1 Though the number of scholars who have addressed the question is small, the evidence is compelling. See Meisel, John, “Religious Affiliation and Electoral Behaviour: A Case Study,” in John, Courtney, ed., Voting in Canada (Toronto: Prentice-Hall, 1967), 144–61Google Scholar; Meisel, John, “Bizarre Aspects of a Vanishing Act: The Religious Cleavage and Voting in Canada,” in John, Meisel, ed., Working Papers on Canadian Politics (2d ed.; Montreal: McGill-Queen's University Press, 1975), 253–84Google Scholar; Irvine, William P., “Explaining the Religious Basis of the Canadian Partisan Identity: Success on the Third Try,” this Journal 7 (1974), 560–63Google Scholar; Irvine, William and Gold, H., “Do Frozen Cleavages Ever Go Stale? The Bases of the Canadian and Australian Party Systems,” British Journal of Political Science 10 (1980), 187–218CrossRefGoogle Scholar; Johnston, Richard, “The Reproduction of the Religious Cleavage in Canadian Elections,” this Journal 18 (1985), 99–113Google Scholar; and Johnston, Richard, “The Geography of Class and Religion in Canadian Elections,” in Joseph, Wearing, ed., The Ballot and Its Message: Voting in Canada (Mississauga: Copp Clark Pitman, 1991)Google Scholar. Canadian studies addressing religious cleavages other than the Catholic/Protestant division are rarer still. For a notable exception, see Laponce, Jean, “Left or Centre? The Canadian Jewish Electorate, 1953–1983,” this Journal 21 (1988), 691–714.Google Scholar

2 Johnston, “The Reproduction of the Religious Cleavage,” 99.

3 See Johnston, Richard, Blais, André, Brady, Henry and Crête, Jean, Letting the People Decide: Dynamics of a Canadian Election (Montreal: McGill-Queen's University Press, 1992), 37–38Google Scholar; and Beck, J. Murray, Pendulum of Power: Canada's Federal Elections (Scarborough: Prentice-Hall, 1968), esp. 46–55.Google Scholar

4 Meisel, “Religious Affiliation and Electoral Behaviour,” and “Bizarre Aspects of a Vanishing Act.”

5 Irvine, “Explaining the Religious Basis.”

6 Lipset, Seymour Martin and Rokkan, Stein, Party Systems and Voter Alignments: Cross National Perspectives (New York: Free Press, 1967)Google Scholar; Russel, Dalton, Paul Allen, Beck and Scott, Flanagan, eds., “Electoral Change in Advanced Industrial Democracies,” in Dalton, , Beck, and Flanagan, , eds., Electoral Change in Advanced Industrial Democracies: Realignment of Dealignment? (Princeton: Princeton University Press, 1984) 3–25.Google Scholar

7 See, for example, Cassel, Carole A., “A Test of Converse's Theory of Party Support,” Journal of Politics 55 (1993), 664–81CrossRefGoogle Scholar; and Martinez, Michael D., “Intergenerational Transfer of Canadian Partisanships,” this Journal 17 (1984), 133–43.Google Scholar

8 Johnston, “The Reproduction of the Religious Cleavage.”

9 Johnston admits, though, that his evidence is more persuasive for the geographic than for the class explanation (see Johnston, “The Geography of Class and Religion”).

10 Gidengil, Elisabeth, “Canada Votes: A Quarter Century of Canadian National Election Studies,” this Journal 25 (1992), 230Google Scholar. Recent work has underlined the importance of examining opinions as patterns of interaction, rather than the property of atomistic individuals. See Huckfeldt, Robert and Sprague, John, Citizens, Politics, and Social Communication: Information and Influence in an Election Campaign (Cambridge: Cambridge University Press, 1995).CrossRefGoogle Scholar

11 Shafer, Byron E., ed., The End of Realignment? Interpreting American Electoral Eras (Madison: University of Wisconsin Press, 1991)Google Scholar; and Kellstedt, Lyman, “Religion, The Neglected Variable: An Agenda for Future Research on Religion and Political Behavior,” in Leege, David C. and Lyman, Kellstedt, eds., Rediscovering the Religious Factor in American Politics (Armonk, N.Y.: M. E. Sharpe, 1993).Google Scholar

12 In Johnston's identification of the Catholic ethos, he acknowledges that his evidence is incomplete: “The estimations are kind to our hypotheses, but glaring exceptions exist”; “our geographic picture although impressively powerful, is still not entirely coherent”; and one part of the results “makes little sense” (Johnston, “The Geography of Class and Religion,” 122, 127, 126).

