Class, crisis, and political ideology in Canada: recent trends

Class, crisis, and political ideology in Canada: recent trends

D O U G B A E R , E D W A R D C R A B B , and W I L L I A M A . J O H N S T O N Western Ontario*

La relation entre appartenance de classe et idiologie politique en periode de dPclin Pconomi- que est itudiie ici au moyen d’un modele B indicateurs multiples (LISREL). Ceci est la premiere analyse de son genre B utiliser du matiriel Canadian. Elle s’intkresse aux diffirences attitudinales entre classes face a la redistribution de la richesse et aux dipenses reliies aux programmes sociaux. Les donnies utilis6es sont des series longitudinales tiries des sondages sur la Qualiti de la vie, de 1977, 1979 et 1981. Nos rPsultats indiquent que les attitudes des repondants sont resties stables malgri le changement du climat Pconomique. Ceci dit, une certaine polarisation des attitudes est mise au jour entre les principales classes sociales, particulikrement entre 1977 et 1979. Nous Pvaluons la signification de ces rPsultats, en prttant une attention particulikre P une exploration his- torique et comparative des liens entre mobilisation politique et changement d’attitudes.

This paper presents the results of applying a multiple indicator model (LISREL) to an examination of the relationship between class location and political ideology in a period of serious economic decline. The analysis, which is the first of its type to be based on Canadian evidence, considers class differences in attitudes about the distribution of wealth and support for social welfare expenditures, using longitudinal data from the 1977, 1979, and 1981 Quality of Life surveys. The findings indicate that, despite the changing economic context, attitudes were generally stable at this time. However, the results demonstrate that at least some modest polarization of attitudes did occur among the major social classes, especially in the period from 1977 to 1979. The implications of these findings are assessed, with particular reference to a comparative and historical exploration of the links between political mobilization and attitude change.

University of

I N T R O D U C T I O N

This paper will examine the relationship between class and political ideology in Canada, particularly the effects of sustained economic crisis on this relationship.

We wish to thank Sam Clark, Fred Evers, and Robert Gardiner for their contributions to the paper. As well, Veronica D’Souza provided excellent-typing services. The order of authorship is alphabetical. This manuscript was received in January, 1986 and accepted July, 1986.

Rev. canad. SOC. & Anth. I Canad. Rev. SOC. & Anth. 24(1) 1987

2 D O U G B A E R , EDWARD GRABB, A N D W I L L I A M A . J O H N S T O N

TABLE I SELECTED ECONOMIC INDICATORS FOR THE PERIOD 1973 TO 1981

Unemployment Consumer Price Bank U.S. Dollar Family Rate Index' G.N.P.' Rate' Per Canadian Income'

'973 5.5 112.7 108 7.25 1.00 11,047* * '974 5.3 125.0 112 8.75 1.01

'975 6.9 138.5 113 9.00 .98 11,818 1976 7.' 148.9 1x9 9.00 .99

1978 8.4 175.2 126 9.50 3 4 12,556

1981 7.6 236.9 134 18.00 .84 12,594

1977 8.1 160.8 122 7.50 .91 12,237

'979 7.8 191.2 130 12.50 .86 1 2 # 4 1980 7.5 210.6 130 12.80 .84 12,784

* Thousand of millions. All statistics are from the Canada Year Book: 1980-81 or Statistics Canada (e.g., Family Incomes (1981) Consumer Income and Expenditure Division).

** Constant dollars for family only - no unattached individuals. 1 Canadian Statistical Review. January 1983. Statistics Canada 2 Mean for Year - Bank of Canada Review. July 1982. Bank of Canada 3 Family Incomes. 1981. Statistics Canada

There is little question that Canada, along with most advanced capitalist countries, has undergone a significant economic downturn in recent years. Especially since the mid 1970s there has occurred, with few exceptions, a general tendency toward higher inflation, higher unemployment, real income declines, and slowed growth within the world's industrial nations.

For Canada, the evidence reveals fairly consistently that the years between the mid-1970s and the early 1980s were difficult economically. As Table I shows, most major indicators suggest a serious recession in progress at this time, with an accelerating rise in the consumer price index, steadily increasing interest rates, slowed growth in gross national product, a weakening dollar, and declining or largely stagnant average incomes for Canadian families. Some fluctuation has occurred for unemployment, including a modest though brief improvement after the late 1970s; nonetheless, the jobless level remained significantly higher after the mid-1970s than at any other time in the post-war era. Moreover, research suggests that non-cyclical unemployment, the so-called 'natural' jobless rate that persists between cyclical peaks, was actually higher between the mid-1970s and 1981 than for any previous period since World War 11 (Riddell, 1985: 18).

Other indicators of economic difficulty include the regular increase in the number of business bankruptcies per year, the general decline in real wages earned, and some modest tendency toward greater income inequality after the mid-1970s (e.g., Calvert, 1984: 109; Wolfson, 1986: xviii). Together, these figures reveal why most analysts agree that this period represents the deepest, most serious, and most prolonged economic decline since the Depression (cf. Gonick, 1978: 108; Mandel, 1978: 9; Willox, 1980: 13; Kaliski, 1985: 129). It is also noteworthy that, despite some current signs of recovery, such as a drop in inflation and interest rates, recently worsening unemployment and further weakening of the dollar have

3 C L A S S , C R I S I S , A N D POLITICAL IDEOLOGY I N C A N A D A

occurred. On balance, analysts project significant economic problems for Canada and other nations for some time to come (cf. Economic Council of Canada, 1984).

