¹ “One Party Dominance and Third Parties,” Canadian Journal of Economics and Political Science XXXIII (August, 1967), 356. The theory has been expanded and qualified in chapters 2, 3, and 4 of his The Rise of a Third Party: A Study in Crisis Politics (Englewood Cliffs, NJ, 1971).

² Wise of a Third Party, 23–6.

³ He also fails to specify what is to be understood by the term “strain,” though for present purposes this is a less serious shortcoming.

⁴ Rise of a Third Party, 22 n. 2.

⁵ Ibid., 37.

⁶ Ibid. In all fairness, we must point out that Pinard recognizes that “the proportion of the votes maintained is only an indicator of one-party dominance, though,” he adds, “it appears to be a very good one.” Ibid., 63, n. 2.

⁷ Ibid., 28.

⁸ Political Parties: Their Organization and Activity in the Modern State 3rd. English ed., tr. by Barbara, and North, Robert (London, 1964), 308.Google Scholar Emphasis added.

⁹ “It does not seem possible to specify clearly the length of time one-party dominance must prevail for the system to become conducive to third parties.” Pinard, Rise of a Third Party, 64, n. 5.

¹⁰ Ibid., 22.

¹¹ “One-Party Politics and the Voter,” American Political Science Review, L (September, 1956), 708.

¹² There is a fundamental problem lurking here regarding the proper level at which Pinard's theory should be tested. It will be brought out into the open, though not resolved, in section III.

¹³ Theory of Collective Behavior (New York, 1963), 15.

¹⁴ “Throughout the article, for the sake of simplicity, dominant-party strength will be considered as equivalent to opposition-party weakness; in other words, we will be dealing with two-party systems, though there is no reason why the analysis could not be extended to multi-party systems.

¹⁵ Pinard notes that in some circumstances, such as in Quebec from 1958 to 1962, this may be a two-step process: “after a long period of dominance by a strong party, a dissatisfied electorate turns in part to the traditional opposition party. But if this party is soon considered to have failed, as the Conservatives were in 1962, then the electorate is not ready to return so rapidly to the dominant party it just repelled; they shift instead to a third party.” Pinard, Rise of a Third Party, 26. Emphasis in original.

¹⁶ Ibid., 31, emphasis in original.

¹⁷ Ibid., 69, emphasis in original.

¹⁸ The comments in this section are made within the context of elaborating and testing the theory with constituency-level data. See pp. 407–8 below for a discussion of testing it at other levels of analysis.

¹⁹ Smelser, Theory of Collective Behavior, 283.

²⁰ Pinard, Rise of a Third Party, 66.

²¹ Ibid., 66–9.

²² Thomas, L.G., The Liberal Party in Alberta: A History of Politics in the Province of Alberta, 1905–1921 (Toronto, 1959), 29.Google Scholar

²³ Pinard, Rise of a Third Party, chap. 4, especially p. 65.

²⁴ Ibid., 65.

²⁵ Ibid.

²⁶ Ibid. See pp. 66–70 for a description of Macpherson's model. See also Macpherson, C.B., Democracy in Alberta (Toronto, 1962Google Scholar).

²⁷ Accordingly, Professor Pinard must be absolved from any responsibility for the extrapolations which will be made about his “special model,” which was, in any event, offered in a very tentative fashion.

²⁸ Pinard, Rise of a Third Party, 65.

²⁹ Ibid., n. 10.

³⁰ See pp. 409–10 and 416 below.

³¹ For one thing, and this is an important point in its own right, strain would seem to be present, almost by definition, in the class deavage model.

³² Although it cannot be considered here, there is a question as to why a strong opposition should be unable or unwilling to contest a substantial number of seats. This would be readily understandable in a previously weak opposition, but it is rather less intelligible in a party that had not been in a weakened condition.

³³ Throughout the paper we make the (admittedly oversimplistic) assumption that voters who harbour ideological disinclinations to support a specific third party – or third parties in general – are not sufficiently numerous to confound the analysis.

³⁴ The dissatisfied voter of course has the option of abstaining, but abstentions should not, ceteris paribus, depend on the strength of any particular party. However, if there is a sufficiently large number of non-voters, it is possible that they will tend to obscure the relationship somewhat.

