnumber conservatives by more than a 2 to 1 ratio. In addition, there are more liberals and fewer conservatives among PAS respondents than among Americans generally, as reported in both the 1990 and 1996 ANES data. However, it is political moderation rather than liberalism that characterizes the institutional participants. Moderates comprise the slight majority of all PAS respondents. In addition, PAS respondents are both less liberal and more moderate than the Jews by Religion in the 1990 NJPS sample and the NJPS reference group, while retaining similar levels of political conservatism. PAS respondents are also more moderate than Americans generally. Because past research has demonstrated that across all American Jews, institutional involvement in general is correlated with lower levels of political liberalism (Kotler-Berkowitz 1997), the apparent displacement of liberals by moderates in the PAS data is not surprising.

While political moderation characterizes the institutionally- affiliated Jews, political variations within the population certainly exist. The political activists are noticeably more liberal and less moderate than others. Indeed, among the JCRC leadership, liberals rival moderates, levels of political liberalism nearly mirror those among the NJPS Jews by Religion (a point we return to later), and the ratio of liberals to conservatives is 5 to 1. It is among the synagogue members that rates of political moderation are highest and liberalism lowest, and the ratio of liberals to conservatives is reduced to just 2 to 1. Across each of the categories for political ideology, those who are just donors fall between the political activists on one side and the synagogue members on the other.

These initial data suggest two characteristics of the politics of PAS respondents: an overall pattern of political cohesion characterized by moderate and liberal preferences in comparison to Americans generally, in conjunction with internal lines of political division based on institutional affiliations. To move beyond these initial assessments, we examined means for each of the nine dependent variables, for the entire sample of respondents as well as for each group, and we conducted difference of means tests among and between the institutional categories for each dependent variable. Here, we use statistically significant differences in means as indicators of political division, while an absence of statistically significant differences signifies political cohesion.

Table 4 presents means on the dependent variables for all PAS respondents and for each institutional group, as well as a test of differences of means across the three institutional groups taken together. Recall that the negative ends of scales represent liberal ideology, Democratic partisanship, liberal preferences on domestic political issues, low degrees of suspicion of anti-Semitism, a dovish position on the Middle East peace process, and support for religious pluralism in Israel, while the positive ends of the scales represent the opposite positions.

Several patterns emerge from the data. First, these institutionally-involved Jews as a whole display a moderate to liberal political ideology, are Democratic in their partisanship, and are decidedly liberal on church-state separation and cultural issues, but they are somewhat to the conservative side of moderate on preferences for government size and provision of services and issues concerning minorities. In addition, they are as a whole less rather than more suspicious of anti-Semitism among other Americans. Concerning Israel, they are slightly dovish with regard to the Middle East peace process, and they are emphatically supportive of religious pluralism in the Jewish state. Overall, the means indicate moderate to liberal political preferences across a variety of domestic and Israeli political dimensions, with mild displays of moderate to conservative stances on selective American political issues.

Table 4: Mean Scores on Dependent Variables^a and Difference of Means Test^b

Dependent Variables All Respondents Political Activists Donors Synagogue Members Difference of Means
Political Ideology -.20 -.36 -.27 -.15 38.64

Partisanship -.45 -.64 -.48 -.41 30.05

Government Size and Services .20 -.04 .11 .26 37.99

Church-State Separation -1.19 -1.76 -1.15 -1.13 44.59

Concern for Minorities .85 .07 .89 .95 104.30

Cultural Liberalism -1.13 -1.38 -1.21 -1.06 20.94

Anti-Semitism -.57 -1.04 -.46 -.54 29.93

Middle East Peace Process -.38 -.82 -.48 -.29 27.55

Religious Pluralism in Israel -2.04 -2.11 -2.15 -2.00 6.76

^a Scales for political ideology, partisanship, and government size and services are three-point interval scales ranging from -1 through 0 to +1. The remaining scales are 7-point interval scales ranging from -3 through 0 to +3.

^b Entries for difference of means column are the F-scores for ANOVA difference of means tests across the four institutional groups.

Second, means for each institutional group support a general pattern of political division. Political activists are the most liberal and synagogue members the most conservative, while the donors occupy the middle ground. Indeed, the political activists are more liberal (or less conservative) than the synagogue members on each and every measure. The JCRC leadership is also more liberal than the donors in eight of the nine instances (the exception is religious pluralism in Israel); in seven of those eight cases, the distance between the political leaders and the donors is less than the distance between the political activists and the synagogue members (here the exception is perceptions of anti-Semitism).

Third, initial difference of means tests displayed in the last column of Table 4 provide additional support for the claim that institutional affiliations structure political divisions within the organized community. The F-scores for ANOVA difference of means tests are all substantially above the threshold for standard levels of statistical significance, indicating that at least one institutional group differs from another group along each dependent variable.

Table 5 probes these differences between means in finer detail, offering group-by-group difference-of-means and accompanying statistical tests. The dependent variables are listed in the left-hand column, starting with ideology and concluding with religious pluralism in Israel. In each of the subsequent three columns (one for each group-by-group comparison), one group is identified as group I, and the other as group J. Entries are the mean of group I minus the mean for each corresponding group J, for the particular dependent variable. Negative entries indicate that group I is more liberal than the corresponding group J, while positive entries indicate group I is more conservative than the group J counterpart. Variables that are statistically significant at the .05 level and below are indicated. Again, we use statistically significant differences as indicators of political division and the absence of significant differences as evidence of political cohesion.

The group-by-group comparisons demonstrate that the political activists are more liberal than both the synagogue members and the donors, as indicated by the statistically significant, negative entries under the two columns where political activists are identified as group I. The one exception, again, is in the case of attitudes toward religious pluralism in Israel. Moreover, examining the size of the difference in means demonstrates that in most cases, the political activists are more politically divided from the synagogue members than from the donors. In the last column, the statistical tests indicate that the donors display different means than the synagogue members on six of nine variables, and in all six cases the donors are more liberal.