13 Iyengar, Shanto and Kinder, Donald R., News that Matters: Television and American Opinion (Chicago: University of Chicago Press, 1987)Google Scholar; Iyengar, Shanto, Is Anyone Responsible? How Television Frames Political Issues (Chicago: University of Chicago Press, 1991)CrossRefGoogle Scholar; Mendelsohn, Matthew, “The Media's Persuasive Effects: The Priming of Leadership in the 1988 Canadian Election,” this Journal 27 (1994), 81–98Google Scholar; and Mendelsohn, Matthew, “The Media and Interpersonal Communications: The Priming of Issues, Leaders, and Party Identification,” Journal of Politics 58 (1996), 112–26.CrossRefGoogle Scholar

14 See, for example, Gerbner, George, Gross, Larry, Morgan, Michael and Signorielli, Nancy, “Growing Up with Television: The Cultivation Perspective,” in Jennings, Bryant and Dolf, Zillmann, eds., Media Effects: Advances in Theory and Research (Hillsdale, N.J.: Laurence Erlbaum, 1994), 17–41Google Scholar; and Morgan, Michael and Signorielli, Nancy, “Cultivation Analysis: Conceptualization and Methodology,” in Nancy, Signorielli and Michael, Morgan, eds., Cultivation Analysis: New Directions in Media Effects Research (Newbury Park: Sage, 1990), 13–34Google Scholar. One researcher from the cultivation analysis school has argued that religiosity decreases amongst heavy TV viewers (see Stewart Hoover, “Television, Religion, and Religious Television: Purposes and Cross Purposes,” in ibid., 123–40).

15 Gerbner et al., “Growing Up with Television,” 28.

16 Altheide, David, Media Power (Beverly Hills: Sage, 1985).Google Scholar

17 McManus, John, Market-Driven Journalism (Thousand Oaks: Sage, 1994).Google Scholar

18 Iyengar, Is Anyone Responsible?

19 Edelman, Murray, Constructing the Political Spectacle (Chicago: University of Chicago Press, 1988).Google Scholar

20 Zaller, John, The Nature and Origins of Mass Opinion (New York: Cambridge University Press, 1992), 39, 111.CrossRefGoogle Scholar

21 The data were collected by the Institute for Social Research at York University, Toronto, for the CES investigators: Richard Johnston, André Blais, Henry Brady and Jean Crête. Overall, 3,609 interviews were completed during the campaign, with an estimated response rate of 57 per cent (for more information, see Johnston et al., Letting the People Decide, Appendix A). We have selected the 1988 data set because it contains well-constructed items on our questions of interest, including an item which measures frequency of church attendance, a variable not included in the 1993 CES.

22 Catholics are defined as those whose self-identified denomination is Catholic, and does not refer to Church attendance or other possible measures.

23 Various operationalizations of the MEDIA variable were attempted, and all produced similar results. The use of a variable that exclusively measures television exposure, as we use, is consistent with the usual practice of cultivation researchers investigating “mainstreaming,” and hence was retained.

24 On issues related to estimation and interpretation of regression models including multiplicative terms see Althauser, R. P., “Multicollinearity and Non-additive Regression Models,” in Blalock, H. M. Jr., ed., Causal Models in the Social Sciences (Chicago: Aldine-Atherton, 1971), 453–72Google Scholar; Cronbach, L., “Statistical Tests for Moderator Variables: Flaws in Analysis Recently Proposed,” Psychological Bulletin 102 (1987), 414–17CrossRefGoogle Scholar; Friedrich, Robert, “In Defense of Multiplicative Terms in Multiple Regression Equations,” American Journal of Political Science 26 (1982) 797–833CrossRefGoogle Scholar; Jaccard, James, Turrisi, Robert and Wan, Choi K., Interaction Effects in Multiple Regression (Newbury Park: Sage, 1990)Google ScholarPubMed; Aiken, Leona S. and West, Stephen G., Multiple Regression: Testing and Interpreting Interactions (Newbury Park: Sage, 1991)Google Scholar; and Pindyck, Robert S. and Rubenfeld, Daniel L., Econometric Models and Economic Forecasts (London: McGraw-Hill, 1981).Google Scholar

25 The exclusion of francophones from the sample in no way changed the results. We have avoided excluding particular groups of Catholics where possible in order to avoid decreasing the sample size.

