1 A poll by United Communications Research (UCR) in late May 1986 showed that the NDP led Social Credit by 38.9 to 35.4 per cent among decided voters. A UCR poll for British Columbia Television in early July, after Bennett announced his intention to resign but before a successor had been chosen, showed a Social Credit lead of 44.5 to 39.5 per cent. We are grateful to UCR and BCTV for permission to use these data. Neither company is responsible for the use made of the data in this article.

2 A study of the most recent Progressive Conservative leadership convention in Alberta had a response rate of 34 percent. See Gibbins, Roger and Hunziker, Margaret, “Issues and Leadership Conventions: The 1985 Alberta Progressive Conservative Leadership Convention,” paper presented to the annual meeting of the Canadian Political Science Association, Winnipeg, June 1986.Google Scholar

3 These data were made available by George Perlin of Queen's University. Professor Perlin bears no responsibility for the use made of these data in this article.

4 Courtney, J. C., “Leadership Conventions and the Development of the National Political Community in Canada,” in Carty, R. K. and Ward, W. P. (eds.), National Politics and Community in Canada (Vancouver: University of British Columbia Press, 1986), 106–07.Google Scholar

5 Only in the constituency of Delta did the MLA not take his place as a constituency delegate. That member, Walter Davidson, declared that as Speaker of the House he would not attend the convention. His place was otherwise filled by the selection meeting.

6 In the end, 41 of the 50 constituencies sent 25 delegates. The largest delegation, from the party's Central Fraser Valley heartland, was 40 members.

7 British Columbia's legislature is probably the most malapportioned in the country. See Carty, R. K., “The Electoral Boundary Revolution in Canada,” American Review of Canadian Studies 15 (1985), 273–87.CrossRefGoogle Scholar

8 Had those two-member ridings been awarded twice as many delegates as the single-member constituencies, Grace McCarthy would have emerged as a much stronger candidate since she attracted more than one-third of the delegates from two-member constituencies. This disadvantage was independent of the more general rural-urban differences in constituency size that worked to the advantage of rural interests.

9 In order to vote at the delegate selection meeting, individuals had to hold membership in the constituency association 30 days prior. The timing of selection meetings meant candidates had little opportunity to sign up new members.

10 For figures on the Liberal and Conservative conventions, see Carty, R. K., “Campaigning in the Trenches: The Transformation of Constituency Politics,” in Perlin, George (ed.), Party Democracy in Canada (Scarborough: Prentice-Hall, 1988), 84–96.Google Scholar

11 Courtney, “Leadership Conventions,” 102. Youth associations, campus clubs, and constituency youth delegates are the source of the many young people at national conventions. At the Social Credit leadership convention many observers were impressed by the number of young people. However, it turns out that most were there to work for candidates, not to vote for them.

12 Janine Brodie, “The Gender Factor and National Leadership Conventions in Canada,” in Perlin, Party Democracy in Canada, 176. An unexpected finding among the Social Credit delegates was the number of married couples who were delegates. An examination of the delegate lists revealed that virtually one-quarter of them (24.4%) were married to another delegate. With such intimate links amongst the delegates one would expect the political anthropology of the campaign and the convention to differ significantly from large-scale national contests. Unfortunately, we have no data to investigate this phenomenon.

13 About 80 per cent of national Conservative constituency delegates had similar educational achievements, and the income distributions were the same in both groups.

14 See Blake, Donald, Two Political Worlds: Parties and Voting in British Columbia (Vancouver: University of British Columbia Press, 1985).Google Scholar

15 Vander Zalm had run both provincially and federally as a Liberal, and also sought unsuccessfully the provincial Liberal leadership earlier in his career. Grace McCarthy was the only candidate who had served in W. A. C. Bennett's cabinet and also emphasized this theme at delegate meetings.

16 For a discussion of the coalition requirements of national as opposed to provincial parties and a comparison between the BC Social Credit party and the national parties, see Blake, Donald E., Carty, R. K. and Erickson, Lynda, “Ideology and Partisanship: A Comparison of Federal and Provincial Activists,” paper presented to the annual meeting of the Canadian Political Science Association, Hamilton, June 1987.Google Scholar

17 On the ideological homogeneity of provincial parties, we also have the example of Alberta, where Gibbins and Hunziker found little evidence of intra-party division on key policy issues and no evidence that policy preferences had any impact on the choice among leadership candidates in 1985. See Gibbins and Hunziker, “Issues and Leadership Conventions.”

18 Blake, Two Political Worlds, 32–35.

19 The “consensus index” equals the absolute value of 50 minus the percentage agreeing with a given statement. If everyone agreed (or disagreed) the score would be 50. A 50/50 split would yield a score of 0 as would a situation in which 50 per cent agreed and the remainder were divided into “disagree” and “no opinion.”