Table 5: Group-by-Group Difference of Means across Dependent Variables
Synagogue Members Synagogue Members -
DependentVariables Group I: Political Activists Political Activists Donors Group J: Donors

Political Ideology -.09^b -.21^c -.11^c

Partisanship -.16^c -.23^c -.07^b

Government Size and Services -.14^b -.29^c -.15^c

Church-State Separation -.60^c -.62^c -.02

Concern for Minorities -.82^c -.88^c -.06

Cultural Liberalism -.17^a -.32^c -.15^c

Anti-Semitism -.58^c -.50^c .08

Middle East Peace Process -.34^c -.53^c -.19^b

Religious Pluralism in Israel .04 -.11 .15^b

^a p<.05 ^b p<.01 ^c p=.000

We confirmed that institutional affiliations structure political division among these American Jews through a series of nine multivariate regression analyses, one for each dependent variable. Here, we employed the synagogue members as the reference group and used dummy variables to indicate the political activists and donors. Statistically significant coefficients for the dummy variables signify political division, while coefficients failing to reach conventional levels of statistical significance indicate political cohesion.

In addition, we controlled for a variety of other explanatory factors, including denominational affiliation (dummy variables representing Orthodox, Conservative, Reform, and Reconstructionist Jews, with secular Jews as the reference group); frequency of synagogue attendance; feelings of ethnic solidarity; and a series of demographic variables that includes income, age, marriage, intermarriage, and gender. Finally, we used political ideology as a control variable in eight of the nine models, and used partisanship as a control when ideology was the dependent variable.¹⁰ As proximate political variables, we expected ideology and partisanship to have a strong impact on the dependent variables, and therefore to provide a tough test for the institutional variables.

Regression results are reported in Table 6. The political divisions between synagogue members and political activists that were suggested in the bivariate difference of means test remain in each multivariate model, as indicated by negative coefficients falling below the .05 level of statistical significance in eight of nine cases. Compared to synagogue members, the JCRC leaders are more liberal ideologically and on specific American political and social issues, more Democratic in their partisanship, less suspicious of anti-Semitism, and more dovish on the Middle East peace process. Only on the issue of religious pluralism in Israel are the political activists and the synagogue members characterized by political cohesion, as the failure of the coefficient to reach standard levels of statistical significance demonstrates.

In contrast to these findings, political divisions between those who are just donors and synagogue members are for the most part erased once other variables are controlled. Across seven of the variables, donors are not significantly different from the synagogue members. In one case, government size and service provision, donors appear more liberal than synagogue members; in a second case, church-state separation, the donors appear more conservative than the members of synagogues. In both cases, though, levels of statistical significance are marginal (p<.10) rather than conventionally decisive (p<.05). Between the two groups, political cohesion rather than division is the dominant tendency, once other explanatory factors are controlled.

We re-specified the regression models in two ways. First, we added a series of dummy variables for city (or county), combining the Federation and JCRC samples in each local area. Here, we used San Francisco and the East Bay as the reference group because it is the most ideologically liberal and because it has the largest percentage of respondents in the survey (11.6 percent). The substantive findings for the differences between the political activists and synagogue members remained the same, as did the findings for the general absence of differences between the donors and synagogue members.¹¹

We also re-specified the original regression models to make donors the reference group, in order to provide a direct multivariate test of the divisions between them and the political activists. The results (not displayed) indicate that at the .01 significance level, political activists are more Democratic in their partisanship; more liberal on government size and services, church-state separation, concern for minorities and cultural issues; less suspicious of of anti-Semitism; and more dovish on the Middle East peace process than the donors. In addition, at the .10 significance level, the political activists are more ideologically liberal and more supportive of religious pluralism in Israel. In sum, the empirical findings demonstrate a consistent pattern of political division between the political activists and the two other groups, while political cohesion rather than division characterizes the relationship between synagogue members and those who are just donors.

Table 6: Multivariate Regression: Determinants of Political Cohesion and Division

Independent Variables Political Ideology Partisanship Government Size/Services Church-State Separation Concern for Minorities

Institutional variables

Political activists -.08^c -.14d -.21^d -.46^d -.70^d

Donors -.03 -.03 -.06^a .10^a -.05

Control variables

Political ideology .54^d .56^d .72^d .92^d

Partisanship .44d

Synagogue attendance .01 -.02^a -.03^a .08^b -.10^d

Ethnic solidarity .00 -.05^d -.05^c -.05^a -.01

Orthodox .32^d .03 .02 1.03^d .10

Conservative .14^d -.10^d .00 -.07 .11^a

Reform .00 -.03 .03 -.28^d .08

Reconstruc tionist -.04 -.12^b -.15^b -.44^d -.25^b

Jewish education -.01 .08^b -.04 .29^c .05

Income -.02^c .05^d .06^d -.07^d -.05^d

Age .00 -.01^b .01 -.02^a .07^d

Married .03 -.01 .01 -.04 .04

Intermarried -.04 .03 -.06 .04 -.20^c

Male .09^d .14^d .24 .03 .05

(Constant) -.07 -.44d .02 -.77^d 1.15^d

R-squared (adjusted) .29 .30 .22 .17 .21

^a p<.10 ^b p<.05 ^c p<.01 ^d p=.000

N= 5124 for each analysis.

Independent Variables Cultural Liberalism Anti-Semitism Middle East Peace Process Religious Pluralism in Israel

Institutional variables

Political activists -.20^d -.42^d -.33^d -.06

Donors .06 .05 .06 .05

Control variables

Political Ideology .69^d .11^c .79^d .33^d

Partisanship

Synagogue attendance .10^d .00 .09^c .12^d

Ethnic solidarity .00 -.09^c .10^c .00

Orthodox .90^d .39^c 1.48^d 2.79^d

Conservative -.13^c .21^c .07 -.24^d

Reform -.25^d .10 -.32^d -.56^d

Reconstructionist -.41^d -.28^b -.81^d -.72^d

Jewish education .35^d -.06 .17 .42^d

Income -.01 -.14^d -.12^d -.03^c

Age -.03^c -.12^d -.01 -.14^d

Married .05 .09 .14^b -.07^a

Intermarried -.07 -.01 -.19^b -.20^c

Male .23^d -.31^d -.13^c .03

(Constant) -1.15^d .46^c .02 -1.43^d

R-squared (adjusted) .24 .05 .20 .39

^a p<.10 ^b p<.05 ^c p<.01 ^d p=.000

N= 5124 for each analysis.

