¹ Epstein, Leon D., “A Comparative study of Canadian Parties”, this Review, 58 (03, 1964), 46–60Google Scholar. I am grateful to Professor Epstein for reading and commenting upon an earlier draft of this article. Although I have benefited greatly from his suggestions, I naturally absolve him of responsibility for any shortcoming herein.

² Leon D. Epstein, “Cohesion of British Parliamentary Parties,” ibid., 50 (June, 1956), 360–377.

³ Op. cit., p. 54.

⁴ Ibid., p. 56.

⁵ Ibid., p. 52.

⁶ Idem, and private communication from Leon D. Epstein.

⁷ The legislative constituences were stratified in terms of region (i.e., Quebec, Ontario, and the residual area) and urban-rural characteristics. I am indebted to Philip E. Converse of the Survey Research Center of the University of Michigan for his assistance in selecting the sample.

⁸ Normally, each party caucuses once a week, generally on a Tuesday, Wednesday, or Thursday morning. The length of the agenda and the importance of the topics to be discussed determine the duration of the meeting. Usually, a caucus lasts approximately one and one half hours. When these interviews were taken, the minority Conservative Government was in danger of being overthrown. Consequently, there were sometimes two or even three caucuses during a week in which crucial votes were scheduled. Probably the most frequent use of the caucus was made by the Social Credit party, whose support on a number of occasions enabled the Conservative Government to survive.

⁹ To help ensure that we were actually delineating caucus functions with the question on caucus discussions, respondents also were asked later: “What do you think are the two or three most important functions a caucus performs?” Unfortunately, we have been unable to use these data. First, we were able to code only a single function because well over half of the respondents did not articulate more than one. More important, virtually all the legislators structured their responses in terms of “caucus ought” rather than “does.” It is interesting to note, however, that there were substantial similarities in their responses to this and the first question. For example, 11% of the M.P.s said a caucus ought to serve as a clearinghouse for information (i.e., when reports of committees could be read and other interesting or useful information exchanged). The largest single group (37%) said the caucus ought to enable the members to plan strategy and devise tactics for impending debate. A third group (27%) said the caucus ought to enable the members to air their grievances and complaints (cathartic), while the remainder (25%) said the caucus should permit participants to hammer out a consensus on party program. In other words, 52% of the respondents felt the caucus ought to perform what we will term consensus-building functions, while 33% indicated in their responses to the first question that the caucus does do this.

¹⁰ Pickersgill, J. W., The Mackenzie King Record (Toronto: University of Toronto Press, 1960), p. 9Google Scholar, quoted in Epstein, op. cit., p. 52. Mr. Pickersgill was for years secretary to the Cabinet and the man sometimes considered closest to both King and his successor as Prime Minister, Mr. Louis St. Laurent. He later stood for Parliament and joined the Cabinet as Minister of Citizenship and Immigration. He fell within the sample for this study and provided us with a fascinating interview which lasted almost three hours.

¹¹ Meisel, John, “The Formulation of Liberal and Conservative Programmes in the 1957 Canadian General Election,” Canadian Journal of Economics and Political Science, 26 (11, 1960), p. 565CrossRefGoogle Scholar. Apparently the Liberal leaders were particularly swayed by their economic advisors. Meisel says they appeared “mesmerized by the cult of the Gross National Product,” ibid., p. 571. For another discussion of the relationship between Liberal party leaders and their administrative advisors in the formulation of party policies, see Hodgetts, J. E., “The Liberal and the Bureaucrat,” Queen's Quarterly, 62, 176–83Google Scholar.

¹² At the time these interviews were taken the four parties in the Canadian House of Commons fell quite naturally along the traditional left-right political continuum. The New Democrats, the party of the left, had a House membership of 19 and the Liberals, the part of the left-center, had elected 100 members. The right-center Progressive Conservatives numbered 119 and formed the minority Government for this Parliament while the right-wing Social Credit party had 30 members. For purposes of analysis, we have grouped the Liberals (N-63) and the New Democrats (N-13) together under “Left-Wing Parties” and the Conservatives (N-66) and Social Creditors (N-23) together under “Right-Wing Parties.”

¹³ This is a statistic proposed by Goodman, Leo A. and Kruskal, William H. in “Measure of Association for Cross Classifications,” Journal of the American Statistical Association, 44 (12, 1954)Google Scholar. The magnitude of the correlations are somewhat greater than equivalent and more frequently used Tau Betas. However, unlike the latter statistic, they are not limited by a “square table” requirement (i.e., that tables be 2×2 and 3×3, etc.).

¹⁴ Good examples of the literature which deal with these latter differences are Trudeau, Pierre Elliot, “Some Obstacles to Democracy in Quebec,” Canadian Journal of Economics and Political Science, 24 (1955), 297–314CrossRefGoogle Scholar; and Maheux, Arthur, “Democracy and the French Canadian,” in Douglas, Grant (ed.), Quebec Today (Toronto: University of Toronto Press, 1958), pp. 341–362Google Scholar. For purposes of analysis we combined the following variables in our index: ethnicity; religion; born Quebec; parochial education; educated in Quebec; and frequent (at least once a week) church attender. A legislator's position on the index was determined by coding him “2” for each “French” characteristic he possessed (i.e., Catholic, parochial education, etc.) and “0” if he did not possess a trait. In the few instances where the information had not been ascertained, the respondent was coded “1,” on the assumption that it was equally probable he did or did not possess a particular characteristic. Then his responses were summed and cutting points arbitrarily established in the scores. Since excessively high variance together with low correlations to other items was not a factor for any of the variables, no special weighting system was employed. Thus, legislators were arrayed along a continuum ranging from those who possessed none of these characteristics (Position “A”) to those who possessed them all (Position “D”). Those legislators in the “B” category had virtually none of the characteristics while those in “C” had virtually all.

