¹ See Rokeach, Milton, Beliefs. Attitudes, and Values (San Francisco: Jossey-Bass, 1970)Google Scholar, and Rokeach, Milton and Ball-Rokeach, Sandra J., ‘Stability and Change in American Value Priorities: 1968–1991’, American Psychologist, 44 (1989), 775–84.CrossRefGoogle Scholar

² Converse, Philip E., ‘The Nature of Belief Systems in Mass Publics’, in Apter, David, ed., Ideology and Discontent (New York: Free Press, 1964), pp. 206–61.Google Scholar

³ Kluckhohn, Clyde et al. , ‘Values and Value Orientations in the Theory of Action’, in Parsons, Talcott and Shils, Edward A., eds, Toward a General Theory of Action (Cambridge, Mass.: Harvard University Press, 1951), pp. 388–433, at p. 395.Google Scholar

⁴ See, for example, Feldman, Stanley, ‘Structure and Consistency in Public Opinion: The Role of Core Beliefs and Values’, American Journal of Political Science, 32 (1988), 416–416;CrossRefGoogle Scholar and Hurwitz, Jon and Peffiey, Mark, ‘How Are Foreign Policy Attitudes Structured? A Hierarchical Model’, American Political Science Review, 81 (1987), 1099–120.CrossRefGoogle Scholar

⁵ See Rokeach, , Beliefs, Attitudes and ValuerGoogle Scholar; Rokeach, Milton, The Nature of Human Values (New York: Free Press, 1973)Google Scholar; Rokeach, Milton, ed., Understanding Human Values (New York: Free Press, 1979)Google Scholar, and Rokeach, and Ball-Rokeach, , ‘Stability and Change in American Value Priorities’Google Scholar; see Tetlock, Philip E.'s value pluralism model of ideological reasoning, in ‘A Value Pluralism Model of Ideological Reasoning’, Journal of Personality and Social Psychology, 50 (1986), 819–27CrossRefGoogle Scholar; and his ‘Structure and Function in Political Belief Systems’, in Praktanis, Anthony R., Breckler, Steven J. and Greenwald, Anthony G., eds. Attitude Structure and Function (Hillsdale, NJ: Lawrence Erlbaum, 1989), pp. 129–51Google Scholar; Schwartz, Shalom H. and Bilsky, Wolfgang, ‘Toward a Universal Psychological Structure of Human Values’, Journal of Personality and Social Psychology, 53 (1987), 550–62;CrossRefGoogle Scholar see also Alwin, Duane F. and Krosnick, Jon A., ‘The Measurement of Values in Surveys: A Comparison of Ratings and Rankings’, Public Opinion Quarterly, 49 (1985), 535–535CrossRefGoogle Scholar; and Krosnick, Jon A. and Alwin, Duane F., ‘A Test of the Form-Resistant Correlation Hypothesis: Ratings, Rankings and the Measurement of Values’, Public Opinion Quarterly, 52 (1988), 526–38.CrossRefGoogle Scholar

⁶ For the distinction between valence and position dimensions of politics, see Stokes, Donald E., ‘Spatial Models of Party Competition’, American Political Science Review, 57 (1963), 368–77.CrossRefGoogle Scholar

⁷ For example, see Cohen, Erik, ‘Citizenship, Nationality and Religion in Israel and Thailand’, in Kimmerling, Baruch, ed., The Israeli State and Society: Boundaries and Frontiers (Albany: SUNY Press, 1989), pp. 66–92Google Scholar; Cohen, Erik, ‘The Changing Legitimations of the State of Israel’, Studies in Contemporary Jewry, 5 (1989), 148–65Google Scholar; Smooha, Sammy, ‘Minority Status in an Ethnic Democracy: The Status of the Arab Minority in Israel’, Ethnic and Racial Studies, 13 (1990), 389–413.CrossRefGoogle Scholar

⁸ The references in this article to greater Israel include the territories of the West Bank of the Jordan River taken in the 1967 Six Days war, referred to by the Likud government as Judea and Samaria. Also taken in that war were the Golan Heights and the Gaza Strip.

⁹ Each of these are here considered ultimate values, or ends in themselves, although it is conceivable that each of them may be instrumental for some respondents, serving as a means for achieving another value. For example, for many on the right and among religious Jews (but not only them), greater Israel is an end in itself. For others, it is a means for achieving peace and security, or for absorbing more immigrants.