20 Detailed figures on attitudes towards government spending are available from the authors.

21 Since 1973, all land in British Columbia has been classified according to its suitability for agriculture. In order to remove land from the agricultural land reserve for commercial or residential development, which normally increases its value substantially, permission must be obtained from a regulatory agency, the BC Land Commission. Since its decisions can be appealed to cabinet, the appeal process frequently leads to political battles. Generally, this form of regulation is criticized by people involved in property development and real estate who are strong supporters of Social Credit.

22 Detailed figures on attitudes towards government regulation are available from the authors.

23 For details on scale construction procedures and a list of items in each scale see the Appendix.

24 The coefficient of variation is simply the standard deviation of a distribution divided by its mean. It provides a simple way of comparing the variation found in two distributions which have different means. The smaller the coefficient of variation, the less variation relative to the mean.

25 This scale was introduced in Blake, Two Political Worlds, as an “individual versus collective responsibility” scale. The change of name here reflects a change in the direction in which scale items were scored.

26 Opinions on the independence issue are linked empirically to populism. Populists are much more willing to bear the economic costs of preserving political independence. Since populist sentiments are widely shared among activists, any setbacks on the free trade question could produce problems in party management.

27 Grace McCarthy, Jim Nielson, Bill Ritchie, Stephen Rogers, Brian Smith. Rogers had just been dropped from the cabinet.

28 Cliff Michael, John Reynolds.

29 Kim Campbell, Bud Smith.

30 Bob Wenman.

31 Mel Couvelier.

32 Their supporters were far more approving of the use of professional organizers, modern advertising techniques and polls, as well as Bennett's party leadership style than were those of McCarthy or Vander Zalm. That said, it must be noted that a majority in most camps (except those of minor candidates Campbell and Rogers) disapproved of Bennett's leadership style and the emphasis on polls and advertising. Overall the disapprovers outnumbered the approvers by a margin of two to one. Moreover, divisions on this issue had little impact on the crucial final ballots, as noted below.

33 John Laschinger managed Bud Smith's campaign; Patrick Kinsella ran Brian Smith's.

34 This was not unnoticed by the delegates. Almost two-thirds of the delegates agreed that “some candidates spent too much money on their campaigns” and 57 per cent agreed that the party ought to put limits on campaign expenditures. These seem large proportions for the highly individualist, free-enterprise Socreds and probably reflect a reaction against the spending extravaganzas of the convention. It is not surprising that the supporters of Bud Smith and Brian Smith were less likely (and Vander Zalm's more likely) to agree with these statements than were delegates in general.

35 Vancouver Sun, July 18, 1987.

36 Geographical differences also characterized the campaigns of the minor candidates, almost all of whom had geographically limited support. Typically they won the bulk of the delegates from their home constituency and added scattered support of a personal kind.

37 Our survey indicates that Vander Zalm won first-ballot support from 64 per cent of the constituencies, Grace McCarthy won delegates from one-half of the constituencies, and all the rest won from far fewer.

38 Regression analysis pairing Bud Smith supporters against those voting for each of the three leading contenders is dominated by regional effects. Attitudinal differences appear only in Bud Smith versus Vander Zalm comparisons, with Smith's supporters significantly less populist than those of Vander Zalm. Here too, however, regional effects contribute more to the explained variance.

39 The data indicate that about one-quarter of the delegates changed their vote between any two successive ballots. Although the numbers abandoning their candidate dropped sharply as the convention moved on, just over 40 per cent voted the same way on the first and last ballots.

40 Only 15 per cent of the delegates committed to one of the eight minor candidates said their commitment was binding after the first ballot. That compares to 60 per cent of McCarthy, 59 per cent of Bud Smith, 54 per cent of Brian Smith, and 46 per cent of Vander Zalm delegates.

41 Given that delegates still had six choices, a majority is a tough test. But even if we were to use a plurality the story would be the same.

42 The pattern of delegates abandoning Nielson and Reynolds after the first ballot was similar. Those delegates appear to have gone in all directions with each of the major candidates winning some, but none of them winning a majority.

43 Only one-third of our respondents replied to the question, but that was virtually all to whom it applied; 62 per cent of the first-ballot voters never saw their candidate endorse another.

44 Of the Smith delegates who answered this question, 55 per cent indicated they had been influenced (though precisely how we do not know) by his crossing of the floor.

45 See, for example, Smiley, D. V., “The National Party Leadership Convention in Canada: A Preliminary Analysis,” this JOURNAL 1 (1968), 373–97.Google Scholar

46 The following independent variables, in addition to those in Table 9, were tested: federal vote (an indicator of connections to the federal Conservatives), all the other attitude scales in Table 7, and a three-item index combining items critical of “machine” politics. None was significant in the regression equations.