¹ For the development of the practice see Rogers, Norman McL., “Federal Influences on the Canadian Cabinet,” Canadian Bar Review, 11, no. 2 (Feb. 1933), 103–21.Google Scholar Even Mr. Trudeau—who when he became Liberal leader had publicly declared his determination to circumvent the restrictions imposed by the necessity of regional and other representation in the administration—apparently found it necessary to add a second minister from British Columbia within a week of announcing the composition of his caretaker cabinet in April. After the June general election, as a consequence of which it was argued that he had a free hand, his rearrangement of the ministry still satisfied all the traditional rules including a remarkably accurate reflection of the distribution of religions in Canada. The fact that the new cabinet is the youngest and probably the most talented in our history is beside the point: the convention has always been observed irrespective of the consequences.

² The Race Question in Canada, Underhill, Frank H., ed. (Toronto, 1966), 113–14.Google Scholar Andre Siegfried's observations were made during the national election of 1904. His book was first published in French in 1906, and an English edition appeared in London in 1907.

³ In 1921 Bryce, Lord observed that the need to avoid racial and religious conflict in Canada meant that “politics is apt to become a series of compromises, and… there is no sentiment of class hostility involved. The rich and the less rich—for one can hardly talk of the poor—the farmers, merchants, manufacturers, shopkeepers, professional men, have been found in both parties, and if the country be taken as a whole, in tolerably equal proportions.” Modern Democracies (London, 1921), I, 527.Google ScholarReid's, Escott pioneering aggregate data study, “Canadian Political Parties: A Study of the Economic and Racial Bases of Conservatism and Liberalism in 1930,” Contributions to Canadian Economics, 6 (1933), 7–39CrossRefGoogle Scholar, did not even mention class divisions as a source of political competition. The economic bases to which he referred were those of urban and rural interests. The absence of the influence of social class in other aspects of our national life has frequently been noticed. See, for example, McDougall, R. L., “The Dodo and the Cruising Auk: Class in Canadian Literature,” Canadian Literature, no. 18 (Autumn 1963), 6–20Google Scholar; and Mealing, S. R., “The Concept of Social Class and the Interpretation of Canadian History,” Canadian Historical Review, 46, no. 3 (Sept. 1965), 201–18.CrossRefGoogle Scholar

⁴ The Vertical Mosaic: An Analysis of Social Class and Power in Canada (Toronto, 1965), 3.

⁵ See ibid., 374–6. Fairly comprehensive bibliographies of the now growing literature on Canadian political behaviour may be found in Courtney, John C., ed., Voting in Canada (Toronto, 1967)Google Scholar; Thorburn, Hugh G., ed., Party Politics in Canada (Toronto, 2nd ed., 1967)Google Scholar; and Fox, Paul, ed., Politics: Canada (Toronto, 2nd ed., 1966).Google Scholar

⁶ The Politics of Democracy: American Parties in Action (New York, 1940).

⁷ “The Canadian Party System in Transition,” Canadian Journal of Economics and Political Science, 9, no. 3 (Aug. 1943), 304.

⁸ “Concerning Mr. King,” Canadian Forum (Sept. 1950), 122, 125.

⁹ Speaking to the 1967 banquet of the Canadian University Liberal Federation, Professor Underhill said: “I may hold my nose at times but I still go into the polling booth and vote Liberal.” Globe and Mail (Feb. 13, 1967).

¹⁰ See his very persuasive analysis of the implications of brokerage in “Toward the Democratic Class Struggle,” Journal of Canadian Studies, 1, no. 3 (Nov. 1966), 3–10.

¹¹ The Vertical Mosaic, 374.