¹⁰ The constitution is not legally binding, but it has great symbolic importance. See Rubinstein, Amnon, The Constitutional Law of the State of Israel, 4th edn (Jerusalem: Schocken [Hebrew], 1991)Google Scholar, chap. 1.

¹¹ In a 1988 survey, 97 per cent said that it was important or very important that the Jewish character of the state be preserved (that there be a Jewish majority) (Katz, Elihu, ‘Forty-nine percent lean towards “transfer” of Arabs’, Jerusalem Post International Edition, 20 08 1988Google Scholar). See also Shapiro, Yonathan, Democracy in Israel (Ramat Gan: Massada [Hebrew], 1977)Google Scholar, chap. 1; Liebman, Charles S. and Don-Yehiya, Eliezer, Civil Religion in Israel: Traditional Judaism and Political Culture in the Jewish State (Berkeley: University of California Press, 1983).Google Scholar

¹² In the survey cited in the previous note, 82 percent said that it was important or very important that the democratic character of the state be maintained (i.e., that every resident have equal rights).

¹³ Harkabi, Yehoshafat, Israel's Fateful Decisions (London: I. B. Tauris, 1988).Google Scholar

¹⁴ See Horowitz, Dan and Lissak, Moshe, Origins of the Israeli Polity: Palestine under the Mandate (Chicago: University of Chicago Press, 1978)Google Scholar, chap. 6; Gorny, Yosef, Zionism and the Arabs, 1882–1948 (Oxford: Clarendon Press, 1987)Google Scholar; and Kimmerling, Baruch, Zionism and Territory (Berkeley: University of California Press, 1983).Google Scholar

¹⁵ Tetlock emphasizes how value conflicts and the trade-offs they require are unpleasant (‘A Value Pluralism Model of Ideological Reasoning’, p. 819Google Scholar). On the psychological level they are unpleasant for cognitive, motivational and affective reasons. Politically, they are unappealing in that the compromises they demand are difficult to justify to the public and to political activists, in particular to those whose value priorities have been forsaken.

¹⁶ The surveys were conducted under the auspices of the National Security and Public Opinion Project of the Jaffee Center for Strategic Studies at Tel-Aviv University, directed by Asher Arian. The 1988 and 1990 surveys were also supported by a grant from the Sapir Center for Development at Tel-Aviv University. Questionnaires were composed and data analysed by the authors. All field work was done by the Dahaf Research Institute.

The interview period in 1990 stretched out for six months in an effort to achieve a national sample with a maximum number of reinterviews. An unanticipated consequence of this protracted period was that 54 per cent of the interviews were conducted before 2 August, the day of the Iraqi invasion of Kuwait, and 46 per cent afterward. For the values discussed here and for demographic variables, the two halves of the survey were very similar, attitudinal differences existed especially regarding perceived threat, use of nonconventional and nuclear weapons, and assessment of the United States. For a discussion, see Arian, Asher, ‘Security and Political Attitudes: The Influence of the Gulf War’, in War in the Gulf: Implications for Israel, Report of a Jaffee Center for Strategic Studies Group (Tel Aviv: Papyrus, 1991).Google Scholar

¹⁷ The notion of a Jewish state was operationalized as a ‘Jewish majority’ in Israel, which is closest to the original Zionist idea, as well as to common usage in Israel (although competing specifications of a Jewish state, such as a state where Jewish culture and tradition are dominant, or a theocratic state where the Torah is the law of the land, are also possible). The Hebrew term for greater Israel is commonly recognized and widely used in political discourse. Democracy was conceived of as equal political rights for all, including the franchise to Arabs under Israel's jurisdiction. Peace was explained as a situation in which there was a low probability of war, so as to clearly set it apart from the more formal aspect of a peace agreement.