We note that we are not claiming that of all the independent variables, the variable capturing political activists offers the greatest amount of explanatory power, or that affiliations with the different institutions are the only source of political division within the community. The standardized beta coefficients (not displayed) indicate that political ideology (and partisanship when ideology is the dependent variable) is usually, though not always, the most powerful explanatory variable. As the only explicitly political predictor in the model, it would be surprising if ideology (or partisanship) did not provide the greatest explanatory power for dependent variables that are also political in nature. Other variables that are statistically significant in the models are also sometimes more powerful than the political activist measure. Moreover, consistent with many previous studies, denominational affiliations are a routine source of political division: relative to secular Jews, Orthodox Jews are frequently more conservative and Reform and Reconstructionist Jews are frequently more liberal. No matter the strength and consistency of other variables, though, the indicator of political activists displays a nearly constant effect across the models, thus supporting our claims that the JCRC leaders are politically divided from the other institutional categories, and that institutional affiliation is one basis of political division among American Jews organized in the Federation and JCRC systems.

Assessment: Cohesion, Division, and Political Representation

The empirical results consistently demonstrate political division among American Jews affiliated with the group's central network of fundraising and political organizations. More specifically, the group's political activists routinely display differences on measures of ideology, partisanship, political and social attitudes, and policy preferences, from the synagogue members and donors. At the same time, institutional affiliations do not structure political divisions between the donors and synagogue members, who stand together as a cohesive bloc once other factors that affect political preferences are controlled.

The one exception to the pattern of political division found in the empirical analysis is the issue of religious pluralism in Israel. The political activists and the synagogue members are cohesive in their preference for the State of Israel to recognize non-Orthodox religious movements and practices. There is, as well, suggestive statistical evidence that the donors may be slightly less liberal on this issue than the political activists, though no less liberal than the synagogue members. Here, we suspect that the political cohesion between the political activists and the synagogue members stems from complementary sources. The political leaders object ideologically to the strong connection between the state and one form of religion in Israel, while the synagogue members, a solid majority of whom are Conservative and Reform, object to the state's refusal to grant legitimacy to their religious movements in Israel.

In addition, consider that the most powerful explanatory variable in the multivariate model predicting preferences on religious pluralism in Israel is Orthodoxy, which displays a strong, positive relationship with support for the exclusive legitimation of Orthodoxy in the Jewish state (see Table 6). Each of the other denominational movements, in contrast, displays smaller, negative relationships with backing for the Orthodox monopoly on state legitimation. These findings provide good evidence that the issue of religious pluralism in Israel creates a serious political division among the synagogue members, specifically between Orthodox and other denominationally-affiliated Jews, but not between synagogue members generally (most of whom are Conservative and Reform) and political activists.

The one exception notwithstanding, the main empirical finding is overwhelmingly in the other direction. The statistical indications of institutional lines of division on nearly all the dependent variables lead, ultimately, to questions about the political representation by group leaders of their constituents (Wald 1997). Recall that the Jewish Community Relations Councils and the Jewish Federations are organizationally linked. The JCRCs are designed, in part, to represent the political interests of those group members who are organized in the Federation system, and the Federations are a major institutional underpinning of the community. Because the synagogue members within the Federations do not differ in their politics from those who are just donors, we can consider them together as a unified bloc; in other words, the question of whether JCRC leaders provide better representation to the synagogue members or the donors need not be addressed. How adequately, then, do the Jewish political leaders represent the political interests, preferences, and views of their constituents, the rank and file affiliated with the community's central fundraising and planning agencies?

The answer depends first on the normative model of representation employed, and second on what that model implies about levels of political cohesion and division between political leaders and their constituents. While the dispute over what constitutes proper or adequate representation is a long-standing and complicated one in the literature on political theory (Pitkin 1967, especially chapter 7), the debate can be fairly well summarized by three competing models. At one extreme is the independence model (Pitkin 1967), under which representatives are free to act as they see fit in pursuing the interests of their constituents. The model sees the representative "as a free agent, a trustee, an expert who is best left alone to do his work" and tends to see "political questions as difficult and complex, beyond the capacities of ordinary men" (Pitkin 1967:147). This normative model of course requires no political cohesion between the political activists and the other institutional participants. But we reject this model for the institutions we are examining, both because an ethnic or religious community that is voluntarily constituted requires some minimal connection between its political representatives and their constituents, and because the model has a rather disparaging and unrealistic view of the political capabilities of ordinary constituents.

At the other extreme, the mandate model (Pitkin 1967) requires representatives to reflect perfectly, or at least nearly perfectly, their constituents' views. Representatives in this model are "mere administrative conveniences�saving the polity the expense of continuing referenda" (Achen 1975:1218), or, translated into the dynamics of ethnic or religious communities, continuing consultations with rank and file institutional participants. Representation according to the mandate model requires strict political cohesion between political activists and the other institutional affiliates, reflected in an absence of statistically significant differences between the institutional categories. In contrast, any statistical differences that signal political division are enough to indicate an absence of proper political representation. Under this model of representation, the political activists fail to provide satisfactory political representation because they are consistently more liberal in their politics than the donors and synagogue members. But we reject this conception of political representation for the Jewish community as well. Such a model is ultimately harmful because it means political representatives must abdicate any role as political experts who are particularly attuned to the political environment and who actively employ their political expertise to represent the community's political interests.

A third model, which we call a flexible accountability model, stands between the independence and mandate models. It sees representatives as specialists in political and public affairs in ways that constituents need not be, while still assuming that constituents are adequately knowledgeable and informed to be aware of their own political interests (Achen 1975). Representatives may be expected on the basis of their expertise to diverge, within reasonable bounds, from the specific preferences of their constituents, as long as their overall approach to politics aligns with that of their constituents. Constituents, in turn, need only be assured that their preferences are being reasonably well reflected among the political elites claiming to represent them. In a voluntarily constituted polity, such a model seems to us to make the most sense. It offers an effective division of labor, allowing political activists to concentrate on political matters, while still holding them accountable in a flexible manner to the political preferences of the polity's members.