¹² Professors Engelmann, and Schwartz, concede in their study that the NDP “may be developing into a party which is differentiated along social class lines on a national scale.” See Political Parties and the Canadian Social Structure (Toronto, 1967), 240.Google Scholar See also Gagne, Wallace and Regenstreif, Peter, “Some Aspects of New Democratic Party Urban Support in 1965,” Canadian Journal of Economics and Political Science, 33, no. 4 (Nov. 1967), 529–50.CrossRefGoogle Scholar In British Columbia the association between class and CCF and NDP support has long been recognized. See Young, Walter D., “The NDP: British Columbia's Labour Party,” in Meisel, John, ed., Papers on the 1962 Election (Toronto, 1964), 181–200.Google Scholar In his secondary analysis of Gallup Poll data Alford found that Ontario exhibited a degree of class-voting often as high as that of the United States. See Party and Society: The Anglo-American Democracies (Chicago, 1963), 266–7. For the evidence which aggregate data provide of class-voting in several major urban areas of Canada see my “The Use of Aggregate Data in the Analysis of Canadian Electoral Behaviour,” paper presented at the Conference on Statistics of the Canadian Political Science Association, Ottawa, 1967.

¹³ See Underhill, Frank H., “Political Stagnation in Canada,” in his In Search of Canadian Liberalism (Toronto, 1960), 251–3.Google Scholar

¹⁴ See, for example, a Canadian Press report by Kelly, Ken, “NDP Win May Have Been One-Shot Deal,” Kitchener-Waterloo Record (Nov. 13, 1964).Google Scholar The argument that the candidate's role is especially important in a by-election is advanced by Young, W. D., “The Peterborough Election: The Success of a Party Image,” Dalhousie Review, 40, no. 4 (Winter 1961), 514.Google Scholar

It has been suggested that Maurice Pinard's explanation of the rise of Social Credit in Quebec (see his “One Party Dominance and Third Parties,” Canadian Journal of Economics and Political Science, 33, no. 3 (Aug. 1967), 358–73) might be applied in the case of Waterloo South. But, as Table I suggests, Liberal opposition to Conservative dominance of the riding before 1964 had not been consistently as weak as his thesis requires. In fact the Liberals won the constituency in 1953 and although the Conservatives had otherwise held it continuously since the First World War (with the exception of a Progressive victory in 1921) their share of the vote had rarely reached 45 per cent. It seems almost certain that the strength of the CCF and the NDP in Ontario has been and is related to much more important factors than the relative position of the two leading parties even in individual constituencies.

¹⁵ The federal and provincial boundaries of Waterloo South were identical and unchanged from Confederation to 1966.

¹⁶ The original purpose of the survey investigation on which this paper is based was to examine the role which the candidate plays in parliamentary elections. For a discussion of the results of this aspect of the study see my “The Myth of Candidate Partisanship: The Case of Waterloo South,” forthcoming in the Journal of Canadian Studies.

¹⁷ For the circumstances surrounding this change see three articles by Scott Young in the Globe and Mail (Sept. 28, 29, and 30, 1964).

¹⁸ For a brief description of the campaign and some of its lighter moments see my “Waterloo South: Breakthrough for the NDP?” Canadian Forum (Dec. 1964), 196–8.

¹⁹ The by-election was held on November 9, 1964. Over the following two weeks a 20-minute questionnaire was administered to a systematically selected sample of the electorate in the principal communities of the constituency: Gait, Preston, Hespeler, Ayr, Baden, and New Hamburg. Altogether 606 interviews were obtained (a response of 84.2 per cent) of which 504 were from people identified as having voted for one of the three parties in the by-election. Departures from these numbers of cases in all subsequent tables will be explained. It will be noticed that the total 1963 vote for each party shown in the last row of Table II does not match the distribution suggested for that election by Table I. In this instance respondents were being asked to recall their vote from over a year and a half earlier and the numbers who either did not know how they had voted in 1963 or did not know if they had voted at all in that election, suggest that the poor fit with the expected distribution may be due to short political memories. In any case it would not be expected that the sample would accurately reflect the electorate's 1963 behaviour because of changes in the population due to removals, deaths, and the arrival of new voters. Over a wide range of other characteristics, however, including age groups by sex, religious affiliation, ethnic origin, and by-election voting behaviour, the chi-square test for goodness of fit showed the sample to be closely representative of the area surveyed. Further details may be had from the Data Bank of the Institute for Behavioural Research at York University, where the data deck and codebook have been lodged.