¹⁸ Kimmerling, Baruch, ‘Between the Primordial and the Civil Definitions of the Collective Identity: “Eretz Israel” or the State of Israel?’, in Cohen, Erik et al. , eds. Comparative Social Dynamics (Boulder, Colo.: Westview, 1985), p. 272Google Scholar; see also Shafir, Gershon, ‘Ideological Politics or the Politics of Demography: The Aftermath of the Six-Day-War’, in Lustick, Ian S. and Rubin, Barry, eds, Critical Essays on Israeli Society. Politics and Culture (Albany, NY: SUNY Press, 1991), p. 52.Google Scholar

¹⁹ The value rankings ranged from 1, the value mentioned last, to 4, the most highly ranked value. Although they are clearly ordinal-level measures, we treat these rank orders throughout the analysis as interval-level variables, following common practice; see Alwin, Duane F. and Krosnick, Jon A., ‘The Measurement of Values in Surveys: A Comparison of Ratings and Rankings’, Public Opinion Quarterly, 49 (1985), 535–535CrossRefGoogle Scholar; O'Brien, Robert M., ‘The Use of Pearson's R with Ordinal Data’, American Sociological Review, 44 (1979), 851–7.CrossRefGoogle Scholar

The standard deviation for the Jewish majority was the smallest in both general surveys, as well as for the panel. The differences in the standard deviations of the four value rankings in the general surveys ranged between 0.92 and 1.16 in 1988, and between 0.99 and 1.10 in 1990. In the panel data, these figures were 0.87 and 1.14.

²⁰ Even the classic valence dimension of economic stability and development may become a position dimension when juxtaposed with a competing value, such as the environment; see Inglehart, Ronald, Culture Shift in Advanced Industrial Society (Princeton, NJ: Princeton University Press, 1990).Google Scholar

²¹ Applying a paired t-test to the responses of the panel showed that the difference in the mean scores of the 1988 and 1990 rankings was higher (and statistically significant) for Jewish majority and peace, while it was smaller and not significant for the two lower-ranked and conflictual values, greater Israel and democracy.

²² The right-left self-placement scale ranged from 1 to 7, with 1 being extreme right and 7 extreme left. As is commonly done, we regarded it as interval level. On the right-left scale, see Inglehart, Ronald and Klingemann, Hans D., ‘Party Identification, Ideological Preference and the Left-Right Dimension among Western Mass Publics’, in Budge, Ian et al. , eds. Party Identification and Beyond (London: Wiley, 1976), pp. 243–73Google Scholar; Arian, Asher and Shamir, Michal, ‘The Primarily Political Function of the Left-Right Continuum’, Comparative Politics, 15 (1983), 139–58CrossRefGoogle Scholar; and Inglehart, , Culture Shift in Advanced Industrial Society.Google Scholar

²³ We used Pearson's r correlations to indicate the degree and in what direction the variables covary. We did not assume one-way causality: right-left self-identification and value priorities both measure to an extent people's political and ideological world-view. Since correlations (as other standardized measures) are a function of both the covariance and the variances of the variables, we wish to emphasize that the relative order of the correlations as discussed in the text held also when the differences in the value rankings' variances were taken into account (see fn. 19).

²⁴ See Conover, J. Pamela and Feldman, Stanley, ‘The Origin and Meaning of Liberal/Conservative Self-Identifications’, American Journal of Political Science, 25 (1981), 617–45CrossRefGoogle Scholar; Kerlinger, Fred N., Liberalism, Conservatism and the Structure of Social Attitudes (Hillsdale, NJ: Lawrence Erlbaum, 1984)Google Scholar; Ventura, Raphael and Shamir, Michal, ‘Left and Right in Israeli Polsities’, State and Government [Hebrew], 35 (1992), 21–50.Google Scholar

²⁵ Right-wing settler groups argued about whether or not they should make explicit their preference for greater Israel over peace. Sensing that peace was highly valued, many were not willing to concede in their public relations campaigns that there was a possible trade-off between land and peace (see Harel, Israel, Haaretz, 6 09 1991).Google Scholar Only at the time of the 1991 peace conference did spokesmen of the extreme right begin to speak in terms that might move peace from being a ‘valence’ to a ‘position’ dimension.

²⁶ All correlations among these rankings are negative as the sum of the rankings for each respondent is constant, but they vary of course in their strength. See, Cattell, Raymond B., ‘Psychological Measurement: Ipsative, Normative and Interactive’, Psychological Review, 51 (1944), 292–303CrossRefGoogle Scholar; Alwin, and Krosnick, , ‘The Measurement of Values in Surveys’, Public Opinion Quarterly, 49 (1985), 535–52.CrossRefGoogle Scholar

