1 Among the few Canadian studies in this area are Erickson, Bonnie H., “Region, Knowledge, and Class Voting in Canada,“ Canadian Journal of Sociology 6 (1981), 121–44;CrossRefGoogle Scholar and Pammett, Jon H. and Whittington, Michael S. (eds.), Foundations of Political Culture: Political Socialization in Canada (Toronto: Macmillan, 1976).Google Scholar Erickson defines ”Knowledge of party class positions“ as ”rating the NDP more working-class oriented than either main party” (127) and hypothesizes that “knowledge of class positions would be related to early training, later socialization and self-exposure, and variations in the amount of political information readily available in a given environment” (141). Level of education, political interest, region and occupation proved to be significant predictors of knowledge.

In the United States, Neuman has noted that the National Election Studies series “includes over 2500 items about the personal characteristics, attitudes, and behavior of a representative sample of American citizens. Yet only ten of these items deal directly with political knowledge.” See Neuman, W. Russell, The Paradox of Mass Politics: Knowledge and Opinion in the American Electorate (Cambridge: Harvard University Press, 1986), 9.Google Scholar

2 Ibid., 14–16.

3 It was necessary in the 1984 Canadian National Election Study to ask the knowledge questions at the end of the interview. When they were placed earlier in the interview during the pre-test, a number of respondents were embarrassed by their ignorance of current political leaders. To minimize these effects, the master question was worded as follows: “We are interested in how well-known the provincial premiers are across Canada. Can you think of their names?” Few respondents were fooled by this effort to make the visibility of the premiers the issue, rather than the knowledge of the respondents.

4 This is based on the symbolic interactionist dictum that “things that are defined as real are real in their consequences.”

5 See Milbrath, Lester W., Political Participation: How and Why Do People Get Involved in Politics? (Chicago: Rand McNally, 1965), 64–72.Google Scholar

6 See Converse, Philip E., “The Nature of Belief Systems in Mass Publics,” in Apter, David E. (ed.), Ideology and Discontent (New York: Free Press, 1964), 209–13;Google ScholarConverse, Philip, “Public Opinion and Voting Behavior,” in Greenstein, Fred I. and Polsby, Nelson W. (eds.), Nongovernmental Politics, Vol. 4 of The Handbook of Political Science (Reading, Mass.: Addison-Wesley, 1975)Google Scholar, chap. 2. It makes sense to employ measures of constraint among normative beliefs which comprise ideology but it is more appropriate to ask about the external validity of descriptive beliefs in the senses described here.

7 Neuman, The Paradox of Mass Politics.

8 Ibid., 195–96.

9 Ibid., 196–98.

10 Ibid., 198–200.

11 Ibid., 200.

12 The 1984 Canadian National Election Study was a multi-stage, stratified cluster sample of the voting population (N = 3,377), with systematic oversampling of the less populous provinces. The weighted sample (N = 3,380) was used here. See Lambert, Ronald D., Brown, Steven D., Curtis, James E., Kay, Barry J. and Wilson, John M., 1984 Canadian National Election Study Codebook (University of Waterloo, February 1986).Google Scholar

13 The names of the 10 provincial premiers at the time the interviews were conducted in late 1984 and early 1985 were as follows: Brian Peckford (Newfoundland), James Lee (Prince Edward Island), John Buchanan (Nova Scotia), Richard Hatfield (New Brunswick), René Lévesque (Quebec), William Davis (Ontario), Howard Pawley (Manitoba), Grant Devine (Saskatchewan), Peter Lougheed (Alberta) and Bill Bennett (British Columbia). None of these premiers had been in office for less than two years. In Ontario, Frank Miller was sworn in as premier on February 8, 1985 but all of the interviews were completed prior to this date. It is worth noting that 25 respondents (all of them in Ontario and Quebec) were interviewed during the first week of February, and that 21 or 83.1 percent of them identified William Davis as the Ontario premier.

14 We did not judge the adequacy of respondents’ definitions of left and right apart from the exclusion of definitions that: (1) were solely evaluative in nature, for example, “unprincipled,” “bad”; (2) did not seem to take the task seriously, for example, associating left-wingers with a position in hockey; or (3) could not be coded. This meant the exclusion of 67 definitions of left and 88 definitions of right. (Compare Lambert, Ronald D., Curtis, James E., Brown, Steven D. and Kay, Barry J., ”In Search of Left/Right Beliefs in the Canadian Electorate,” this JOURNAL 19 [1986], 547.Google Scholar) The coding of this variable produced a three-point scale, as follows: respondent defined both left and right and rated the NDP to the left of the Liberals and Conservatives = 3; respondent defined both left and right or rated the NDP on the left = 2; respondent failed to define both left and right and failed to rate the NDP on the left = I.

15 Following Erickson, “Region, Knowledge, and Class Voting,” 127.

16 Lambert, Ronald D., “Question Design, Response Set and the Measurement of Left/Right Thinking in Survey Research,” this JOURNAL 16 (1983), 140;Google Scholar Lambert et al., “In Search of Left/Right Beliefs,” 556.