²⁰ The most recent detailed assessment of the lack of a class base to our national electoral behaviour may be found in ALford's Party and Society. He does not claim that the influence is entirely absent but rather (p. 250) that “the association between class and voting behaviour… is lower in Canada than in any of the other Anglo-American countries.”

²¹ Cairns argues that “given the historical (and existing) state of class polarization in Canada the electoral system has made sectionalism a more rewarding vehicle for amassing political support than class.” See Cairns, Alan C., “The Electoral System and the Party System in Canada, 1921–1965,” this Journal, 1, no. 1 (March 1968), 75.Google Scholar

²² Only religious affiliation, ethnic origin, and social class are considered here as relevant variables. There are, of course, many other possibilities but an analysis of the influence of length of residence in the constituency, home ownership, and sex showed no connection between these factors and the by-election voting pattern. Education and income were found to reflect broadly the same relationship as that revealed by an analysis of the role of social class as defined by occupation, and space has therefore not been taken for an account of these variables. The Conservative dominance amongst those over the age of 65 is indicated by the fact that for five age groups and the by-election vote x ² = 31.18 (with 8 d/f p < .001), whereas removing the eldest age group gives x ² = 7.01 (with 6 d/f p > .30).

²³ Sources for the data from other studies presented in Table III are, respectively: Regenstreif, Peter, The Diefenbaker Interlude: Parties and Voting in Canada (Toronto, 1965), 104Google Scholar, calculated from CIPO data in Table IX; Meisel, John, “Religious Affiliation and Electoral Behaviour: A Case Study,” Canadian Journal of Economics and Political Science, 22, no. 4 (Nov. 1956), 486CrossRefGoogle Scholar, calculated from Table II; George Perlin, “St. John's West,” in Meisel, ed., Papers on the 1962 Election, 10, calculated from Table I; Anderson, Grace M., “Voting Behaviour and the Ethnic-Religious Variable: A Study of a Federal Election in Hamilton, Ontario,” Canadian Journal of Economics and Political Science, 32, no. 1 (Feb. 1966), 30CrossRefGoogle Scholar, calculated from Table I; Havel, J. E., Politics in Sudbury (Sudbury, 1966), 65Google Scholar, calculated from Table 36; and John C. Courtney and Smith, David E., “Voting in a Provincial General Election and a Federal By-election: A Constituency Study of Saskatoon City,” Canadian Journal of Economics and Political Science, 32, no. 3 (Aug. 1966), 349Google Scholar, calculated from Table X.

²⁴ The percentage distribution among the 75 Roman Catholics in the sample who remembered how they had voted in 1963 was Conservative 29, Liberal 52, and NDP 19.

²⁵ These data may be compared with Meisel, “Religious Affiliation and Electoral Behaviour,” 486.

²⁶ For all seven religious groups and the three-party vote, x ² = 54.27 (with 12 d/f p < .001). For three religious groups (Protestant, Roman Catholic, and Other) and the three-party vote, x ² = 39.77 (with 4 d/f p < .001).

²⁷ Respectively, x ² = 4.77 (with 6 d/f p > .50) and x ² = 1.70 (with 2 d/f p > .30). It is not necessarily being suggested that the Liberal and Conservative parties either should or could be combined for electoral purposes. The device merely serves as an analytical tool in explaining the riding's electoral behaviour.

²⁸ Havel, Politics in Sudbury, 75–6.

²⁹ These data may be compared with ibid., 69.

³⁰ For five ethnic groups and the three-party vote, x ² = 26.01 (with 8 d/f p < .005). For two ethnic groups (British and non-British) and the three-party vote, x ² = 18.84 (with 2 d/f p < .001).

³¹ For all five groups and the hypothetical two-party vote, x ² = 2.08 (with 4 d/f p > .70). For two groups (British and non-British) and the hypothetical two-party vote, x ² = 0.52 (with 1 d/f p > .30).

