¹ Alford, Robert R., Party and Society (Chicago, 1963)Google Scholar; Meisel, John, Working Papers on Canadian Politics (Montreal, 1972)CrossRefGoogle Scholar, chap. 2; Laponce, Jean A., People vs Politics (Toronto, 1969), 150Google Scholar; and the careful review of the literature in Winn, Conrad, Spatial Models of Party Systems: An Examination of the Canadian Case, unpublished phd dissertation, University of Pennsylvania, 1972Google Scholar, chap. 2.

² Meisel, John, “Conclusion: An Analysis of the National (?) Results,” in Papers on the 1962 Elections, ed. Meisel, (Toronto, 1964), 286.Google Scholar

³ Stokes, Donald, “Spatial Models of Party Competition,” in Campbell, Angus, et al., Elections and the Political Order (New York, 1966)Google Scholar; and Downs, Anthony, An Economic Theory of Democracy (New York, 1957).Google Scholar

⁴ Ibid.; Converse, Philip E., “The Problem of Party Distances in Models of Voting Change,” in The Electoral Process, ed. Jennings, M. Kent and Zeigler, L. Harmon (Englewood Cliffs, N.J., 1966)Google Scholar; Converse, Philip and Valen, Henry, “Dimensions of Cleavage and Perceived Party Distances in Norwegian Voting,” Scandinavian Political Studies, 6 (1971), 107–52CrossRefGoogle Scholar; Borg, Olavi, “Basic Dimensions of Finnish Party Ideologies: A Factor-Analytic Study,” Scandinavian Political Studies, 1 (1966), 94–120CrossRefGoogle Scholar; Pedersen, Mogens N., Damgaard, Erik, and Olsen, P. Nannestad, “Party Distances in the Danish Folketing 1945–1968,” Scandinavian Political Studies, 6 (1971), 87–106CrossRefGoogle Scholar; Garvey, Gerald, “The Theory of Party Equilibrium,” American Political Science Review, 60 (1966), 29–38CrossRefGoogle Scholar; Niemi, Richard, “Majority Decision Making with Partial Unidimensionality,” American Political Science Review, 63 (1969), 488–97CrossRefGoogle Scholar; Hinich, Melvin and Ordeshook, Peter, “Plurality Maximization vs. Vote Maximization: A Spatial Analysis with Variable Participation,” American Political Science Review, 64 (1970), 772–91CrossRefGoogle Scholar; Wittman, Donald A., “Parties as Utility Maximizers,” American Political Science Review, 67 (1973), 490–8CrossRefGoogle Scholar; and Kornberg, Allan, Mishler, William, and Smith, Joel, “Elite and Mass Perceptions of Canadian Party Location on Issue Space: Some Tests of Two Theories,” paper presented at International Political Science Association meetings, Montreal, August 1973.Google Scholar

⁵ “Dimensions of Cleavage,” 114–16, 135, 147, 149.

⁶ See the seminal article by Lemieux, Vincent, “La composition des préférences partisanes,” Canadian Journal of Political Science, 2 (1969), 397–418CrossRefGoogle Scholar, especially pp. 412–14.

⁷ Irvine, William P., “Voting Shifts in Canada: A Comparison of 1963–65 and 1965–68,” paper presented to the Canadian Political Science Association meetings, Montreal, August 1973.Google Scholar

⁸ Winn, Spatial Models of Parly Systems, 76–7, 99, has suggested a “salience” dimension which is quite similar to the size criteria. Salience for Winn means likelihood of winning, and the prediction he makes is that voters will shift disproportionately to parties with better prospects of winning.

⁹ “Spatial Models.”

¹⁰ Referring only to material already cited, lack of unidimensionality has been demonstrated for the United States (Stokes), Finland and France (Converse), Norway (Converse and Valen), Denmark (Pedersen, et al.), Canada (Kornberg, et al.), and Quebec (Lemieux).

¹¹ “Spatial Models,” 168–9.

¹² Ibid., 168.