Linking such a model to a framework of cohesion and division means that strict political cohesion is not required for adequate political representation, and that a certain amount of political division is acceptable. Consequently, a more substantive assessment of the extent of political cohesion and the depth of division between political activists and other institutional affiliates is necessary.

We argue that under this model, the political activists are providing adequate political representation to their constituents. Consider first that whatever political division occurs among the institutionally-affiliated Jews studied here, it does so within an overall pattern of political cohesion. Though not as notably liberal as American Jews generally, these institutional affiliates all hold political preferences that fall within a relatively bounded range of moderate to liberal. They are also more liberal and moderate and less conservative than Americans as a whole, a distinctive political profile that contributes to their political cohesion vis-à-vis the society in which they live.

Consider as well some further statistical details about the depth of the political division that separates the activists from the other institutional participants. In no cases do the divided groups take stands that place them at opposite ends of the scales measuring political preferences. In fact, the group means, even when significantly different in a statistical sense, never display distances from each other of more than .29 on the 3-point scales and never more than .88 on the 7-point scales (Table 5). Controlling for other explanatory variables reduces the distance between the political activists and the synagogue members on the dependent scales in every multivariate regression model (Table 6), where the largest regression coefficient has an absolute value of .21 on the 3-point dependent scales and an absolute value of .70 on the 7-point scales. In addition, in five of the nine cases between the political activists and donors (results not displayed), the distance between the groups is also reduced in the multivariate models, and in three of the cases it remains essentially unchanged from the bivariate specification.

The only characteristic of the political division between the political activists and their constituents that raises concerns under this model is the consistent direction of the division: the activists are always more liberal than the other institutional participants. It would perhaps be preferable if the activists were more liberal on some issues and more conservative on others, in which case the liberal and conservative divisions would partially cancel each other out over the series of political issues that political activists routinely address on behalf of the organized community.

On the other hand, we do not believe that a consistently liberal bias is without some redeeming value. Indeed, as analysts who are concerned with the community's political affairs, we think there may be some benefit to political division. To the extent that JCRC leaders (and the professionals who work with them) are aware of the political division between themselves and their constituents, then the political differences may act as an inducement to the activists to routinely evaluate and, where necessary, reformulate the political preferences and policy positions they advocate on behalf of the community. In other words, political division between the activists and the rank and file may just keep the leaders from stagnating in their political thinking.

In sum, our assessment is that the political division between the political activists and the other institutional participants is only rarely substantial (concern for minorities), usually rather modest (government size and services, church-state separation, cultural liberalism, anti-Semitism, and Middle East peace process), and occasionally slight (political ideology and partisanship). To borrow a phrase from Wald (1997:144) and apply it to our analysis, if the political activists are not perfectly representative of the institutional rank and file, neither are they acutely out of step with it. We do not discount the consistent, statistical evidence of institutional lines of political division; indeed, we emphasize such evidence. However, under a model of flexible accountability that views representatives as specialists in political affairs, we think that concerns about a lack of accurate political representation within these organizations are unwarranted.

Moreover, the JCRC activists may be performing their representative function quite well with regard to the wider Jewish masses. Analyzing linkages between JCRCs and Jews at the mass level puts us on somewhat slippery ground: JCRCs do not legitimately claim to represent American Jews who have no connection to the Federation system, and in any case, it is doubtful that American Jews would recognize claims to be represented by institutions with which they are not affiliated. As an analytic strategy, however, linking the institutional and mass levels through the prism of political cohesion and division and the normative model of representation yields an interesting observation.

To do so, we return to the distribution of political ideologies displayed in Table 3. The data appear to indicate greater political cohesion between the JCRC leaders and American Jews generally (represented by the 1990 NJPS Jews by Religion) than between the political activists and the other institutional affiliates. The percentage of political activists claiming a liberal orientation to politics is roughly equal to the percentage of liberals among the NJPS Jews by Religion. There is some disparity between moderates and conservatives within the two groups - the JCRC leadership is still more moderate and less conservative in its politics than the Jewish masses - but the plurality position at the mass level, political liberalism, seems well represented among the JCRC activists. To the extent, then, that the JCRCs serve as a central network of political organizations for the Jewish community, they seem to display a fairly high degree of political cohesion with Jews at the mass level, a combination of those who are and are not institutionally affiliated, and therefore seem to represent the general political orientation of American Jews adequately.

Conclusion

No matter whether they are defined by ethnicity, religion, race, class, or other objective categories, social structural groups are characterized by both political cohesion and division. Moreover, when political divisions occur, they are not randomly distributed. Instead, they are frequently the result of characteristics that are tied to a group's objective definition, such as religious divisions within a religious group, or as we have shown here, to different types of institutional affiliations within a social group. The result is that from larger, objective social categories emerge subgroups, the members of which are similar to each other in both their political preferences and some additional social characteristic, but are politically divided from other group members with whom they do not share a social similarity and connection.^12d Put in slightly more colloquial terms, not all members of objectively-defined social categories are the same, and the differences between them may be politically significant.

American Jews are no exception to this rule. They display both relatively high levels of political cohesion along with political division. At the mass level, American Jews are generally liberal in their preferences, and distinctively so relative to Americans generally, but studies have repeatedly shown that Orthodox Jews tend to hold different political views and electoral preferences than other Jews. Here, we have shown that participants in the group's most centralized set of institutions display political preferences within a fairly narrow range, denoting political cohesion, and continue to be more moderate and liberal and less conservative than Americans generally. We have also demonstrated empirically that within the organized Federation and JCRC system, the community's political activists report different political preferences than its synagogue members and those who are just donors. We reiterate, though, that in our assessment the political division is not severe and does not compromise political representation within the organized community. Moreover, a reasonable level of political division may even provide benefits by challenging political leaders to evaluate consistently their understanding of the community's political interests.¹³ Analytically, understanding that social groups are both cohesive and divided in their politics is crucial: without doing so, researchers may assume political cohesion where it does not exist and may fail to identify lines of division where they do. Consequently, they will be unable to offer substantive assessments of political cohesion and division that characterize the social categories' members, and they will be unable to evaluate what varying levels of cohesion and division signify for political representation within groups.