³² “Recent Changes in Canadian Parties,” in Thorburn, ed., Party Politics in Canada, 41–2. Although the Liberals gained seats in the West in the 1968 federal election, a careful examination of the pattern of the vote suggests that the regional disparities of the last five years remain.

³³ See, for example, the data in Porter, The Vertical Mosaic, 101–3.

³⁴ The term social class is one which has been given a number of meanings. For the purposes of this study an objective scheme of class assignment was developed broadly following that described in Moser, C. A. and Hall, J. R., “The Social Grading of Occupations,” in Glass, D. V., ed., Social Mobility in Britain (London, 1954), 29–50Google Scholar, and using the nature of the occupation of the head of the household in which the respondent lived as the reference. Non-manual occupations were called middle class and manual occupations working class. Within the middle class four categories were distinguished and in Table VIII these are combined into two for purposes of simplicity. The three standard working class categories were used. Each of the respondents was also graded according to status scales developed by Pineo, Peter C. and Porter, John in “Occupational Prestige in Canada,” Canadian Review of Sociology and Anthropology, 4, no. 1 (Feb. 1967), 24–40CrossRefGoogle Scholar; and Bernard R. Blishen in “A Socio-Economic Index for Occupations in Canada,” ibid., 41–53. A comparison of the classification scheme used here with each of these scales showed a general agreement although there were minor differences due to the fact that some non-manual jobs have a very low prestige. Unless otherwise stated all references in this paper are to objective social class as determined by this scheme.

³⁵ The percentage of each ethnic group in non-manual occupations was English 40, Irish 44, Scots 41, German 34, and Other 35.

³⁶ See, for example, the data in Table 13, Blondel, J., Voters, Parties, and Leaders: The Social Fabric of British Politics (Harmondsworth, 1963), 57Google Scholar; and in Robinson, Alan D., “Class Voting in New Zealand: A Comment on Alford's Comparison of Class Voting in the Anglo-American Political Systems,” in Lipset, Seymour M. and Rokkan, Stein, eds., Party Systems and Voter Alignments: Cross-National Perspectives (New York, 1967), 97.Google Scholar

³⁷ See the data reported in Lipset, Seymour Martin, Political Man: The Social Bases of Politics (Garden City, 1963), 252–4.Google Scholar

³⁸ Of the 606 respondents in the sample 374 (62 per cent) were manual workers.

³⁹ For 1963,x ² = 35.23 (with 2 d/f p < .001). For 1964, x ² = 37.41 (with 2 d/f p < .001).

⁴⁰ If Alford's index of class voting is calculated for the data in Table IX it appears that virtually no change has occurred because the NDP made gains in the middle class proportionate to those made in the working class.

⁴¹ No one familiar with the NDP in Ontario (or in some other provinces such as British Columbia) will doubt that its principal appeal to the electorate is as a labour party. The extent to which this is found attractive by voters is another matter.

⁴² Respondents were asked to place themselves in one of three classes: upper, middle, and working. Those who chose the working class were then asked to identify themselves as skilled, semi-skilled, or unskilled. Of the 606 people interviewed 12 said they were upper class, 168 middle class, and 404 working class, while 22 were not able to answer. For two classes (upper combined with middle, and working) and the three-party vote, x ² = 6.14 (with 2 d/f p < .05).

⁴³ Studies in Europe and the United States suggest that roughly three-quarters of the members of an objective class will identify themselves as belonging to it. While 83 per cent of the objective working class respondents in the Waterloo South sample claimed working class status for themselves, nearly half of the objective middle class did so as well. This tendency was less marked among older people, however, and in the case of the youngest age group nearly all of those objectively middle class respondents who said they were working class (63 per cent of the group) had working class parents.

⁴⁴ x ² = 37.21 (with d/f p < .001).

⁴⁵ The treatment of the data in this section follows the method outlined in Coleman, James S., Introduction to Mathematical Sociology (New York, 1964)Google Scholar, chap. 6.