²⁷ These value packages are ideologically constrained, notwithstanding their political roots. Land and people are basic notions in all particularistic, nationalist, right-wing ideologies, and in Israel they are also tied to religious tradition. The recognition of the rights of others stems from universalistic standards, and brings together democracy and the willingness to compromise. In a study of the meanings attached to left and right in Israel, Ventura and Shamir showed that the issue of the territories is the major conflict dimension in contemporary Israeli politics, but that right-wingers define it more often with reference to the land, whereas on the left there is more reference to the Palestinians or Arabs, that is to the people and not to the land (see Ventura, and Shamir, , ‘Left and Right in Israeli Polities’, pp. 39–40).Google Scholar

²⁸ Shamir, Michal, ‘Realignment in the Israeli Party System’, in Arian, Asher and Shamir, Michal, eds., The Elections in Israel – 1984 (New Brunswick, NJ: Transaction, 1986), pp. 267–96Google Scholar; Arian, Asher, Shamir, Michal and Ventura, Raphael, ‘Public Opinion and Political Change: Israel and the Intifada’. Comparative Politics, 24 (1992), 317–34.CrossRefGoogle Scholar

²⁹ The variable was constructed from two questions, the first asking preference with a ‘Leave as is’ option, and a follow-up forcing a choice for respondents who chose the ‘Leave as is’ option (see Arian, Asher, ‘Security and Political Attitudes in Israel’, Public Opinion Quarterly, 56 (1992), 116–28).CrossRefGoogle Scholar The responses in 1987,1988 and 1990 respectively were: (A) Territories for peace, 31.8, 39.5, 43.9; (B) ‘A’ on probe, 10.8, 9.9, 6.8; (C) Leave as is, 2.5, 2.1, 2.8; (D) ‘E’ on probe, 25.7, 22.5, 16.8; (E) Annex, 28.3, 26.0, 29.8. The effective N in each year was 1,081, 846 and 1,235.

³⁰ The alpha reliability coefficients for the policy scale in the national samples were 0.72, 0.80 and 0.75 in 1987, 1988 and 1990 respectively. In the panel data the corresponding figures were 0.69, 0.81 and 0.77. We used correction for attenuation formulas to check whether the changes in the reliability of the policy scale over time altered any of our substantive conclusions in this section and the next one. All estimates of R and R² increased of course, but their relative magnitude did not change. For more complete discussions of the policy scale and its dimensionality, see Arian, Asher, ‘A People Apart: Coping with National Security Problems in Israel’, Journal of Conflict Resolution, 33 (1989), 605–31CrossRefGoogle Scholar, and Arian, Asher, Shamir, Michal and Ventura, Raphael, ‘Public Opinion and Political Change: Israel and the Intifada’, Comparative Politics, 24 (1992), 317–34.CrossRefGoogle Scholar

³¹ We present the results for the panel respondents in addition to the general samples in order to establish their equivalence, as in later analyses we depend more heavily on the panel data.

³² The relative order of the correlations as discussed in the text held also when the differences in the value rankings' variances were taken into account (see fn. 19).

³³ A proper interpretation of the R ²S requires us to examine the variances of the variables, and these variances of the independent variables measured at different time points were all very similar. There was also no difference among the variances of the territories question at the three points in time. On the policy scale, the variance is somewhat higher for 1988 and 1990 than in 1987, peaking in 1988. The effect of these differences runs in the opposite direction of the results, thus presenting no interference with the validity of our interpretation. The regressions used only cases with valid values on all variables.

The other attitudinal factors included in the regression analysis were constructs which have been shown to have an effect on policy positions in the Israeli context: God-and-us, go-it-alone, threat, and overcome; see Arian, Asher, Talmud, Ilan and Hermann, Tamar, National Security and Public Opinion in Israel (Boulder, Colo.: Westview, 1988).Google Scholar The God-and-us construct relates to the special, mystical relation perceived by many between God, Israel and Jewish history, the go-it-alone construct to feelings of isolation and to the belief that ultimately Jewish destiny depends on the Jews.