¹³ Ibid., 170.

¹⁴ Ibid., 175. Of course, the frame of reference of which the actor is aware may not be the one which triggers the action; but I am concerned more with describing cognitive and perceptual structures than with causes of behaviour.

¹⁵ Converse, “Party Distances,” 196–7, has made the same observation.

¹⁶ The article by Lemieux has some suggestive comments on the interplay of perceptions and competition in two Quebec constituencies.

¹⁷ The data are available through the Inter-University Consortium for Political Research, University of Michigan, and the Institute for Behavioural Research, York University.

¹⁸ Laponce, Jean A., “Non-Voting and Non-Voters: A Typology,” Canadian Journal of Economics and Political Science, 33 (1967), 75–87CrossRefGoogle Scholar, has a good discussion of several of the factors reducing reliability in self-reported behavioural data.

¹⁹ In fact, for our purposes, the question is ambiguous, too difficult for most respondents, and begs the question of dimensionality.

²⁰ As small as some of the interparty ratings are in Table i and Figure i, most are statistically significant. This is particularly true for the Liberals and Conservatives, since ratings by them are based on quite large numbers of respondents; but even among the Créditistes, some of the differences are significant.

²¹ Winn, Spatial Models of Party Systems, 164–9, also found the Créditistes difficult to pin down. Using other data, however, he found that they placed themselves on the left while other party electorates located them on the far right. Most people find it difficult to locate a relatively new party which is active in only a few areas, since it is peripheral to their vision, interests, and geography.

²² Social Credit, of course, suffers from many of the same “peripheral” problems as does the Créditiste party, since both are spin-offs of the same national party (in 1963). Some respondents were probably confused by the similarity of names, though the Francophone version of the questionnaire is the more confusing, referring to Créditistes in several different ways (Créditiste, Parti Créditiste, le Ralliement des Créditistes, le Ralliement des Créditistes de M Caouette).

²³ Laponce, Jean A., “Note on the Use of the Left-Right Dimension,” Comparative Political Studies, 2 (1970), 483–4CrossRefGoogle Scholar, found the same ordering. However, in his book, People vs Politics, chap. 7 and pp. 149–53, he found the classic ordering in most cases with the above order only in a subset of the sample.

²⁴ The text has emphasized differences between parties, but the large standard deviations in Table i and the regional differences within each party (presented below) demonstrate the lack of consensus internal to parties as well.

²⁵ It is particularly interesting to find the left-right dimension not adequate here, since a left-right scale has been imposed on respondents.

²⁶ A Theory of Data (New York, 1964); and Torgerson, Warren, Theory and Methods of Scaling (New York, 1958)Google Scholar, chap. 14. Winn has also applied the unfolding technique to Canadian party space but with different data and somewhat different results.

²⁷ Of course, in some cases party i is also party j, as when a person who identifies with party i says he would vote for that party and no other.

²⁸ Coombs, Theory of Data; see also Torgerson, Theory and Methods, who discusses these assumptions under the heading of the law of comparative judgment.

²⁹ For an exceptional situation, see Almond, Gabriel, The Appeals of Communism (Princeton, 1954).CrossRefGoogle Scholar He found that some former members of the Communist Party switched to right-wing parties, apparently to maximize the distance from their past and perhaps to satisfy a need for authoritarian relationships.

³⁰ The reader interested in pursuing this topic may consult Coombs or Torgerson.

³¹ Using the same procedures, I also analysed the preference rankings implied by cross-tabulating party identification, second choice, and last choice. The results were nearly identical to those in Table ii. Alternatives, A, B, and C (in Table ii) received by far the most support.

