1 There is a multitude of evidence documenting this observation, including several major sociopolitical surveys and a number of public opinion polls. Among the more important published findings, see Meisel, John, Working Papers on Canadian Politics (Montreal: McGill-Queen's University Press, 1973), 58–59CrossRefGoogle Scholar; Regenstreif, Peter, The Diefenbaker Interlude (Toronto: Longman Canada, Ltd., 1965), 57Google Scholar; and Schwartz, Mildred, Politics and Territory (Montreal: McGill-Queen's University Press, 1974), 187–90.CrossRefGoogle Scholar

2 The 1965 Election Study was conducted by Philip Converse, John Meisel, Maurice Pinard, Peter Regenstreif and Mildred Schwartz. The 1968 Election Study was conducted by John Meisel. Data from these studies were made available by the Inter-University Consortium for Political Research. Neither the original collectors of the data, nor the Consortium, bear any responsibility for the analyses or interpretations presented here. In 1965, the survey questions employed were: “In recent elections, the political party which has formed the government in Ottawa has not had a majority of the seats in Parliament. Do you think it makes a great deal of difference, some difference, or not much difference whether the party forming the government has a majority of seats or not?” and “How important, in deciding how to vote, did you find the issue of whether Canada should have a majority government?” In 1968, the question employed was: “Before the election last June, how important to you was the issue of whether Canada should have a majority government, that is whether a party would have more seats in Parliament than all other parties combined? Was this extremely important, very important, somewhat important, or not too important?”

3 The 1974 Election Study was conducted by Harold Clarke, Jane Jenson, Lawrence LeDuc and Jon Pammett. The majority government questions included in this study were asked of a random half-sample of respondents. Information regarding the sample and approximations of sampling error for proportions cited in this study and for differences of proportions from the 1965 study may be found in Appendix A of this paper. For a more detailed discussion of the sample design, see LeDuc, Lawrence, Clarke, Harold, Jenson, Jane and Pammett, Jon, “A National Sample Design,” this Journal 7 (1974), 701–08.Google Scholar

4 The question employed was: “In recent years, the political party which has formed the government in Ottawa has frequently not had a majority of seats in Parliament. Some people think that it is better when the government has a majority of seats in Parliament, while others feel that minority governments can accomplish more. Which do you feel is better?”

5 This particular question was: “Canada has had a number of minority governments in Ottawa during the past ten years or so—that is, a Parliament in which the party in power has no clear majority over all other parties. On the whole, in your opinion, do you think that a minority government is good or bad for the nation?” C.I.P.O., The Gallup Report, April 25, 1973.Google Scholar

6 This question appears in identical form in the 1965 and 1974 studies, as follows: “If you believed that one party was more likely to be able to form a majority government, but you did not ordinarily vote for that party, how likely would you be to vote for it in order to have a majority government—very likely, somewhat likely, or not at all likely?”

7 The effect of issues in elections is discussed in some detail by Butler, David and Stokes, Donald in Political Change in Britain (London: St. Martin's Press Inc., 1969), 173–92.Google Scholar In particular, they argue that three conditions must be met before an issue will alter the net strength of political parties in an election—the issue must be perceived and attitudes toward it formed by the voters, there must be a skewness of opinion on the issue, and it must be associated differently with the parties in the public mind. See also our attempt to operationalize these conditions in an analysis of issues in the 1974 election. Pammett, Jon, LeDuc, Lawrence, Jenson, Jane and Clarke, Harold, “The Perception and Impact of Issues in the 1974 Federal Election,” this Journal 10 (1977), 93–126.Google Scholar

8 That is, the desirability of a periodic rotation of the party in power as an end in itself. The question employed in the British studies was: “Do you think that in a country like this control of the government should pass from one party to another every so often or that it's all right for one party to have control for a long time?” In 1963, 54 per cent of a sample of British voters felt that alternation was desirable. Butler and Stokes, Political Change in Britain, 431–37.

9 For example, in the 1974 study, a preference for majority government was indicated by 64 per cent of the Liberals, 56 per cent of the Conservatives, 28 per cent of the NDP, and 32 per cent of the Social Credit. See also the distributions on another question in Table 8.

10 It is well to remember that region and party preference are also correlated, and that this relationship is particularly distinct in 1974. Intensity of partisanship is also an important intervening variable which exhibits some correlation with the majority government items, particularly the item dealing with hypothetical switching. As is noted by Schwartz in arguing that unwillingness to switch is due to entrenched partisanship, “If this is the case, then theories of cognitive consistency would suggest that partisanship should lead voters to downgrade the significance of majority government.” Politics and Territory, 190.

