1 Teather, Lynne, “The Feminist Mosaic,” in Matheson, Gwen (ed.), Women in the Canadian Mosaic (Toronto: Peter Martin Associates, 1976), 334–35Google Scholar.

2 For an analysis of these changes in a comparative perspective, see O'Connor, Karen and McGlen, Nancy E., “Sexual Politics in Three Nations: Canada, Great Britain and the United States,” a paper presented at the 1977 Midwest Political Science Association,April 1977Google Scholar.

3 The 1965 and 1974 election studies, obtained from the Inter-University Consortium for Political and Social Research, were both national post-election surveys with N-sizes of 2,118 and 2,445, respectively. For the 1974 study, however, only a random half of all respondents were asked questions about specific political participation acts. The data in 1965 were originally collected by John Meisel, Maurice Pinard, Peter Regenstreif and Mildred Schwartz; the 1974 data, by Harold Clarke, Jane Jenson, Lawrence Leduc and Jon Pammett. Of course, neither the archive nor the original collectors of the data bear any responsibility for the analysis or interpretations presented here.

4 See, for example, Loon, Rick Van, “Political Participation in Canada: the 1965 Election,” this Journal 3 (1970), 388–89Google Scholar and Report of the Royal Commission on the Status of Women (Ottawa, 1970), 352–55Google Scholar.

5 Kirkpatrick, Jeane J., Political Women (New York: Basic Books, 1974), 13–15Google Scholar.

6 Ibid., 17–19.

7 Ibid., 18–19, 67. For a discussion of the impact of motherhood, see Lynn, Naomi B. and Flora, Cornelia B., “Motherhood and Political Participation: The Changing Sense of Self,” Journal of Political and Military Sociology 1 (1973), 91–103Google Scholar.

8 Unfortunately, only a handful of women in the 1965 and 1974 surveys specifically mention women-related issues. However, this by itself is not necessarily evidence of the irrelevance of the proposition. First, neither survey makes it a point to delve into these areas. Second, more frequent issue responses involving economic problems, welfare problems, and so forth may, in fact, mask issues such as equal pay, equal rights and the like. Nevertheless, this proposition will be examined at least indirectly by considering the political activity of women in Quebec during a period of extreme agitation for women's rights.

9 For a report on these trends in Canada and their future implications, see Cooke, Gail C. A. (ed.). Opportunity for Choice: A Goal for Women in Canada (Ottawa, 1976)Google Scholar.

10 Research has, however, been much more extensive in the United States. Among the first to look at differences between the sexes with regard to voting were Campbell, A., Converse, P. E., Miller, W. E. and Stokes, D. E., The American Voter (New York: J. Wiley & Sons, 1960), 483–93Google Scholar. Susan Welch has examined overtime differences with respect to voting, campaign and spectator activities, “Women as Political Animals? A Test of Some Explanations for Male-Female Participation Differences,” American Journal of Political Science 21 (1977), 711–30CrossRefGoogle Scholar. S. Verba and N. H. Nie, in their extensive study of a wide variety of political acts in 1967, found that women were overrepresented among those totally inactive and among those whose participation was limited to voting. At the other extreme, however, among “complete activists” women were only marginally underrepresented. Participation in America (New York: Harper and Row, 1972), 97–100Google Scholar.

11 For the exact wording of all questions used, see the Inter-University Consortium for (Social and) Political Research's 1965 and 1974 Canadian National Election Surveys. The response sets for the three campaign items (reading, meetings and aiding candidates) and for vote proselytizing differ in the two studies. In 1965, respondents replied yes (coded 1) or no (0) to the listed activities, but in 1974, the alternatives included: often, sometimes, seldom and never. Hence, to maintain comparability with the 1965 data set, the first three response possibilities in 1974 were all coded as 1, the latter response, 0. The campaign activism index itself was created by simply adding (the three) individual item scores and thus it ranged from 0 to 3. Vote proselytizing was not incorporated into the index because previous work had indicated sharper empirical differences between the sexes compared to the other campaign acts. See the authors' “Changing Patterns of Political Participation Between Canadian Men and Women, 1965–1974,” a paper presented at the Canadian Political Science Association,June 1977Google Scholar. The electoral turnout index was based on whether or not individuals had voted in the most recent federal and provincial elections. With turnout at both levels coded 0 for abstention, 1 for voting, the index ranged from 0 to 2.