³⁵ Pinard's individual-level data indicated that the stronger the dominant Liberal party had been, the higher was the proportion of Liberals who switched to Social Credit (ibid., 30). Since, however, the Conservatives did field a full slate of candidates in the 1962 election, this finding cannot be cited as support for our hypothesis.

³⁶ In this, as in other parts of the paper, we must recognize the dangers involved in making what Converse calls “the assumption of minimal change” – the assumption that the only change is that which is visible as net change in aggregate data. Converse, Philip E., “The Problem of Party Distance in Models of Voting Change,” in The Electoral Process, ed. Jennings, M. Kent and Zeigler, L. Harmon (Englewood Cliffs, NJ, 1966), 177.Google Scholar For an insightful discussion of the proportions of voters defecting from parties in more normal times, see Butler, David and Stokes, Donald, Political Change in Britain (New York, 1969Google Scholar), chap. 13.

³⁷ Rise of a Third Party, 31.

³⁸ “Implications of Within-Nation Variations and Regional Imbalances for Cross-National Research,” in Comparing Nations: The Use of Quantitative Data in Cross-National Research, ed. Merritt, Richard L. and Rokkan, Stein (New Haven, Conn., 1966), 340.Google Scholar

³⁹ Rise of a Third Party, 28.

⁴⁰ Ibid., 37.

⁴¹ Ibid. Another study of province-level data, employing more rigorous definitions, examined all elections since the turn of the century and found Pinard's theory to be, on the whole, not supported. See the paper by André Blais, “Third Parties in Canadian Provincial Politics,” below.

⁴² Rise of a Third Party, 23–30.

⁴³ See n. 15 above.

⁴⁴ See Lemieux, Vincent, “Les dimensions sociologiques du vote créditiste au Québec,” Recherches sociographiques VI (1965), 185–90Google Scholar; and Pinard, Maurice, “La faiblesse des Conservateurs et la montée du Crédit social en 1962,” ibid., VII (1966), 360–3.Google Scholar The original operationalization used by Pinard in the Quebec case was rather more ingenious, and apparently more valid (Pinard, Rise of a Third Party, 23–5).

⁴⁵ The usefulness and, we would argue, the validity of the S-index is suggested in that, while the correlation (Pearsonian r) between the index and the actual percentage of vote never fell below + .90, the other correlations were, in every case, higher (and in some cases substantially higher) when they were computed using the S-index than they were when the actual percentage was used.

⁴⁶ Rise of a Third Party, 69.

⁴⁷ Ibid., 65, n. 10.

⁴⁸ Ibid., 42. In only 3 of the 46 constituencies with UFA candidates were there also Conservative candidates.

⁴⁹ “A Comparison of Prairie Political Movements in Saskatchewan and Alberta,” Journal of Canadian Studies, IV (February, 1969), 18. My emphasis.

⁵⁰ See Thomas, The Liberal Party in Alberta, chap. 4.

⁵¹ Ibid., 146.

⁵² Democracy in Alberta: Social Credit and the Party System, 20.

⁵³ Rise of a Third Party, 70.

⁵⁴ The Progressive Party in Canada (Toronto, 1950), 37.

⁵⁵ Thomas, The Liberal Party in Alberta, 154.

⁵⁶ Ibid., 189.

⁵⁷ Ibid., 203.

⁵⁸ Raw electoral data for all Alberta elections were taken from the “Returns of the General Election of the Province of Alberta,” various years, kindly supplied by the Clerk of the Alberta Legislative Assembly. Candidates’ affiliations for the elections of 1917 and 1921 were taken from Hopkins, Castell, The Canadian Annual Review of Public Affairs 1917 and 1921 (Toronto, 1917 and 1921), 806Google Scholar (1917) and 854 (1921).

⁵⁹ Although there were redistributions in 1913 and 1917, the constituencies under consideration were unaffected.

⁶⁰ At least they were supposed to be overseas – there is some evidence that at least one acclaimed Liberal got no closer to the front than the Saskatchewan border. Thomas, The Liberal Party in Alberta, 169.

