1 On the instability of party identification in Canada and the concomitant importance of short-term forces in federal elections see Clarke, Harold D., Jenson, Jane, LeDuc, Lawrence and Pammett, Jon, Political Choice in Canada (Toronto: McGraw-Hill Ryerson, 1979)Google Scholar, and Clarke, Harold D., Jenson, Jane, LeDuc, Lawrence and Pammett, Jon, Absent Mandate: Interpreting Change in Canadian Elections (2nd ed.; Agincourt, Gage, 1991)Google Scholar.

2 Reviews of the literature are Lewis-Beck, Michael S., Economics and Elections: The Major Western Democracies (Ann Arbor: University of Michigan Press, 1988)Google Scholar, and Miller, William L., “Studying How the Economy Affects Public Attitudes and Behavior: Problems and Prospects,” in Clarke, Harold D., Stewart, Marianne C. and Zuk, Gary, eds., Economic Decline and Political Change: Canada, Great Britain, the United States (Pittsburgh: University of Pittsburgh Press, 1989)Google Scholar.

3 See, for example, Archer, Keith and Johnson, Marquis, “Inflation, Unemployment and Canadian Federal Voting Behaviour,” this Journal 21 (1988), 569–585Google Scholar; Clarke et al., Political Choice in Canada, chaps. 8, 11, 12; and Clarke etal., Absent Mandate, chaps. 4, 6, 7.

4 See, for example, Carmichael, Calum M., “Economic Conditions and the Popularity of the Incumbent Party in Canada,” this Journal 23 (1990), 713–726Google Scholar; Clarke, Harold D. and Zuk, Gary, “The Politics of Party Popularity: Canada, 1974–1979,” Comparative Politics 20 (1987), 299–316CrossRefGoogle Scholar; Happy, Jack R., “Economic Performance and Retrospective Voting in Canadian Federal Elections,” this Journal 22 (1989), 377–387Google Scholar; and Monroe, Kristen and Erickson, Lynda, “The Economy and Political Support: The Canadian Case,” Journal of Politics 48 (1986), 616–647CrossRefGoogle Scholar.

5 The surveys were funded by the US National Science Foundation and involved telephone interviews of national probability samples of the Canadian electorate. Fieldwork was conducted by Canadian Facts, Ltd., Toronto. The weighted national sample sizes for the cross-sectional surveys are: 1983, N = 2,117; 1987, N = 1,877; 1988 pre-election, N = 2,215; 1988 post-election, N = 2,010; 1990, N = 1,967. The 1983, 1988 and 1990 surveys contain interlocking panel components; the 1987 survey is a cross-section only. The weighted sample size for the 1988 post-election-1990 national panel employed below is N = 1,120. Details concerning the sampling design and other technical aspects of the surveys are available from the authors upon request.

6 Key, V. O. Jr., The Responsible Electorate: Rationality in Presidential Voting, 1936–1960 (New York: Vintage Books, 1968), 7Google Scholar.

7 Question wordings are in note 11 below.

8 See Peffley, Mark, “The Voter as Juror: Attributing Responsibility for Economic Conditions,” in Eulau, Heinz and Lewis-Beck, Michael S., eds., Economic Conditions and Electoral Outcomes: The United States and Western Europe (New York: Agathon Press, 1985)Google Scholar.

9 See, for example, Lewis-Beck, Economics and Elections, chaps. 3, 4.

10 Unlike exploratory factor analysis (EFA), confirmatory factor analysis (CFA) provides goodness-of-fit tests of theoretically specified models, correlations between factors which are parameter estimates (rather than arbitrary values as in an obliquely rotated EFA) and correlations among error variances may be specified. See Bollen, Kenneth, Structural Equations with Latent Variables (New York: Wiley Interscience, 1989), chaps. 6, 7CrossRefGoogle Scholar. Here, CFA is implemented using LISREL 7's weighted least squares (WLS) estimation procedure. WLS relaxes the assumptions of interval-level measurement and multivariate normality and, therefore, suits attitudinal data that involve dichotomous or ordinal measures and skewed distributions. See Joreskog, Karl and Sorbom, Dag, LISREL 7: A Guide to the Program and Applications (Chicago: SPSS Inc., 1988), 192–193Google Scholar.