The two constructs were related but shown to be different; the God-and-us construct statements were supported by more than two-thirds of the respondents, and the statements of the go-it-alone construct by about half. The God-and-us scale was comprised of Likert-type responses to statements such as ‘The God of Israel will not lie’, ‘Massada will not fall again’, and ‘The guardian of Israel will neither slumber nor sleep’. Go-it-alone statements included ‘World criticism of Israeli policy stems mainly from antisemitism’, and ‘Israel is and will continue to be “A people dwelling alone”’. In addition, scales measuring perception of threat and the ability of Israel to overcome a series of challenges were used. See Arian, Asher, ‘A People Apart’, Journal of Conflict Resolution, 33 (1989), 605–31.CrossRefGoogle Scholar

The last stage in both stepwise analyses contained all important socio-economic variables: education, ethnic background, income, density of living conditions, religiosity, age and sex. In addition, we included the psychological construct of dogmatism, shown in previous research to have considerable impact on positions towards the territories and related attitudes. See Shamir, Michal and Sullivan, John L., ‘The Political Context of Tolerance: The United States and Israel’, American Political Science Review, 77 (1983), 911–28.CrossRefGoogle Scholar

All attitudinal constructs and dogmatism were measured by scales; for the listing of the items and the alpha coefficients of the scales, see Arian, , ‘A People Apart’.Google Scholar The socio-economic and psychological variables were considered causally prior to the attitudinal predictors of policy position, and were therefore entered in the third and final stage.

³⁴ A problem of interpretation would arise if the importance of values would be low in 1987 and high in 1988 and 1990, and the opposite pattern were to hold for the God-and-us and go-it-alone constructs. The charge of deficient measurement would be plausible and hard to dismiss. But, as can be seen in Tables 3 and 4, this was not the case.

³⁵ While the value rankings account for similar proportions of the variance of the policy scale and of the territories question, the other variables are able to account for more of the policy scale's variation than of the territories question. The R ² for the policy scale range in the 0.40s and one even exceeds 0.60. On average, our regression equations ‘explain’ almost half of the variance in the policy scale. The R ² for the territories question vary between 0.24 and 0.46, averaging 0.34.

The figures in Table 3 are R ²s. Since we have fifteen predictors in the full equations, and the panel samples were quite small, it is wise to check the adjusted R ²s. They are lower by 0.01–0.02 for the estimates of the value rankings' contributions (rows 2 and 3 in Table 3). The two averages, when we take the adjusted R ²s into account, are 0.22 and 0.10, instead of 0.23 and 0.11. As to the total R ²s, the corresponding adjusted R ²s are 0.37,0.56 and 0.39 for the policy scale in the three panel samples, and 0.15, 0.39 and 0.30 for the territories question.

³⁶ See Markus, Gregory B., Analyzing Panel Data (Beverly Hills, Calif.: Sage Series, Quantitative Applications in the Social Sciences, no. 18, 1979), pp. 45–8CrossRefGoogle Scholar; and Kessler, Ronald C. and Greenberg, David F., Linear Panel Analysis: Models of Quantitative Change (New York: Academic Press, 1981).Google Scholar

³⁷ In the first stage, the lagged dependent variable was regressed on all the variables included in the cross-sectional equations at t and t + 1. The predicted values from this first stage then served as the instrumental variable, replacing the actual lagged dependent variable in the second stage of the regression analysis. This two-stage least squares procedure produces consistent estimators for the coefficients in the model.

³⁸ The socio-demographic and psychological variables were not included in these equations, as they were assumed to be constant; indirectly they have an effect through the lagged policy position.

For the go-it-alone and God-and-us constructs the 1987 scores were used as indicators for 1988 values.

³⁹ The adjusted R ²s are lower than the raw R ²s by 0.01–0.02 for the value rankings' contributions. The adjusted R ²s for the full equations are 0.51, 0.36, 0.34 and 0.28 (first row of Table 4).

⁴⁰ Arian, , Shamir, and Ventura, , ‘Public Opinion and Political Change’.Google Scholar

⁴¹ Ajzen, Icek and Fishbein, Martin, ‘Attitude-Behavior Relations: A Theoretical Analysis and Review of Empirical Research’, Psychological Bulletin, 84 (1977), 888–918CrossRefGoogle Scholar; Ajzen, Icek, ‘Attitude Structure and Behavior’, in Praktanis, Anthony R., Breckler, Steven J. and Greenwald, Anthony G., eds. Attitude Structure and Function (Hillsdale, NJ: Lawrence Erlbaum, 1989), pp. 241–74.Google Scholar

⁴² Sniderman, Paul M. and Tetlock, Philip E., ‘Interrelationship of Political Ideology and Public Opinion’, in Hermann, Margaret G., ed., Political Psychology (San Francisco: Jossey-Bass, 1986), 62–96, at p. 75.Google Scholar