⁶¹ It is perhaps significant that the highest mean UFA S-index scores were registered in ridings where candidates had been acclaimed in 1917: 31.2 in the five with acclaimed Liberals and 56.0 for the two in which Conservative candidates had been returned unopposed (the remaining acclamations had been in the cities). This suggests that rejection of the old parties may have been most pronounced in areas where the shortcomings of the party system – in the form of highly suspect acclamations – had been most in evidence.

⁶² For purposes of computing these correlations, and those presented elsewhere in this paper, a riding in which a party did not run a candidate was assigned an S-index of one less than the party's lowest S-index for any other riding. Similarly, acclamations were scored as one greater than the party's highest index elsewhere. Unless otherwise indicated, though, the correlations reported are those computed excluding acclamations and uncontested ridings. Since we are dealing with the population of ridings (not a sample), significance levels are not appropriate.

⁶³ Rise of a Third Party, 43. The series referred to is the “Social Credit in Alberta: Background and Development” series edited by S.D. Clark. In addition to the volumes by Morton, Macpherson, Thomas, and Irving cited in this paper, the series includes Mallory, J.R., Social Credit and the Federal Power in Canada (Toronto, 1954Google Scholar); Mann, W.E., Sect, Cult and Church in Alberta (Toronto, 1955Google Scholar); and Clark, S.D., Movements of Political Protest in Canada 1640–1840 (Toronto, 1959CrossRefGoogle Scholar), and others.

⁶⁴ Flanagan, Thomas, “Ethnic Voting in Alberta Provincial Elections, 1921–1971,” Canadian Ethnic Studies, in (December, 1971), 150.Google Scholar This figure includes the entire province; if the urban areas, in which the UFA did not offer candidates, were excluded, the UFA vote would be considerably higher. The Flanagan paper, we might point out, provides a useful alternative to the present analysis, the narrow limits of which preclude consideration of a number of important factors, ethnicity being one.

⁶⁵ Smith, “Comparison of Prairie Political Movements,” 22.

⁶⁶ Ibid., 24.

⁶⁷ Macpherson, Democracy in Alberta, 142.

⁶⁸ The Social Credit Movement in Alberta (Toronto, 1959), 336, 4.

⁶⁹ Ibid., 341.

⁷⁰ The creation of the ridings of Clover Bar and Grande Prairie prior to the 1930 election necessitates that the analysis for the 1926 election be based on only 44 cases. Beyond this, there were no redistributions between 1926 and 1935.

⁷¹ The solitary rural constituency which did not drop into the Social Credit maw in 1935, Grouard, was, significantly, the only riding which had remained Liberal throughout the entire period of UFA ascendency. According to Flanagan, Grouard's atypical behaviour was in part attributable to the disproportionately large population of French Canadians in this remote riding. Flanagan, “Ethnic Voting in Alberta,” 146–7, 152.

⁷² The apparent inconsistency in the correlations between 1935 Social Credit strength and the S-indices of the UFA and the Liberal party in the previous election stems from the presence of several Conservative candidates, and also from the different number of constituencies (44 and 28 respectively) on which the correlations were computed.

⁷³ The relationship between the UFO and the ILP has been stated succinctly: “the two movements had much in common… [but] disagreed on too many issues to merge. However, they did cooperate in the election campaign.” Tennyson, Brian D., “The Ontario General Election of 1919: The Beginnings of Agrarian Revolt,” Journal of Canadian Studies, IV (February, 1969), 33.Google Scholar The ILP has been excluded from the analysis due to the small number of its candidates in the election (19), and also because it would require, as an urban phenomenon, separate treatment from the exclusively rural UFO, which we do not have the space to present.

⁷⁴ The Progressive Party in Canada, 83.

⁷⁵ Rise of a Third Party, 49.

⁷⁶ Tennyson, “The Ontario General Election of 1919,” 29–30. Emphasis added.