11 The six economic evaluation indices were constructed as follows: (1) Past, Self: Respondents were asked (a) “Do you think that you are financially better off now than you were four years ago, worse off, or are things about the same?” (b) “Do you think that government has had a great deal, something, or not much at all to do with this?” Responses were scored “better off” = +l, “same” = 0 and “worse off” = −1; “great deal” = + 2, “something” = + 1, “not much” = 0. Scores for (a) and (b) were multiplied to yield an index ranging from + 2 to − 2. (Z) Present, Self: An index ranging from +4 to − 4 was constructed by multiplying responses to (a) “Now let's think about the things you can buy or do, all the things which make up your material standard of living. Would you say that you are very satisfied (+2), fairly satisfied (+1), a little dissatisfied (−1), or very dissatisfied (−2) with the material side of your life right now?” (b) “Do you think government has a great deal (+2), something (+1), or not much at all (0) to do with this?” (3) Future, Self: An index ranging from +2 to −2 was constructed by multiplying responses to: (a) “Still thinking about the material side of things and looking ahead over the next three or four years, do you think that you will be better off (+1) or worse off (−1) or will things stay about the same (0)?” (b) “Will government have a great deal (+2), something (+1), or not much at all (0) to do with this?” (4) Past, National: “Would you say that the federal government has done a very good job (+ 2), a good job (+ 1), a poor job (−1), or a very poor job (−2) in handling the economy?” Qualified answers, for example, “depends,” were scored 0. (5) Present, National: Unlike the other categories no question referring explicitly to the federal government is available. The following is used as a proxy. “Thinking generally about how the Canadian economy is doing these days, would you say it is doing very well (+2), fairly well (+1), or not very well (0)?” Qualified answers were scored +1. (6) Future, National: An index ranging from +2 to −2 was constructed by multiplying responses to: (a) “Do you think the Canadian economy will get better (+1), worse (−1), or stay about the same (0) over the next year or so?” (b) “Will the government have a great deal (+ 2), something (+ 1), or not much at all (0) to do with this?” For all six indices “don't know” responses were scored 0.

12 See Kornberg, Allan and Clarke, Harold D., Citizens and Community: Political Support in a Representative Democracy (New York: Cambridge University Press, 1992), chap. 2Google Scholar. Note the three-factor model includes a correlated error covariance for the retrospective and prospective personal economic evaluations indicators.

13 When testing the goodness-of-fit of a CFA model, “[l]arge χ² values correspond to bad fit and small χ² values to good fit. The degrees of freedom serve as a standard by which to judge whether χ² is large or small.” See Joreskog and Sorbom, LISREL 7, 42. In this regard, note that a simple one-factor model for the 1983, 1988 and 1990 data has a markedly worse fit than the three-factor alternative. In 1990, for example, testing a one-factor model yields χ²₉= 116.69, p <. 001.

14 Frizzell, Alan, Pammett, Jon H. and Westell, Anthony, The Canadian General Election of 1988 (Ottawa: Carleton University Press, 1988), 95Google Scholar.

15 Respondents were asked if they “agreed' or “disagreed” that: (a) “the free trade agreement ensures Canada's future prosperity”; (b) “the free trade agreement threatens Canada's political independence” (c) “economically the free trade agreement helps some industries but it hurts others”; (d) “the free trade agreement threatens Canadian culture and the arts”; (e) “the free trade agreement benefits all of Canada economically, not just certain regions or provinces”; (f) “the free trade agreement could threaten important programmes such as unemployment insurance and medical care.” Items (a)-(e) were asked in the 1988 pre- and post-election surveys and in the 1990 survey; (f) was asked in the latter two surveys only.