³² Returning briefly to Table ii, it will be noted that alternatives A, C, and F are consistent with our ordering of the “central” features referred to above; they all contain the ranking ndp-lib-pc with varying positions for sc. Alternatives B, D, and E, on the other hand, give a contrary ordering (ndp-pc-lib) for the central features. Looking at the “unique percentage” for these alternatives, we may say that 45.2 per cent of the sample uniquely favour ndp-lib-pc, compared to only 15.8 per cent for ndp-pc-lib, a ratio of about 3:1. Since the judgment ratings and the “most alike” ratings also favour the “classic” ordering (ndp-lib-pc), the evidence so far is clearest for that ranking of central features.

³³ A “tie” means that the parties are seen as equidistant from the point of unfolding. A tie hinders unfolding only if it occurs among a group at one end of the scale, as the ndp in the Prairies and bc, but not when it is found among supporters of an “interior” party such as the Liberals in Quebec.

³⁴ If we do not allow ties in unfolded scales, then the ndp preferences cannot be unfolded.

³⁵ Since Laponce, “Note on the Use of the Left-Right Dimension,” found the same order in bc, perhaps this order is not a quirk of the present data.

³⁶ One must recall that I have combined the Socreds and the Créditistes. The presence of one in Quebec and of neither in Ontario makes these two party systems very different, even if their orderings are congruent in these tables. In so far as we devote our attention only to the adequacy of the left-right dimension, then Quebec and Ontario may be said to yield similar results.

³⁷ The Maritimes lack a party in two senses: there are virtually no Social Credit identifiers, and the other party supporters seem to have only the vaguest awareness of Social Credit's position outside the Maritimes.

³⁸ This raises again the issue of whether distance between the parties could be measured by the relative size of the parties (in votes or seats). As long as we must seek at least one other dimension in addition to left-right, it is worthwhile considering whether the two largest parties (ndp and Social Credit in bc) can be considered “near” on this measure as well as distant on left-right. Multidimensionality might be hidden in provinces (like Ontario) where the left-right ordering and the relative size of the major parties (Liberal and Conservative) yield the same relative placement.

³⁹ Stein Rokkan has been mainly responsible for explicating and popularizing this cleavage. See, for example, Rokkan, S., “Geography, Religion and Social Class: Cross-Cutting Cleavages in Norwegian Politics,” in Parly Systems and Voter Alignments, ed Lipset, S.M. and Rokkan, S. (New York, 1967)Google Scholar, and Rokkan, S. and Valen, H., “The Mobilization of the Periphery,” in Rokkan, S., Citizens, Elections, Parties (Oslo, 1969).Google Scholar

⁴⁰ Winn, Spatial Models of Party Systems, chap. 4, concludes from a factor analysis of party distances that there are three dimensions in Canada: (1) cultural (French vs English), (2) regional (actually periphery: rural bc vs rural Quebec), and (3) left-right. One might conceptualize these as the two dimensions on which I focus: besides left-right which is common to his analysis and to mine, his first factor refers to the “centre” pole and his second to the “peripheral” pole.

⁴¹ Others have also thought in terms of populism, whether calling it by that label or not. See, for example, Laponce, People vs Politics, 162; and Meisel, Working Papers, 50–1.

⁴² When other dimensions are identified, multidimensional unfolding techniques can be applied to test their adequacy. For discussions of techniques in several dimensions, see Bennett, J.F. and Hays, W.L., “Multidimensional Unfolding: Determining the Dimensionality of Ranked Preference Data,” Psychometrika, 25 (1960), 27–43CrossRefGoogle Scholar; Hays, W.L. and Bennett, J.F., “Multidimensional Unfolding: Determining Configuration from Complete Rank Order Preference Data,” Psychometrika, 26 (1961), 221–38CrossRefGoogle Scholar; and McElwain, D.W. and Keats, J.A., “Multidimensional Unfolding: Some Geometrical Solutions,” Psychometrika, 26 (1961), 325–32.CrossRefGoogle Scholar

⁴³ We assume that people support parties “nearest” to themselves. If there are different party systems even at the federal level, one can safely hypothesize that differences may exist between the party spaces of federal and provincial levels in any one region or among a given group of people. This might explain the apparent fact that some voters identify with and vote for different parties in provincial and federal elections.