⁴⁶ It is also worth noticing that although the number of cases is small the propensity of Roman Catholics to vote Liberal is much more marked among people of non-British origin. The same phenomenon can be observed for 1963 in Table XII.

⁴⁷ The values associated with each variable in Table XIII are estimates of the effect of each upon the vote of the party in question. These are calculated in each case by averaging the percentage differences in each pair of controlled comparisons shown in Tables XI and XII. The two values r and s represent the total effect of what Coleman calls “random shocks”—r in the direction of the behaviour being examined (i.e., voting for a particular party) and s in the opposite direction (i.e., not voting for that party). In the case of each party all five values aggregate to unity, indicating that all influences have been accounted for even though all are not identified. The probability that the observed effect could have occurred by chance is tested by estimating the variance of each effect parameter, finding the standardized normal deviate, and consulting tables of the standardized cumulative normal distribution. For examples of these particular calculations see Coleman, Introduction to Mathematical Sociology, 205–7.

⁴⁸ If age is added as a fourth attribute in explaining the NDP vote over two-thirds of the variation can be accounted for. In this case the effect of class is .39, of ethnic origin .07, of religious affiliation .06, and of age .16. Random shocks r and s are reduced respectively to .06 and .26. The effect of age in relation to the three principal variables on the vote of the other parties is negligible.

⁴⁹ Some 85 per cent of the respondents in the sample recalled being canvassed by one or more of the parties, and of these 94 per cent were seen by the NDP while only about two-thirds were visited by either the Liberals or the Conservatives. An additional side-effect of the opportunity for a concentration of effort which the by-election offered was that the carefully organized NDP campaign generated many more local party workers than had been available before—so that in later elections outside help was not required. The by-election was the occasion for a major trial of the new electioneering techniques which were developed by the party in Toronto's east end, and which have produced a number of striking successes—notably the important gains of the 1967 Ontario election. The method is fully described in an extremely interesting manual published by the federal party: Morton, Desmond, With Your Help (Ottawa, 1966).Google Scholar

⁵⁰ One of the shortcomings of the present study is its failure to probe the identification between the working class and the NDP which it is being suggested exists in Waterloo South. The reason for that gap in the data is that the original purpose of the survey, as was noticed at the beginning of the paper, was to examine the role of the candidate in parliamentary elections. However, recent survey data from other studies indicate that the party's image as a group especially sensitive to the interests of the working class is increasing. A survey of the Ontario electorate taken in January 1967 found that nearly half of those who answered the question: “What are all the things you like about the provincial New Democratic Party here in Ontario?” mentioned its attitude to labour. See John Wilson and David Hoffman, “Ontario: A Three-Party System in Transition,” in a volume of essays on party politics in the Canadian provinces forthcoming from Prentice-Hall of Canada.

⁵¹ Ibid.

⁵² Watt, F. W., “The National Policy, the Workingman, and Proletarian Ideas in Victorian Canada,” Canadian Historical Review, 40, no. 1 (March 1959), 1–26.CrossRefGoogle Scholar

⁵³ See McKenzie, R. T. and Silver, Allan, “Conservatism, Industrialism and the Working-Class Tory in England,” in Rose, Richard, ed., Studies in British Politics (London, 1966), 21–33.CrossRefGoogle Scholar

⁵⁴ See Ostry, Bernard, “Conservatives, Liberals, and Labour in the 1870's,” Canadian Historical Review, 41, no. 2 (June 1960), 104—10.CrossRefGoogle Scholar For an assessment of labour reaction to these overtures see Robin, Martin, “The Working Class and the Transition to Capitalist Democracy in Canada,” Dalhousie Review, 47, no. 3 (Autumn 1967), 326–43.Google Scholar

⁵⁵ See Wilson and Hoffman, “Ontario: A Three-Party System in Transition.”