16 Respondents were asked if they “agreed” or “disagreed” that: (a) “the GST will reduce the size of the federal deficit”; (b) “the GST will start a new round of inflation”; (c) “the GST will help pay for important social programmes like medical care and unemployment insurance”; (d) “the GST is unfair because poor people will pay the same rate as the rich.”

17 Factor scores provided by a three-factor principal components analysis model are used to build indices measuring sociotropic, egocentric and future-oriented economic evaluations.

18 See, for example, Clarke, Harold D. and Stewart, Marianne C., “Partisan Inconsistency and Partisan Change in Federal States: The Case of Canada,” American Journal of Political Science 31 (1987), 383–407CrossRefGoogle Scholar, and Martinez, Michael, “Partisan Reinforcement in Context and Cognition: Canadian Federal Partisanships, 1974–79,” American Journal of Political Science 34 (1990), 822–845CrossRefGoogle Scholar.

19 See, for example, Archer, Keith, “A Simultaneous Equation Model of Canadian Voting Behaviour,” this Journal 20 (1987), 533–572Google Scholar, and Clarke and Stewart, “Partisan Inconsistency and Partisan Change.”

20 The federal and provincial party identification measures are scored: very strong Conservative = +3, fairly strong Conservative = +2, not very strong or leaning Conservative = + 1, non-identifier=0, not very strong or leaning other party identifier= −1, fairly strong other party identifier= −2, very strong other party identifier = −3.

21 Age is measured in years; annual family income has nine categories ranging from under $ 10,000 per year = 1 to $80,000 per year or more = 9; gender is women = 1, men = 0; formal education is elementary or less = 1, some secondary = 2, completed secondary or technical, community college = 3, some university = 4, completed university (B.A., B.Sc. or more) = 5. Region/ethnicity is a set of dummy variables with Ontario as the reference category.

22 Persons in favour of the FTA are scored 1; those opposed are scored 0. “D.K.'s” and “no opinions” are removed from the analysis. Opinions about the GST are scored the same way.

23 See Aldrich, John and Nelson, Forrest, Linear Probability, Logit and Probit Models (Beverly Hills: Sage, 1984), 43CrossRefGoogle Scholar.

24 The correlation (r) between two thermometer scores is +.79. Cases (N = 8) with missing data are rescored to the thermometer means. CFA analyses of the 1983, 1988 and 1990 survey data indicate the utility of the governing party and governing party leader thermometers for measuring support for an incumbent federal government. These analyses show that Canadians do distinguish among their feelings about incumbent political authorities, the national political regime and the national political community. A four-factor support model specifying separate authorities, regime (parliament-civil service and judiciary) and community factors has an excellent fit in every case (for example, for 1990, χ²₇ = 13.37, p=.064, AGFI =.998). When the authorities-regime distinction is collapsed, the fit deteriorates markedly (for example, for 1990, χ¹₂=676.50, p=.000, AGFI=.912). The 1983 and 1988 results are presented in Kornberg and Clarke, Citizens and Community, chap. 4.

25 Respondents were asked “Overall, were you in favour of the Meech Lake Accord or opposed to it?” Those stating they were in favour were scored + 2, those opposed were scored 0 and “D.K.'s” were scored + 1.

26 The principal findings are reviewed in Lewis-Beck, Economics and Elections, chaps. 3, 4.

27 The vote intention question sequence is: (a) “If a federal election were held today, which party do you think you would vote for?” [If “refused,” “undecided” or “don't know,”] (b) “Well, are you leaning in any special direction? Which party do you favour right now?” Those providing a “party” answer to (a) or (b) are considered as intending to vote for that party. Those not providing a party answer are included in the base for calculating the percentages displayed in Figure 1.

28 See Archer and Johnson, “Inflation, Unemployment and Federal Voting Behaviour,” 580–84, and Clarke, Absent Mandate, 143–45.

29 The 1985 and 1986 party identification data are from national cross-sectional surveys (weighted N's= 1,852, and 2,000, respectively) conducted by Canadian Facts Ltd. Details are available from the authors upon request.