Appendix: Variable Construction

This appendix provides detailed information on variables, their construction, and ranges. In total, we employed nine dependent variables across all the analyses, and 17 independent variables in the multivariate regression analyses.

Dependent Variables

Partisanship, ideology, and government size and services are ordinal variables that we treat as interval, each ranging from -1 to +1. On the respective measures, Democrats, liberals and those favoring larger government with many services were coded -1; independents, moderates, and those having no preference on government size and services were coded 0; Republicans, conservatives and those favoring smaller government with fewer services were coded +1.

The remaining dependent variables are ordinal scales (again treated as interval), each composed from a series of components and each ranging from -3 to +3. Factor analyses (usually confirmatory but in one case exploratory) were performed for the components of each scale, to ensure that they loaded together on a single underlying dimension. In the empirical analyses, however, the ordinal scales were employed because the values on factor scales are not as readily interpretable as they are on ordinal indices.

The measure of church-state separation was constructed from questions about the Ten Commandments being displayed in courtrooms, government aid for religious school tuition, and religious displays in public areas. Respondents received -1 point for strongly agreeing that the Ten Commandments should not be hung in any courtroom, and +1 points for strongly disagreeing. Respondents garnered -1 point for answering "no" to the question: "I believe that government aid should be given to families for tuition in private religious schools," and +1 point for answering "yes" to the same question. Lastly, respondents received -1 point for answering that they would always object to temporary religious displays in public parks and +1 point for answering they would never object to such displays. Negative and positive points were added to create the final church-state separation scale, with -3 representing the extreme separationist view and +3 the most accomodationist perspective. The factor loadings for the components were .76 for the Ten Commandments, .63 for government aid for religious school tuition, and .74 for religious displays.

For our two indices of social issues, we relied on explanatory factor analysis to combine different variables into scales. Six constructed variables were initially factor analyzed. The factor analysis revealed that positions on immigration, affirmative action, and welfare loaded together (rotated factor loadings of .69, .64, and .73, respectively), while stances on abortion, vouchers for non-religious schools, and the rights and benefits of unmarried and same-sex partners loaded together (rotated factor loadings of .76, .62, and .50, respectively).

Concern for minorities, therefore, was constructed by adding negative and positive points for responses to questions on immigration, affirmative action, and welfare, producing a range from -3, the most liberal, to +3, the most conservative. Respondents received -1 point for stating the number of legal immigrants to the U.S. annually should be increased, and +1 point for stating the number should be decreased. The constructed measure of affirmative action was itself composed of two variables. Respondents received -1 point for strongly disagreeing to the following statement and +1 point for strongly agreeing: "The use of racial, ethnic or gender quotas should not be permitted in hiring or in college admissions." In addition, respondents received -1 point for strongly agreeing and +1 point for strongly disagreeing to the following statement: "Race, gender, or ethnicity can each be considered as one factor among others when choosing qualified candidates for jobs or in college admissions." Points on these two responses were totaled, and the initial -2 to +2 variable was collapsed at each end to yield a -1 to +1 affirmative action variable. The constructed measure of welfare was composed of four initial variables. Respondents were given -1 point for stating they strongly disapprove of limiting the number of years in which recipients can receive welfare, requiring welfare recipients to work if able, ending increases in welfare payments to women who give birth to children while on welfare, and shifting welfare programs from the federal to the state governments; they received +1 point for each of the above questions to which they expressed strong approval. The initial -4 to +4 variable was collapsed on each side, yielding a final -1 to +1 variable for preferences on welfare policy.

Cultural liberalism was constructed by adding negative and positive points on preferences on abortion, vouchers for non-religious schools, and the rights and benefits of unmarried and same-sex partners. Again, the final variable ranges from -3 to +3, referencing those who are most liberal and those who are most conservative. Here, each component was constructed from one variable in the data set. Respondents received -1 point for stating that abortions should remain legal as they currently are; no point for asserting that abortions should be legal only in cases of saving the mother's life, incest, or rape; and +1 point for answering that abortions should never be legal. Respondents were granted -1 point for answering "no" to the question: "I believe that government aid should be given to families for tuition in private non-religious schools," and +1 point for answering "yes" to the same question. Survey participants garnered -1 point for answering "yes, without conditions" and +1 point for answering "no, under no circumstances" to the following statement: "Unmarried couples, including same-sex partners, should have the same rights and benefits as married couples."

The anti-Semitism index taps feelings of whether anti-Semitism is increasing or decreasing in several realms. Respondents received -1 point for answering that anti-Semitism is decreasing in employment, in politics, and in government; they received +1 point for stating that anti-Semitism is greatly or somewhat increasing in those areas. Those who answered that anti-Semitism is remaining about the same received no points. Negative and positive points were added to construct the final scale, so that -3 represents those who are least suspicious of anti-Semitism and +3 those who are most wary of anti-Semitism. Factor loadings were .83 for anti-Semitism in employment, .90 for anti-Semitism in politics, and .89 for anti-Semitism in government.

The Middle East peace process scale was constructed from seven components. Respondents received positive points for hawkish views and negative points for dovish views. Survey participants garnered -1 point for strongly disagreeing with the statement that "You can never trust the PLO to make a real peace with Israel," and +1 point for strongly agreeing with it; -1 point for strongly agreeing that "Palestinians have a right to an independent state as long as it doesn't threaten Israel," and +1 point for strongly disagreeing; -1 point for strongly agreeing that "Israel should freeze further settlements on the West Bank," and +1 point for strongly disagreeing; -1 point for strongly disagreeing with the statement that "It is proper for Israel to build new Jewish housing - such as Har Homa - in east Jerusalem," and +1 point for strongly agreeing; -1 point for strongly disagreeing that the Netanyahu government "should have renounced the Oslo Agreement," and +1 point for strongly agreeing; -1 point for strongly agreeing that the Netanyahu government "should try to work more closely with Arafat," and +1 point for strongly disagreeing; -1 point for stating that the Clinton administration should place pressure only or mainly on Benjamin Netanyahu, and +1 point for responding that the Clinton administration should pressure only or mainly Yasser Arafat. An initial scale ranging from -7 to +7 was recoded to yield a final 7-point scale ranging from -3, the most dovish perspective, to +3, the most hawkish. Factor loadings were .65, .60, .75, .74, .58, .69, and .58 respectively for the components detailed above.