¹⁵ There was also a correlation of .22 between the dependent variable and metropolitan-non-metropolitan constituency differences as well as a correlation of .20 between length of parliamentary experience and perceptions of caucus. However, these correlations undoubtedly reflect the fact that the parties of the right and left were disproportionately distributed in terms of these two variables. For example, fully 60% of the right-wingers had between 1–6 years of parliamentary experience while only 19% of the left-wingers enjoyed equivalent tenure. In addition, 40% of the left-wing legislators represented constituencies situated wholly within metropolitan regions while only 10% of the right-wing legislators represented such constituencies.

¹⁶ Supra, pp. 85–86.

¹⁷ Although, to the best of our knowledge, there are no empirical data which delineate Congressional responses to such a cohesion question, a number of studies strongly support our assumption, albeit indirectly. For example, in U.S. Senators and Their World, (Chapel Hill: University of North Carolina Press, 1960)Google Scholar, Donald R. Matthews demonstrated the importance of constituency, interest groups and other Senators upon Senatorial attitudes and behavior. Froman's, Lewis study, Congressmen and Their Constituencies (Chicago: Rand McNally, 1963)Google Scholar illustrated the importance of selected constituency characteristics upon Congressional voting behavior. An important multivariate analysis of the relationship between Congressional voting and independent variables such as region and constituency was made by MacRae, Duncan, The Dimensions of Congressional Voting (Berkeley: University of California Press, 1958)Google Scholar. Perhaps the most sophisticated attempt to illustrate the relative salience of party, constituency, individual predispositions and administrative cues upon congressional attitudes and behavior was the research carried out by Miller, Warren and Stokes, Donald and reported in “Constituency Influence in Congress,” this Review, 57 (1963), 45–57Google Scholar.

¹⁸ For his views, see op. cit., p. 56. In a recent article, Howard Scarrow lists the arguments usually given by those who feel Canadian parties and politics are more programmatic and ideological than American and those who do not. See Scarrow, , “Distinguishing Between Political Parties—The Case of Canada,” Midwest Journal of Poltiical Science, 9 (02, 1965), 61–76CrossRefGoogle Scholar.

¹⁹ See, for example, the Hansard reports on the debates over the pensions legislation, the new flag, or the alleged immorality of the Liberal Cabinet in the 26th Parliament.

²⁰ Also known as the Lippit Iterative Program. For a description of the program, see Pelz, Donald E. and Andrews, Frank, The S.R.C. Computer Program for Multivariate Analysis: Some Uses and Limitations (Ann Arbor: Survey Research Center, 1961)Google Scholar.

²¹ Huitt's, Ralph K. study, “The Morse Committee Assignment Controversy: A Study on Senate Norms,” this Review, 51, (1957), 313–330Google Scholar affords an excellent illustration of how difficult it has been to curb the “maverick” tendencies in American Congressmen and how infrequently sanctions have been invoked by the parties to punish deviant behavior. Froman, Lewis, in “The Importance of Individuality in Voting in Congress,” Journal of Politics, 35 (1963), 324–32CrossRefGoogle Scholar, argues that we must be cognizant of the importance of individual predispositions if we are to understand Congressional voting behavior. Another report by Huitt, , “Democratic Party Leadership in the Senate,” this Review, 55 (1961), 333–45Google Scholar, as well as Lewis Froman and Randall Ripley's “Conditions for Party Leadership: The Case of the House Democrats,” ibid., 59 (1965), 52–63 also illustrate the difficulties entailed in obtaining cohesive action from a Congressional party. Further, a basic criticism of American parties made by responsible-party theorists such as E. E. Schattschneider and James M. Burns is that American Congressional parties do not really behave like parties—at least on the British model.

²² Epstein quotes the late H. McD. Clokie, an expert on Canadian politics, who in an insightful essay, “The Machinery of Government,” contended that the unity of Canadian parties was expected by the public as the very condition of party government of the Cabinet variety. See Epstein, op. cit., p. 52. That this particular group of legislators may have been aware of constituency expectations is suggested by the fact that 59% of them said most of the time they knew how rank-and-file constituents felt about the major issues coming before parliament. An additional 24% said they were sometimes aware, while only 17% said they were seldom aware of constituency opinions. Furthermore, 42% said they visited and talked with their constituents every week; 44% said they communicated with their constituency once a month and only 14% said they did not visit their constituency at least once a month.

²³ For example, there was a correlation of .04 between such motives and length of parliamentary experience; one of .09 between motives and legislators' perception of political competition in their constituencies; a correlation of .11 with variations in respondents' educational backgrounds; and one of .15 with metropolitan-nonmetropolitan constituency differences.

²⁴ For the strongest relationship other than regional background, 43% of the legislators representing metropolitan constituencies were party- and 13% system-oriented. The equivalent proportions for those representing nonmetropolitan constituencies were 49% and 15% respectively.