26 Catholics are of course a heterogeneous group. Regional and ethnic differences between Catholics are important, with these two variables often corresponding to one another, as well as corresponding to various waves of Catholic immigration (for example, many Irish and Scottish Catholics settled in the Maritimes during the wave of immigration in the late eighteenth and first half of the nineteenth centuries). Regional and ethnic dummy variables were thus entered in the various equations, as were interactives created by multiplying CATH by these regional and ethnic dummies. We could find no differences between various groups of Catholics on the questions of importance. We thus focus our analysis on the larger picture and treat Catholics outside Quebec as one group, hypothesizing that they possess a “distinct Catholic ethos” and that this ethos is weakened by exposure to the mass media.

27 See, for instance, Kellstedt, Lyman and Green, John C., “Knowing God's Many People: Denominational Preference and Political Behavior,” in David, Leege and Lyman, Kellstedt, eds., Rediscovering the Religious Factor in American Politics (Armonk, N.Y.: M. E. Sharpe, 1993)Google Scholar, Table 3.1, 62.

28 The 1991 Canadian census reports that 31.7 per cent of Canadians outside Quebec are Catholic (6,474,045), while 46.0 per cent are Protestant (9,381,980). In the CES sample, 31.3 per cent of respondents outside Quebec are Catholic (757), while 48.6 per cent are Protestant (1176). These numbers indicate that our sample is representative of the population, though Protestants are slightly over-represented and Others are slightly underrepresented in the CES. This small discrepancy should not affect our results.

29 See Wald, Kenneth D., Religion and Politics in the United States (2d ed.; New York: St. Martin's, 1992)Google Scholar, chap. 8; and Castles, Francis, “On Religion and Public Policy: Does Catholicism Make a Difference?” European Journal of Political Research 25 (1994), 19–40.CrossRefGoogle Scholar

30 Our expectations concerning our variables are based on the political socialization of Catholics in Canada, not the religious socialization of Catholics generally, and might be different expectations for, say, Catholics in Italy.

31 See Byrnes, Timothy, “The Politics of Abortion: The Catholic Bishops,” in Timothy, Byrnes and Mary, Segers, eds., The Catholic Church and the Politics of Abortion (Boulder: Westview, 1992)Google Scholar, for a description of Catholic bishops leadership on the abortion issue.

32 See Stortz, Gerald, “Serving the Poor: J. J. Lynch and Our Lady of Lourdes Parish,” in John, Duggan and Terry, Fay, eds., Spiritual Roots: Historical Essays on the Roman Catholic Archdiocese of Toronto at 150 Years of Age (Toronto: Our Lady of Lourdes, 1991), 46–72.Google Scholar

33 Gilley, Sheridan, “Labour and the Catholic Church: A Tale of Three Cardinals,” in Terrence, Murphy and Cyril, Byrne, eds., Religion and Identity: The Experience of Irish and Scottish Catholics in Atlantic Canada (St. John's: Jesperson, 1992), 128–43.Google Scholar

34 See various statements from the Canadian Conference of Catholic Bishops (CCCB) on the need to support unions, particularly Documents 8–14, 58, 59, in Sheridan, E. F., ed., Do Justice: The Social Teaching of the Canadian Catholic Bishops (Sherbrooke: Éditions Paulines, 1987).Google Scholar

35 The choice of the dependent variable is supported by the fact that attitudes toward unions have been identified as the most important component of class politics in Canada (Johnston et al., Letting the People Decide, 97–99).

36 See CCCB Documents 52–54, 57, in Sheridan, ed., Do Justice.

37 The purchase of nuclear-powered submarines was the only variable in the data set that measured support for the military, and thus our range of choices for the dependent variable was constrained. Nonetheless, we are comfortable with this choice because military spending on large projects is consistently the highest profile defense issue (Environics Focus Canada Reports, 1984–1994 [Kingston: Centre for the Study of Public Opinion, Queen's University]).

38 Johnston, “The Geography of Class and Religion.”

39 See, for example, Clarke, Harold, Jenson, Jane, Leduc, Lawrence and Pammett, Jon, Absent Mandate: Interpreting Change in Canadian Elections (2d ed.; Toronto: Gage, 1991).Google Scholar

40 See, for example, Johnston et al., Letting the People Decide.

41 An earlier version of this article examined the differences between Catholics and Others (with no respondents excluded) and all findings were virtually identical.