The measure of religious pluralism in Israel was constructed from three components. Respondents received -1 point for strongly agreeing that the State of Israel should fully recognize conversions performed by Conservative and Reform rabbis when conducted in Israel; -1 point for strongly agreeing that the state should fully recognize such conversions when conducted outside Israel; and -1 point for strongly agreeing that men and women should be allowed to pray together in the vicinity of the Western Wall in Jerusalem. They received +1 point for strongly disagreeing to each of the above statements. Negative and positive points were added to yield the final index, ranging from -3, indicating the greatest levels of support for religious pluralism in Israel, to +3, referencing the lowest levels of support for religious pluralism in the state. Factor loadings were .90 for conversions outside Israel, .91 for conversions in Israel, and .74 for prayer at the Western Wall.

Independent Variables

Our primary independent variable is the institutional status of respondents. Respondents were divided into three groups according to their institutional affiliations: those who are members of their local JCRC boards (political activists); those who donated to the local Federation but did not claim any other institutional affiliation (donors); and those who donated to the local Federation and belong to a synagogue but are not on the JCRC board in their community (synagogue members). These institutional groupings provide the basis for the difference of means tests reported in Tables 4 and 5.

For the multivariate regression analyses, dummy variables reference the political activists and the donors, with the synagogue members as the reference group. In addition, we introduced a series of other variables in the regression analyses. Several of them are demographic. Income is an ordinal scale measuring annual household income, with the following categories and codes: under $30,000 (coded 1); $30,000-49,999 (2); $50,000-74,999 (3); $75,000-99,999 (4); $100,000-199,999 (5); and $200,000 or more (6). Age is also an ordinal variable, with the following categories and codes: under 35 (coded 1); 35-44 (2); 45-54 (3); 55-64 (4); 65-74 (5); and 75 over (6). Married, intermarried, and male are each dichotomous variables, coded 1 to indicate that respondents are married, married to someone who is not Jewish, and male.

Other independent variables capture aspects of respondents' cultural and psychological characteristics. A series of dichotomous variables reference respondents who consider themselves Orthodox, Conservative, Reform, and Reconstructionist. The synagogue attendance measure is an ordinal variable which we treat as interval; respondents were coded 1 for never attending synagogue, 2 for attending several times per year, 3 for attending about once a month, and 4 for attending about once a week or more. Jewish education is a dichotomous variable, with respondents coded 1 to indicate they attended a Jewish day school for most of their primary and secondary education. Ethnic solidarity is a measure of psychological ties to other Jews. Respondents received -1 point for answering that they felt somewhat or very distant to other American Jews and to the Israeli people, and +1 point for stating that they felt very close to those two groups. The initial -2 to +2 variable was collapsed at each end to produce a final scale ranging from -1 (the lowest level of identification with other Jews) to +1 (the highest level of identification).

Lastly, two dependent variables (described above) also served as independent variables in the regression analyses. Partisanship was employed as an independent variable when ideology was the dependent measure; in all other regression analyses, ideology served as an independent variable.

* * *

Notes

* The authors jointly thank Steven M. Cohen for helpful comments on an earlier version of the article, while retaining full responsibility for the analysis. Kotler-Berkowitz thanks the Lady Davis Fellowship Trust and the Department of Political Science, Hebrew University, for support while preparing the final manuscript for publication. The research reported here was initially presented at the Midwest Political Science Association Annual Meeting, April 15-17, 1999, Chicago, Illinois.

1. Two comments concerning our methods and objectives are in order. First, we assert that survey data and statistical methodologies provide a valid basis for analyzing patterns of political attitudes and behavior in the Jewish community (for a recent defense of survey-based research in political science generally, see Brady 2000). At the same time, we recognize that alternative methodologies, more qualitative in nature, are available for studying patterns of cohesion and division in the community (for an example of non-survey based methods in the study of cohesion and division, see Kotler-Berkowitz's 1997 analysis of the Joint Program Plans of the National Jewish Community Relations Advisory Council). Second, in undertaking this analysis, we purposefully leave aside what has been the dominant objective in the study of American Jewish politics: explaining American Jewish liberalism. For a recent addition to this literature, see Cohen and Liebman (1997). Levey (1995a, 1995b) offers critiques of much of this literature.

2. In most cases, JCRCs are constituent agencies of local Federations; this tends to be the case in midsize and smaller communities in which JCRCs are the major and sometimes only organized Jewish political organization. In other cases, typically in larger cities, JCRCs may be somewhat independent from the Federations, and include both their own members directly as well as representatives from other Jewish political organizations active in the area. In these latter cases, JCRCs occupy a dual role, operating as both political organizations in their own right and umbrella organizations encompassing other political groups. In addition to their vertical linkage to Federations, JCRCs are also linked horizontally to a national organization, the Jewish Council for Public Affairs (JCPA), recently renamed from the longstanding National Jewish Community Relations Advisory Council (NJCRAC). Federations are likewise linked horizontally to a national agency, the United Jewish Communities (UJC), itself recently renamed and reconstituted from the Council of Jewish Federations (CJF). Nationally, a long and varied history links CJF/UJC and NJCRAC/JCPA. For more details on CJF, NJCRAC, and local Federations and community relations councils, see Elazar (1995).

3. The survey was sponsored by the Nathan Perlmutter Institute for Jewish Advocacy at Brandeis University, the Jewish Council for Public Affairs, and a supporting cast of Jewish Federations, Jewish Community Relations Councils, and other funding organizations and individuals (see Raab and Sternberg 1998 for a complete list). None of these organizations or persons bears responsibility for the analysis and interpretations presented here.