⁵⁶ A comparison of the percentage share of the vote won by the three parties in federal elections in Toronto between 1953 and 1965 shows virtually no change for the Liberals, a decline for the Conservatives, and a corresponding increase for the NDP. The sudden growth in the Liberal vote in this area in 1968 appears to have come equally from both the other parties leaving the long-term shift unaffected. It is this phenomenon which has enabled the Liberals in recent elections to win nearly every seat in what used to be known as “Tory Toronto.” The change is especially obvious in the east end of the city, which was once dominated by the Conservatives and is now a particularly strong area for the New Democrats. All that distinguishes this part of Toronto from the heavily Liberal and working class ridings in the central downtown and west end areas is the fact that its population is very much more British in origin. At the same time, in other cities in the province where it might have been expected that industrialization would promote the growth of a class politics, there is some evidence that it is indeed the ethnic cross-pressure which interferes with the development. In Kitchener, survey data collected by University of Waterloo undergraduates indicate a much stronger Liberal bias among workers of German origin than among those of British origin. For the view that past Liberal successes in Sudbury have been due to the cross-pressure of French origin on the working class see my “The Use of Aggregate Data,” 29–30. It is intriguing to notice that in both these constituencies the voting pattern has lately shown much greater support for the NDP than at any previous time.

⁵⁷ A preliminary analysis of aggregate data by census tract for the 1968 election shows a general improvement of the NDP vote in working class areas in nearly every part of Ontario, including a growth in support in cities such as Brantford and Windsor where the party has never before made a strong showing federally. The pattern is in fact very close to the result of the 1967 provincial election in Ontario, which appears to have marked the emergence of a more clearly defined class base to the province's political system. See Wilson and Hoffman, “Ontario: A Three-Party System in Transition.”

⁵⁸ It could be argued that Canada is especially fertile ground in any case for the development of a class politics because of our more “elitist” value system. For an extensive discussion of the differences between Canadian and American society in this respect see Lipset, Seymour Martin, “Revolution and Counter-Revolution: The United States and Canada,” in Ford, Thomas R., ed., The Revolutionary Theme in Contemporary America (Lexington, Ky., 1967), 21–64.Google Scholar

⁵⁹ The 1966 interim census shows that nearly half the Canadian people live in major urban areas with populations larger than 100,000.

⁶⁰ For a summary of the American data on this point see Lane, Robert E., “The Politics of Consensus in an Age of Affluence,” American Political Science Review, 59, no. 4 (Dec. 1965), 874–95.CrossRefGoogle Scholar There appears to be no evidence at all that working class identification is declining in Canada. Indeed, in the Waterloo South data—contrary to what might have been expected—the proportion of objectively working class people who assign themselves to the working class is almost identical (roughly 85 per cent) in every age group except the eldest. Here the discrepancy appears to be due to the large number of older people who describe themselves as middle class “now that we're retired.” Those who had been employed in manual occupations before retirement were assigned objective working class status.

⁶¹ “Toward the Democratic Class Struggle,” 9–10.

⁶² The proposition that joint acceptance by the two older parties of a class base to the Canadian political system might result in greater success for their cause is very strikingly borne out by the two most recent elections in Waterloo South. If the analysis offered in this paper is correct the NDP ought to have won the provincial riding easily in 1967—since it had been reduced by redistribution to the areas which had been its strongest fighting ground in the federal by-election. In fact, however, the Conservatives held the seat by a margin of 721 votes—entirely as a consequence of the defection to them of a good part of what was left of the Liberal vote. Conversely, in the 1968 federal election, Mr. Saltsman was able to hold the new riding of Waterloo only because the huge Liberal majority in the areas which were added to the old riding by the 1966 redistribution was insufficient to overcome that party's weakness in Gait, Preston, and Hespeler. At the same time, the Conservatives—who are much stronger than the Liberals in the old riding—ran nearly as badly as the NDP in the predominantly middle class city of Waterloo. The lesson seems obvious.

⁶³ Such a marriage took place in British Columbia after the 1941 election with ultimately disastrous consequences for the provincial Liberal and Conservative parties. For the details see Ormsby, Margaret A., British Columbia: A History (Toronto, 1958), 473–8 and 487–90.Google Scholar