⁷⁷ Oliver, Peter, “Sir William Hearst and the Collapse of the Conservative Party,” Canadian Historical Review, LIII (March, 1972), 34–5.Google Scholar

⁷⁸ Tennyson, “The Ontario General Election of 1919,” 27. See also Young, W.R., “Conscription, Rural Depopulation and the Farmers of Ontario, 1917–19,” Canadian Historical Review, LIII (September, 1972), 290.Google Scholar

⁷⁰ Trowbridge, R.W., “War Time Discontent and the Rise of the United Farmers of Ontario 1914–1919,” MA thesis, University of Waterloo, 1966.Google Scholar

⁸⁰ Young, “Conscription, Rural Depopulation and the Farmers,” 318.

⁸¹ Tennyson, “The Ontario General Election of 1919,” 29.

⁸² Ibid., 27.

⁸³ Young, “Conscription, Rural Depopulation and the Farmers,” 313.

⁸⁴ Smelser, Theory of Collective Behavior, 282–6.

⁸⁵ Raw data for Ontario, as well as the exact changes affected by redistributions, are taken from Lewis, Roderick, Centennial Edition of a History of the Electoral Districts, Legislatures and Ministries of the Province of Ontario 1867–1968 (Toronto, 1968Google Scholar). Particularly with respect to candidates’ affiliations, this source seems to be more complete than the Canadian Annual Review from which Pinard's data were taken; this accounts for the slight discrepancies between our figures and those of Pinard (Rise of a Third Party, 49–50). A redistribution was carried out prior to the 1914 election, but in the main it affected only the urban areas. A few ridings have, however, been excluded from some phases of the analysis because their boundaries were substantially revamped at this time. Finally, the 1919 election marked the first full enfranchisement of women. This vast expansion of the electorate was doubtless of major significance, but the problem of dealing with it empirically, as is the case with so many difficulties of aggregate data analysis, can only be recorded, not resolved.

⁸⁶ In an attempt to further illuminate the processes at work in the rise of the UFO, an indicator of social strain was brought into the analysis of the electoral data. The indicator employed was outmigration from rural areas, as a percentage of total population, during the period 1911–21. (Data taken from Young, “Conscription, Rural Depopulation and the Farmers,” appendix A; we should like to thank Mr Young for permission to use this data.) Given the importance of rural depopulation as a source of strain (ibid., passim, and page 417 above), this would appear a not unreasonable choice.

The migration data, aggregated by county, was matched to the electoral districts as closely as possible, though inclusion of large urban areas in some counties and gross mis-matches between some counties and ridings necessitated that the analysis be based on only 50 cases. (In addition, for a considerable number of these rural population movements were not separated out from those affecting incorporated towns and villages; ibid., notes to Appendix A.)

In the 24 ridings with Liberal candidates, results were much as might have been expected. The Conservative S-index in 1914 correlated with 1919 UFO strength at r = −.31 (the discrepancy between this figure and that in the text, r = −.36, is attributable to the different number of cases employed); when the effects of strain (outmigration) were partialled out, r fell to −.23, indicating the importance of strain in the causal sequence. Strain correlated with UFO strength at r = .24; partialling out previous Conservative strength had virtually no effect (r = −.23). The simple correlation between Conservative strength in 1914 and outmigration was −.44, meaning that the Tories, not unexpectedly, fared worst in areas of heaviest depopulation.

For the 26 ridings not contested by the Liberal party in 1919, the data are frankly in-comprehensible to this writer. The zero-order correlation between 1914 Conservative strength and UFO strength in 1919 was −.43, rising slightly to −.45 when migration was partialled out. Strain correlated with the UFO S-index very weakly (r = .07), although the partial correlation, controlling for Tory strength in 1914, rose to .23. This unexpected result appears to reflect the influence of the relationship between strain and Tory fortunes in 1914, which was, strange to say, positive (r = .27). In other words, in those ridings in which no Liberal candidates were present in 1919, the Conservatives had attracted their greatest 1914 strength in areas most affected by outmigration. This finding is puzzling on two counts: (1) why should the relationship be positive? and (2) why should the relationship, in 1914, differ according to whether there were or were not Liberal candidates in 19191 At the present time, no satisfactory answers – or speculations – can be provided to either of these questions.

⁸⁷ Thorson, T. L., Biopolitics (Toronto, 1970Google Scholar), chaps. 3–5.

⁸⁸ “Les dimensions sociologiques,” 187; see also Pinard, Rise of a Third Party, 72–8.