11 Ibid., 189.

12 By contradictory responses here is meant the scattering of respondents who gave seemingly opposite opinions to questions in the same survey, for example the 2 per cent of the 1965 respondents who indicated that majority government was “not important” but nevertheless said that they would switch parties because of it, or the 4 per cent of 1974 respondents who would switch parties to obtain a majority government but who had previously indicated a preference for a minority. While it is not impossible to develop a rationale for some of these responses, it is more likely that a small amount of measurement error is involved here, and that other cells in the tables are subject to comparable levels of response error. Small variations therefore should be treated with appropriate caution.

13 With regard to the role played by intensity of partisanship as an intervening variable, it may be noted that the relationship of this variable to the majority government questions, particularly the one dealing with switching, is stronger in 1965 than in 1974. See the comparison of Cramer's V between the two studies in Table 2.

14 This is the general conclusion of an analysis of the 1965 data alone. See LeDuc, Lawrence, “Public Opinion and Minority Government,” paper presented to the 46th Annual Meeting of the Canadian Political Science Association, Toronto, June 1974.Google Scholar Note that Figures 3 and 4 in the present analysis combine votes for the NDP, Social Credit, and other “minor” parties under a single heading. This is done primarily to alleviate the small frequencies that would otherwise result, particularly in analysis of the 1974 half sample. However, the patterns evidenced in Figures 3 and 4 are primarily the result of fluctuations in the NDP vote, because supporters of that party are more numerous in both elections.

15 Support for Liberal candidates by voters first entering the electorate in 1974 was much in evidence in our survey. See the discussion of this pattern in Pammett, Jon, LeDuc, Lawrence, Jenson, Jane and Clarke, Harold, “The 1974 Federal Election: A Preliminary Report,” Carleton Occasional Papers Series, No. 4 (Ottawa: Department of Political Science, Carleton University, 1975).Google Scholar There is no reason, however, to believe that the majority government issue is related to this pattern as responses of new voters to the majority government questions do not differ significantly from those of the total sample, either in 1974 or in 1965.

16 Political Change in Britain, 432–33.

17 Politics and Territory, 253–55.

18 In the analysis shown in Table 9, the three majority government questions are dichotomized on a “positive-negative” basis, that is, majority-other, would-would not switch, and influenced-not influenced.

19 For example, in a regression analysis of the 1965 data treating Liberal vote as a dummy dependent variable and Liberal partisanship, region, subjective social class, religion, age, sex, and a composite majority government variable as dummy independent variables, the contribution of majority government over the proportion of variance explained by partisanship and region is marginal. LeDuc, “Public Opinion and Minority Government.”

20 See Converse, Philip, “The Concept of a Normal Vote,” in Campbell, Angus, Converse, Philip, Miller, Warren and Stokes, Donald, Elections and the Political Order (New York: John Wiley & Sons, Inc., 1966), 9–39.Google Scholar

21 LeDuc, , “Public Opinion and Minority Government,” reports the results of such an analysis of the 1965Google Scholar data.

22 See the discussions of this point in Sniderman, Paul M., Forbes, H. D. and Melzerx, Ian, “Party Loyalty and Electoral Volatility,” this Journal 7 (1974), 268–88Google Scholar; Schwartz, Mildred, “Canadian Voting Behaviour,” in Rose, Richard(ed.), Electoral Behaviour: a Comparative Handbook (New York: The Free Press, 1974), 543–618Google Scholar; Jenson, Jane, “Party Loyalty in Canada: The Question of Party Identification,” this Journal 8 (1975), 543–53Google Scholar; and Jenson, Jane, Clarke, Harold, LeDuc, Lawrence and Pammett, Jon, “Patterns of Partisanship in Canada: Split Identification and Cross-time Variation,” paper presented to the Annual Meeting of the American Political Science Association, San Francisco, September 1975.Google Scholar

23 We would expect, for example, some over-reporting of past Liberal voting behaviour and some distortion in the direction of party identification. The latter will be modified somewhat by intensity of partisanship. See the discussions of response validity and recall by Clausen, Aage, “Response Validity and Vote Report,” Public Opinion Quarterly 32 (1968), 588–606CrossRefGoogle Scholar; Weir, Blair, “The Distortion of Voter Recall,” American Journal of Political Science, 19(1975), 53–62CrossRefGoogle Scholar, which examines American panel data. The use of recall in this paper is similar to that by Sniderman, et al. and Schwartz in their investigations of stability of voting choice.

24 Because we are working here with rather small percentages, considerations of sampling error are particularly important. (See the general discussion of sampling error for these studies in Appendix A.) It is important to bear in mind that many of the percentages cited in this paper are based on total samples, thus making small fluctuations rather more meaningful. The finding, for example, that the net shift to the Liberals by voters expressing concern for majority government was 1 per cent less in 1974 than in 1965 is statistically significant at the .05 level.

25 Both of these effects are discussed in further detail in Pammett et al., “The 1974 Election: A Preliminary Report.”