12 Balch has differentiated the “classic” political efficacy scale into what he terms “external” and “internal” dimensions. The former is routinely measured by the questions: “I don't think that the government cares much what people like me think” and “People tike me don't have any say about what the government does.” These, he argues, refer “to the respondent's perceived probability of success at influencing public officials.” On the other hand, the two following items, he shows, tap “internal political efficacy”: “Sometimes politics and government seem so complicated that a person like me can't really understand what's going on” and “Voting is the only way people like me can have any say about how government runs things.” These, he suggests, “refer to the perceived presence or absence of limits to personal activity … the individual's confidence in his own abilities, regardless of political circumstances.” Balch, George I., “Multiple Indicators in Survey Research: The Concept of Political Efficacy,” Political Methodology 1 (1974), 1–43, quotations on 24. Responses to the two external efficacy items were scored 1 for disagree in the 1965 study and, to maintain comparability, in the 1974 survey both the responses disagree strongly and disagree were coded as 1; agree and agree strongly responses were coded as 0. Added together, these measures provided an index with the scores 0, 1, 2. For our measure of internal political efficacy, we employed only the single “politics is complicated” question because it alone was asked in both 1965 and 1974. While the 1965 study did use the “vote is the only way” measure, the 1974 study used the following item, which we regard as incomparable: “So many people vote in federal elections that it doesn't matter whether I vote or not.” Coded similarly as the two external efficacy items, our single internal efficacy measure distributed between 0 and 1Google Scholar.

13 The 1965 political interest question used was: “How much interest do you generally have in what is going on in politics?” The responses were coded: great deal (2), some (1), and not much (0). The 1974 version was: “We would also like to know whether you pay much attention to politics generally. I mean from day to day, when there isn't an election campaign going on. Would you say that you follow politics very closely (coded 2), fairly closely (1) or not much at all (0).”

14 A related problem with the measures of political involvement is that the concrete participation acts are limited to only one aspect of behaviour, albeit an important one, electoral activity. Questions dealing with contacting officials and solving local community problems were asked in 1974 but not in 1965. Thus, while a broader set of measures would have been preferable, we are constrained by our focus on change over time. Nevertheless, we feel an examination of male-female differences in electoral politics in addition to some general attitudinal measures can serve as a useful starting point. As well, for interest sake we report data from 1974 on contacting and community problems in footnote 16.

15 In using mean scores we recognize the data is ordinal at best. The justification is, as stated in the text, to provide a summary of the data. Apart from the fact many researchers do make the ordinal-to-interval assumption, later we employ a multivariate technique based on the assumption of strict ordinality.

16 Women also trailed men somewhat in 1974 on measures of contacting officials and solving local community problems. With regard to the former, 33 per cent of the men and 25 percent of the women indicated they did so at least “seldom.” With reference to the latter item, the figures are 38 per cent and 30 per cent, respectively.

17 In the United States, education, age and marital status have been found to be associated with support for efforts to strengthen and change the status of women in society. Harris, L., The Anguish of Change (New York: W. W. Norton, 1973), 95Google Scholar. The justification for examining the other variables is based more on Canadian-specific arguments. See below.

18 Nancy E. McGlen, “The Impact of Parenthood on Political Participation,” Western Political Quarterly (forthcoming).

19 See Cleverdon, Catherine L., The Woman Suffrage Movement in Canada (Toronto: University of Toronto Press, 1974), chap. 7Google Scholar.

20 Quebec women received the provincial vote in 1940, PEI women in 1922 and New Brunswick women in 1919. In Newfoundland a limited franchise was granted in 1925 and full equality in 1949. O'Connor and McGlen, “Sexual Politics in Three Nations,” Table III. Our classification of Maritimers as more traditional will be bome out by the data; however, it is worth noting that with regard to turnout, Maritime women participated more than men in 1965.

21 As will be seen, we take Ontario as a benchmark province for cosmopolitans. As Teather points out, the most recent women's movement had its origins in Ontario and continues to be most active there (“The Feminist Mosaic”).