4. National Jewish Population Survey data were provided by the Council of Jewish Federations in conjunction with the Mandell Berman Center-North American Jewish Data Bank and The Graduate School and University Center, City University of New York. None of these organizations bears responsibility for our use and interpretation of the NJPS data.

5. The slight drop in the correlation after adding secular Jews to the end of the ordinal scale is consistent with previous research showing secular Jews are less politically liberal than members of the most religiously liberal denominations, Reform and Reconstructionism. Indeed, putting secular Jews between Reform and Conservative Jews on the denominational scale raises the correlation with political ideology among PAS respondents back to .16 (p=.01), an indication that secular Jews are better placed empirically between Reform and Conservative Jews with regard to their political ideology.

6. The JCRC leadership is expected to contribute to Federation campaigns and, as reported in the text, nearly all report doing so.

7. We originally had four groups in our analysis. Two groups corresponded exactly to the composition of donors and synagogue members described in the text. Within the JCRC sample, we created two groups: a minority group comprised of respondents who belong to the JCRC leadership but are not members of synagogues, and a majority who belong to both institutions. We did this in order to determine if those who are joint participants in the group's religious and political institutions differed in their politics from those who are political activists only and those who are synagogue members. The empirical results consistently showed that joint political-religious affiliates did not differ from those who were only members of the JCRC leadership, but did differ from the synagogue members. Put slightly differently, the joint participants were politically similar to the exclusively political activists, but not to the synagogue members. Consequently, we discarded the joint participant group, re-merging it with the other JCRC leaders to form the political activist category described in the text.

8. Just under one-third of all NJPS respondents (768 out of 2,441) were randomly selected to answer the survey's political ideology question. Consequently, the number of respondents in the Jews by Religion category in Table 3 is 546 and the number of respondents in the reference group is 246.

9. American National Election Studies (ANES) data were obtained from the American National Election Studies 1948-1997 CD-ROM (Sapiro et al. 1998) and were made available by the Inter-University Consortium for Political and Social Research. ICPSR bears no responsibility for our use and interpretation of ANES data.

10. Ideology and partisanship are correlated at .50 (p=.000), and therefore cannot be entered simultaneously into a single regression model. We opted to use ideology as a control in most analyses because it is a broader concept that theoretically should help shape specific issue positions and policy preferences more strongly than partisanship.

11. When controlling for city, some minor changes occurred in the coefficients and levels of statistical significance for the political activists and donors. In six cases - partisanship, government size and services, church-state separation, concern for minorities, cultural liberalism, and the Middle East peace process - the political activist coefficients changed in size by an absolute value of .03 or less. There were no changes in the coefficients' levels of statistical significance in any of these models except cultural liberalism, where the level of significance dropped from .00 to .01. In the case of religious pluralism, the JCRC coefficient was reduced to -.02 and remained statistically not significant. In both the ideology and anti-Semitism models, the coefficients for political activists and their levels of statistical significance were unchanged.

Regarding the donors, in two models - government size and services and church-state separation - the coefficients dropped below the .10 level of statistical significance and were each reduced in size by .02. In contrast, the donor coefficient in the model of cultural liberalism increased to .10 and became statistically significant at the .05 level. In all other cases, the coefficients remained statistically not significant and changed in size by an absolute value of .03 or less.

Not surprisingly, each of the other cities or counties exhibited some tendency to be more conservative than San Francisco. Dallas was more conservative on seven measures; Atlanta and Houston on six; Indianapolis, Philadelphia and Detroit on five; Milwaukee on four; and Cleveland, Seattle, Los Angeles, and Bergen County on three.

12. In other cases, political division may be the result of external social factors that cut across the group's objective status, such as class divisions that slice through ethnic or religious groups. The result is the same: the creation of subgroups within the larger social group, the members of whom are politically and socially cohesive among themselves but politically and socially divided from other subgroup members. We do not emphasize external, cross-cutting social factors theoretically or empirically in this study.

13. Additional methodologies, particularly qualitative ones, would allow further investigation of the lines of political divisions that the statistical analysis has identified and the implications of political division for political representation. See note 2.

* * *

References

Achen, C.H. 1975. "Mass Political Attitudes and the Survey Response." American Political Science Review 69:1218-1231.

Brady, H.E. 2000. "Contributions of Survey Research to Political Science." PS: Political Science and Politics 33(1):47-57.

Cohen, S.M. 1983a. American Modernity and Jewish Identity. New York: Tavistock.

Cohen, S.M. 1983b. "The 1981-1982 National Survey of American Jews." American Jewish Year Book 83: 89-110.

Cohen, S.M. 1989. The Dimensions of American Jewish Liberalism. New York: American Jewish Committee.

Cohen, S.M. and G.B. Bubis. 1990. "The Impact of Denomination: Differences in the Israel-Related Opinions of American Rabbis and Jewish Communal Workers." Jewish Political Studies Review 2(1-2):137-163.

Cohen, S.M. and C.S. Liebman. 1997. "American Jewish Liberalism: Unraveling the Strands." Public Opinion Quarterly 61:405-430.

Elazar, D.J. 1989. People and Polity: The Organizational Dynamics of World Jewry. Detroit: Wayne State University Press.

Elazar, D.J. 1995. Community and Polity: The Organizational Dynamics of American Jewry. Philadelphia: Jewish Publication Society.

Goldstein, S. and A. Goldstein. 1996. Jews on the Move: Implications for Jewish Identity. Albany, NY: State University of New York Press.

Gordon, M. 1964. Assimilation in American Life. New York: Oxford University Press.

Goren, A.A. 1999. The Politics and Public Culture of American Jews. Bloomington: Indiana University Press.

Kosmin, B., S. Goldstein, J. Waksberg, N. Lerer, A. Keysar, and J. Scheckner. 1991. Highlights of the CJF 1990 National Jewish Population Survey. New York: Council of Jewish Federations in association with the Mandell Berman Institute-North American Jewish Data Bank and The Graduate School and University Center, City University of New York.