42 See Aldrich, John H. and Nelson, Forrest D., Linear Probability, Logit, and Probit Models (Beverly Hills: Sage, 1984)CrossRefGoogle Scholar; and Demaris, Alfred, Logit Modelling (Newbury Park: Sage, 1992).CrossRefGoogle Scholar

43 The relative size of the maximum likelihood coefficients for the main cleavages and the interactive terms was virtually indistinguishable from the OLS estimates. The MLE coefficients for the four equations were: Liberal Vote: Catholic 1.17(.31)^a (where^a and^b indicate significance levels as defined in Table 2), Catholic × Media −1.01(.41)^a; Military Spending: Catholic .38(.11)^a, Catholic × Media −.35(.16)^a; Unions: Catholic .44(.12)^a, Catholic × Media −.46(.16)^a; Abortion: Catholic .53(.11)^a, Catholic × Media −.28(.15)^b. For work highlighting the robustness of OLS regression, see Bohrnstedt, G. W. and Carter, T. M., “Robustness in Regression Analysis” in Costner, H. L., ed., Sociological Methodology (San Francisco: Jossey-Bass, 1971)Google Scholar; and Borgatta, E. F. and Bohrnstedt, G. M., “Level of Measurement: Once Over Again,” Sociological Methods and Research 9 (1980), 147–60CrossRefGoogle Scholar. For interesting perspectives on the measurement level controversy, see Maxwell, S. and Delaney, H. D., “Another Look at ANCO Versus Blocking,” Psychological Bulletin 95 (1984), 136–47CrossRefGoogle Scholar; and Townsend, J. and Ashby, F., “Measurement Scales and Statistics: The Misconception Misconceived,” Psychological Bulletin 96 (1984), 394–401.CrossRefGoogle Scholar

44 The Catholic variable is statistically significant in all equations even when excluding the interactive terms.

45 Media exposure is often related to other measures of political awareness, like education or interest in politics (see Zaller, The Nature and Origins of Mass Opinion, and Luskin, Robert, “Measuring Political Sophistication,” American Journal of Political Science 31 [1987], 856–99)CrossRefGoogle Scholar. We thus also tested a CATH × EDUCATION interactive and a CATH × INTEREST interactive (when testing the INTEREST interactive, the original INTEREST variable was also included in the models). Three of the eight interactives were significant, all in the expected direction, but their inclusion did not weaken any of the media relationships.

46 Regression analysis, as used thus far, cannot establish with certainty the path of causality. In fact, it is theoretically conceivable that media exposure may itself be the dependent variable: those Catholics who are more exposed to the media may hold opinions indistinguishable from Protestants not because of media exposure per se, but because it is secular Catholics who are more heavily exposed to the media, and they hold opinions different from more religious Catholics due to this secularization, not due to media exposure. Thus, in order to determine that our model is not misspecified, we must show that Catholics more highly exposed to the media are not less religious than other Catholics, for if they are, it is possible that integration into the Catholic community may determine both the manifestation of the Catholic ethos and media exposure. We examined this possibility, and found it to be ungrounded: non-practising Catholics were not more exposed to the media than practising Catholics. In fact, practising Catholics were more heavily exposed to the media than any other group (the correlation coefficient for practising Catholics and media exposure is r =.08^a) indicating that the interaction between religion and the media is not spurious.

47 European researchers have also rediscovered the religious cleavage, suggesting that various countries’ public policy patterns can be explained by the strength of the Catholic presence. For example, see Castles, “On Religion and Public Policy.”

48 Reichley, A. James, Religion in American Public Life (Washington: Brookings Institute, 1985), 219–20.Google Scholar

49 Wald, Religion and Politics in the United States, chaps. 4 and 8. As mentioned earlier, Catholics in each country have unique historical experiences, with Canadian Catholics developing more reticence toward the military due to Canada's involvement with wars in defence of the British Empire, while American Catholics did not become more pacifistic until after Vatican II.

50 Novak, Michael, Catholic Social Thought and Liberal Institutions: Freedom with Justice (2d ed.; New Brunswick, N.J.: Transaction, 1989Google Scholar), particularly chap. 7 on the issue of peace and chap. 8 on the primacy of labour over capital; and Novak, Michael, The Catholic Ethic and the Spirit of Capitalism (New York: Free Press, 1993).Google Scholar

51 Bartels, Larry, “Uninformed Votes: Information Effects in Presidential Elections,” American Journal of Political Science 40 (1996), 206, 215.CrossRefGoogle Scholar

52 This is a reference to Irvine's original comment: “The religious cleavage has indeed been shown to have been the man who came to dinner—early in this century or before” (Irvine, “Explaining the Religious Basis,” 563).

53 Kellstedt, Religion, the Neglected Variable, 299.

54 Irvine, , “Comment on ‘The Reproduction of the Religious Cleavage in Canadian Elections,’” this Journal 18 (1985), 417.Google Scholar

55 Johnston, “The Geography of Class and Religion,” 128.

56 See, for example, Ginsberg, Benjamin, The Captive Public: How Mass Opinion Promotes State Power (New York: Basic Books, 1986).Google Scholar