22 The number of dummy variables (and in some instances their coding) used to represent demographic factors was constrained by statistical considerations, in view of the multivariate (probit) program used (see below). Overlapping characteristics and very high intercorrelations among subsets of independent variables proved to be the main problem to avoid due to multicoUinearity. On the problem of multicollinearity, see Blalock, H. M., “Correlated Independent Variables: The Problem of Multicollinearity,” in Tufte, E. (ed.), The Quantitative Analysis of Social Problems (Reading: Addison-Wesley, 1970), 418–25Google Scholar.

23 See McKelvey, Richard D. and Zavoina, William, “A Statistical Model for the Analysis of Ordinal Level Dependent Variables,” Journal of Mathematical Sociology 4 (1975), 103–20CrossRefGoogle Scholar and Aldrich, John and Cnudde, Charles E., “Probing the Bounds of Conventional Wisdom: A Comparison of Regression, Probit and Discriminant Analysis,” American Journal of Political Science 29 (1975), 571–608CrossRefGoogle Scholar.

24 The probit package utilized here was devised by Richard D. McKelvey. The model is discussed in McKelvey and Zavoina, “A Statistical Model.” An attempt was made to incorporate “interactive” dummies into the analysis, but for the most part these new variables were so highly correlated with the existing ones that multicollinearity became a serious problem.

25 In particular, each MLE “represents the amount of change in the dependent variable on its (hypothesized) underlying scale which is brought about by a unit change in the independent variable” (ibid., 114).

26 Aldrich and Cnudde, “Probing the Bounds,” 582.

27 All coefficients and other data mentioned below, but not presented, are available from the authors upon request.

28 This follows from the assumption of unit variance and the fact that probit also (necessarily) estimates points on the underlying scale associated with the observed ordinal response categories. These, in effect, constitute cutting points on that scale. Assuming, then, unit variance around the proneness figure on the scale, it is possible to determine what proportion(s) of the curve is cut off by the cutting point(s). In effect, this is equivalent to evaluating z scores with, of course, proportions interpreted in terms of probabilities. See McKelvey and Zavoina, “A Statistical Model,” especially, 114–15, 117–19. See also footnote 29.

29 In this instance, two cutting points were estimated, 0 and .97. (Probit routinely sets one to 0). Proneness figures below 0 would substantively mean individuals failed to vote in both elections while figures between 0 and .97 imply voting in one election only and above .97, voting in both elections, or, in terms of the actual ordinal response categories, fell into categories 0, 1 and 2, respectively. Now to determine the probabilities that Quebec francophone traditionalist women with an estimated proneness figure of 2.10 fall into the three categories, one simply evaluates the appropriate z scores. First, 0–2.10=-2.10 and hence the probability is .98 such individuals fall above the cutting point 0; therefore, the probability is .02, they fall below 0. Since the probability of falling above the cutting point .97 is .87—given a z score of (.97–2.10)=-1.13—the probability for the interval between 0 and .97 is (.98-.87=).11.

30 For a discussion of these events in Quebec, the issues involved and the impact the movement had, see Caroline Pestieau, “Women in Quebec,” in Matheson, Women in the Canadian Mosaic, 57–69. See also Dolment, Marcelle and Barthe, Marcel, La Femme au Québec (Montréal: Presses Libres, 1973)Google Scholar and Jean, Michele (ed.), Québécoises du 20e Siècle (Montréal: Presses Libres, 1974)Google Scholar.

31 Pestieau, “Women in Quebec,” 66.

32 Figures reported in Cook, Opportunity for Choice, 97, indicate an increase of 11 per cent (29% to 40%) in the labour force participation rate of women in Canada over the period 1961–1971. Other data for mothers show that between 1967 and 1973 the rate increased from 21 per cent to 35 per cent.

33 Other important social changes in this period that probably also signalled an easing in the restrictiveness of women's familial roles include declines in fertility rates, delays in childbearing and increases in childlessness (ibid., chap. 2). Indeed, it is possible that at least some of the changes reported above were due to some of these factors, although with the lack of appropriate data their impact cannot be tested.

34 There may be another reason for the unique effect of nonprofessional employment among francophone women. Because of the limited opportunities and incentives found in other aspects of their life, at least historically, francophone women may benefit from work force activity, whatever the kind.