Kotler-Berkowitz, L.A. 1995. "Political Division and Diversification in the American Jewish Community." Journal of Jewish Communal Service 71:266-274.

Kotler-Berkowitz, L.A. 1997. "Ethnic Cohesion and Division among American Jews: The Role of Mass-Level and Organizational Politics." Ethnic and Racial Studies 20:797-829.

Layman, G.C. 1997. "Religion and Political Behavior in the United States: The Impact of Beliefs, Affiliations, and Commitment from 1980 to 1994." Public Opinion Quarterly 61:288-316.

Levey, G.B. 1995a. "Review Article: The Liberalism of American Jews - Has It Been Explained?" British Journal of Political Science 26:369-401.

Levey, G.B. 1995b. "Toward a Theory of Disproportionate American Jewish Liberalism." Studies in Contemporary Jewry 11:64-85.

Liebman, C.S. 1973. The Ambivalent American Jew: Politics, Religion and Family in American Jewish Life. Philadelphia: Jewish Publication Society.

Lipset, S.M. and E. Raab. 1984. "The American Jews, the 1984 Elections, and Beyond. The Tocqueville Review 6:401-419.

Lipset, S.M. and E. Raab. 1995. Jews and the New American Scene. Cambridge and London: Harvard University Press.

Pitkin, Hanna Fenichel. 1967. The Concept of Representation. Berkeley: University of California Press.

Raab, E. and L. Sternberg. 1998. National Jewish Community Public Affairs Survey. New York and Waltham, MA: Jewish Council for Public Affairs and Nathan Perlmutter Institute, Brandeis University.

Sapiro, V., S.J. Rosenstone, W.E. Miller, and the National Election Studies. 1998. American National Election Studies, 1948-1997 (CD-ROM). ICPSR ed. Ann Arbor, MI: Inter-university Consortium for Political and Social Research (producer and distributor).

Sigelman, L. 1991. "Jews and the 1988 Election: More of the Same?" In J.L. Guth and J.C. Green, eds., The Bible and the Ballot Box. Boulder, CO: Westview Press, pp. 188-203.

Wald, K. 1997. Religion and Politics in the United States. Washington, D.C.: Congressional Quarterly Press.

Weber, M. 1946. "Class, Status, Party." In H.H. Gerth and C.W. Mills, eds., From Max Weber: Essays in Sociology. New York: Oxford University Press, pp. 180-195.

Zuckerman, A.S. 1989. "The Bases of Political Cohesion: Applying and Reconstructing Crumbling Theories." Comparative Politics 21: 473-495.

Zuckerman, A.S. 1990. "The Structural Sources of Cohesion and Division in the American Jewish Community." In Michael Shapiro, ed., Divisions Between Traditionalism and Liberalism in the American Jewish Community. Lewiston, ME: Edwin Mellen Press, pp. 15-34.

* * *

LAURENCE A. KOTLER-BERKOWITZ is Lady Davis Postdoctoral Research Fellow, Department of Political Science, Hebrew University. His articles have been published in the British Journal of Political Science (forthcoming) and Ethnic and Racial Studies, and his co-authored works have appeared in Comparative Political Studies and the European Journal of Political Research. His current research focuses on political cohesion and division within ethnic, religious, and racial groups.

LAWRENCE STERNBERG is Director of the Nathan Perlmutter Institute for Jewish Advocacy, a member of the faculty of the Hornstein Program in Jewish Communal Service, and Associate Director of the Maurice and Marilyn Cohen Center for Modern Jewish Studies at Brandeis University. He is co-editor of Changing Jewish Life: Service Delivery and Planning in the 1990s (Greenwood Press) and has published articles on a variety of topics, including Jewish public opinion and political activism, intermarriage, Jewish identity in America, and Jewish social policy.

The opinions expressed herein do not necessarily reflect those of the Board of Fellows of the Jerusalem Center for Public Affairs.

The above essay appears in the Spring 2000 issue of the Jewish Political Studies Review, the first and only journal dedicated to the study of Jewish political institutions and behavior, Jewish political thought, and Jewish public affairs.

Published by the Jerusalem Center for Public Affairs (http://www.jcpa.org/), the JPSR appears twice a year in the form of two double issues, either of a general nature or thematic, with contributors including outstanding scholars from the United States, Israel, and abroad.

Jewish Political Studies Review

ORDER FORM

Invoice No. ________

Date _______________

Annual Subscription Rates:

Individual Institutions
& Libraries Students

Outside Israel: $26 $40 $20

In Israel: NIS 70 NIS 110 NIS 40

Back Issues or Single issues - $12 each

Enclosed is my check for US$/NIS: ____________

Name: _____________________________________________________________

Address: _____________________________________________________________

_____________________________________________________________

All checks should be made payable and mailed to:

IN THE US:
Center for Jewish Community Studies
Baltimore Hebrew University
5800 Park Heights Avenue
Baltimore, MD 21215 IN ISRAEL:
Jerusalem Center for Public Affairs
13 Tel Hai Street
Jerusalem 92107 ISRAEL

Jewish Political Studies Review

Laurence A. Kotler-Berkowitz and Lawrence Sternberg

Data

Table 1: Comparison Of 1997 PAS Respondents and 1990 NJPS Reference Group

Measures

Table 2: Institutional Affiliations of 1997 Pas Respondents

Political Cohesion and Division: Empirical Findings

Table 3: Political Ideologies of American Jews and Americans Generally

Table 4: Mean Scores on Dependent Variablesa and Difference of Means Testb

Table 5: Group-by-Group Difference of Means across Dependent Variables

Table 6: Multivariate Regression: Determinants of Political Cohesion and Division

Assessment: Cohesion, Division, and Political Representation

Conclusion

Appendix: Variable Construction

Dependent Variables

Independent Variables

Notes

References

Jewish Political Studies Review

ORDER FORM

Table 2: Institutional Affiliations of
1997 Pas Respondents

Table 4: Mean Scores on Dependent Variables^a and Difference of Means Test^b