17 See Ogmundson, Rick, “On the Measurement of Party Class Positions: The Case of Canadian Federal Political Parties,” Canadian Review of Sociology and Anthropology 12 (1975), 566–69;Google ScholarGuppy, Neil, Freeman, Sabrina and Buchen, Shari, “Representing Canadians: Changes in the Economic Backgrounds of Federal Politicians, 1965–1984,” Canadian Review of Sociology and Anthropology 24 (1987), 423–25.Google Scholar

18 See, for example, Ogmundson, Rick, “On the Use of Party Image Variables to Measure the Political Distinctiveness of a Class Vote: The Canadian Case,” Canadian Journal of Sociology 1 (1975), 171.CrossRefGoogle Scholar

19 There are four components in Neuman's knowledge dimension, of which the knowledge of political figures component corresponds most closely to our factual knowledge dimension. Likewise, his conceptualization dimension is made up of differentiation and integration components, neither of which corresponds neatly with our conceptual knowledge dimension. Over nine studies, Neuman found average correlations of .34 and .30 between knowledge of political figures, on the one hand, and his measures of conceptual differentiation and integration, on the other. See Neuman, The Paradox of Mass Politics, 202.

20 Hyman, Herbert H., Wright, Charles R. and Reed, John Shelton, The Enduring Effects of Education (Chicago: University of Chicago Press, 1975), 18.Google Scholar For their analysis of Canadian data linking education to knowledge, see 118–22, 199–203.

21 See, for example, Beeghley, Leonard, “Social Class and Political Participation: A Review and an Explanation,” Sociological Forum 1 (1986), 499.CrossRefGoogle Scholar

22 Respondents were asked the following: “Some people do quite a lot in politics while others find they haven't the time or the interest. Thinking about federal politics, how often, if at all, havex you done any of the following things—often, sometimes, seldom or never? [Show card 3; read list]

-d. Try to convince friends to vote the same as you?

-e. Attend a political meeting or rally?

-g. Spend time working for a political party or a candidate?

-h. Contribute money to a political party or candidate?”

The distribution of answers for each question was skewed toward inactivity. We created an index ranging from 0 (respondent did not engage in any of the four activities “often” or “sometimes”) to 4 (respondent participated “often” or “sometimes” in all four activities). It should be noted that the four items used in this index represent the three degrees of political participation identified in Milbrath's ”hierarchy of political involvement,” that is, “convince friends” is what he calls a spectator activity, “attend a political meeting” and “contribute money” are transitional activities, and “work for a party” is gladiatorial activity. See Milbrath, , Political Participation, 18.Google Scholar

23 Respondents were asked about the following two activities, in the context of the question described in footnote 22, above: “How often do you read about politics in the newspapers and magazines?” “How often do you watch programs about politics on TV?”

24 We are positing the same kind of relationship between power and knowledge among the provinces that prevails between Canada and the United States.

25 See, for example, Curtis, James E. and Scott, William G. (eds.), Social Stratification: Canada (2nd ed.; Scarborough: Prentice-Hall, 1979), 1–17.Google Scholar

26 Lambert, Ronald D., Curtis, James E., Brown, Steven D. and Kay, Barry J., “Effects of Identification with Governing Parties on Feelings of Political Efficacy and Trust,” this JOURNAL 19 (1986), 718:Google ScholarKay, Barry J., Lambert, Ronald D., Brown, Steven D. and Curtis, James E., “Gender and Political Activity in Canada, 1965–1984,” this JOURNAL 20 (1987), 851–63.Google Scholar

27 MCA resembles multiple regression with dummy variables. This procedure produces eta and beta coefficients which, when squared, provide rough measures of the proportion of the total variance in the dependent variable that is “explained” by each predictor variable (1) without controls (eta) and (2) with controls for the other predictorxs (beta). The magnitude ofx betas indicates the relative importance of the various predictors used in the analysis. See Andrews, F., Morgan, J., Sonquist, J. and Klem, L., Multiple Classification Analysis (Ann Arbor: Institute for Social Research, 1973).Google Scholar A fuller version of Table 3, reporting unadjusted and adjusted means, is available from the authors on request.

28 Another 6 per cent cited radio and television, 19 per cent television and newspapers, and 11 per cent claimed to rely on all of them equally. Only 3 per cent mentioned radio only. See also Clarke, Harold D., Jenson, Jane, LeDuc, Lawrence and Pammett, Jon H., Political Choice in Canada (Toronto: McGraw-Hill Ryerson, 1979), 287–91.Google Scholar

29 In supplementary analyses, we used the parties for which respondents reported voting in 1984 (Liberal; PC; NDP; other; didn't vote) as a predictor variable in lieu of the dichotomous variable. The other nine variables in Table 3 were cxontrolled. The effects of this party vote variable were significant for both factual (F = 4.18; p ≤ s .01) and conceptual knowledge (F = 4.34; p ≤ .01). In both cases, New Democrats were most knowledgeable, non-voters were least knowledgeable, and supporters of the Liberals, Conservatives and “other” parties clustered around the grand means.

30 Examining the independent effects of political interest (Neuman's “salience” variable) showed that while its effects were statistically significant for both dependent variables, it was clearly less important than the other predictors.

31 See Ajzen, Icek and Fishbein, Martin, Understanding Attitudes and Predicting Social Behavior (Englewood Cliffs: Prentice-Hall, 1980)Google Scholar, chap. 6.

32 As opposed to the following: normative beliefs or opinions about what ought o prevail; evaluations of what is, in light of certain values or standards of judgment; affect or feelings of like/dislike or attraction/repulsion.

33 Lambert et al., “Effects of Identification with Governing Parties.”

34 Hyman et al., The Enduring Effects of Education, 18.

35 See Fletcher, Frederick J. and Gottlieb, Daphne F., “The Mass Media and the Political Process,” in Whittington, Michael S. and Williams, Glen (eds.). Canadian Politics in the 1980s (Toronto: Methuen, 1981)Google Scholar, chap. 8.