The persistence of serious economic difficulties and the contrast between this period of decline and the rest of the post-war era raise questions concerning the effects of the changing economic context on political attitudes. The key question is whether the persistent and deepening economic decline experienced in Canada has had a discernible impact on the political orientations of different classes.

The obvious source on such a question is Marx’s classical analysis of capitalist economic crises. According to this view, capitalism’s recurrent and increasingly serious economic problems should gradually bring about the objective conditions of class polarization necessary, if not sufficient, for proletarian class awareness, class consciousness, and the ultimate establishment of socialism. Implicit in this basic formulation is the hypothesis that economic crises help promote, over time, a widening gap in the political ideologies of the major classes in capitalism (cf. Marx, 1844: 241,258-9; Marx and Engels, 1848: 7 0 - ~ 7 5 ; Marx, 1867: 625-8; Engels, 1895: 171-2).

Since Marx’s writings on this issue, several other analysts, both Marxist and non-Marxist, have made broadly similar arguments. Certain contemporary Marxists have offered quite strong versions of this same viewpoint (e.g., Nasser, 1975: 5, 23, 28). However, most recent Marxists do not claim an automatic or inevitable link between crisis and revolutionary consciousness. The prevalent view within modern Marxism seems to be that a fundamental economic crisis is a necessary but not sufficient element in the process of socialist transformation, and that capitalism’s recurrent downturns could spur heightened class conflict and ideological divergence between classes (e.g., Wright, 1975 : 35-6; Gamble and Walton, 1976: 197-206; Wright, 1978: 88-90,245-51; Poulantzas, 1978: 242-4; Anderson, 1980: 194-5).

Apart from the range of Marxist treatments of the issue, it is possible to find parallel views among non-Marxist writers concerning the impact of economic conditions on political ideology and behaviour. Non-Marxists, however, typically discuss this topic in terms of the absence of class conflict in good times as opposed to the existence of struggle during bad times. They suggest a lack of any clear ideological conflict or disparity across classes in most capitalist societies and attribute this to the prosperous economy and avoidance of crisis existing until recently (cf. Crosland, 1957; Dahrendorf, 1959; 1979; Galbraith, 1968; Lipset, 198ra; 1981b). Some of these writers, though, allow for the possibility that economic crises will surface periodically and that, during such declines, the resulting economic deprivation could provide a basis for ideological dissensus and potential political conflict between classes (e.g., Lipset, 1960: 243-9; 1981b: 5’19-21; see also Gurr, 1970: 352-3).

From this brief review it is evident that writers occupying a wide spectrum of theoretical and political orientations have either assumed or concluded that there is a likely connection between worsening economic conditions and increasing divergence in the political outlooks of different classes. It may be surprising, therefore, to find a shortage of clear and consistent empirical support for this hypothesis in the existing literature.

4 D O U G BAER, EDWARD GRABB, A N D W I L L I A M A. JOHNSTON

One problem with the available evidence is that it is far from extensive, especially in Canada where empirical analyses have not yet addressed this problem. Some relevant studies have been conducted elsewhere, primarily in the United States. Even these treatments, however, tend to shed only limited or indirect light on the possible relationship between economic conditions and class-based political ideology.

One body of work has assessed the contention of a pronounced rightward shift in the political attitudes of the American public. Certain observers see the recent electoral victories of conservative parties in the United States and elsewhere as evidence to support this claim; these developments, however, may not reflect some general change in public opinion as a whole but rather the enhanced visibility of the new right at the party level (cf. Hughey, 1982). More direct studies of public opinion indicate a possible shift to the right on economic and social policy issues during the current downturn. For example, there appears to have been a modest tendency for an increasing proportion of Americans to favour reduced government spending on social welfare and human services. Still, the small nature of this shift, coupled with the stable attitudes expressed on other domestic issues and on foreign policy, lends little support to the view that worsening economic conditions have significantly altered public opinion (Page and Shapiro, 1982; Chafetz and Ebaugh, 1983; Rielly, 1983).

Several other analyses that have sought to gauge the political impact of changing economic trends focused principally on voting behaviour. Generally speaking, these studies suggest that economic swings have an insignificant impact on voting patterns, except perhaps to reduce voter participation during hard times (e.g., Rosenstone, 1982). Although some earlier discussions indicate otherwise, recent evidence reveals no obvious tendency, for example, for people to vote either against incumbents during downturns or for incumbents during upswings (e.g., Owens and Colson, 1980). Some studies suggest that general economic conditions may affect political party preferences or evaluations. However, such effects seem to be tempered by a clear inclination for citizens to see their own economic grievances - personal unemployment or financial difficulties, for instance - in highly individualistic terms. In other words, rather than looking to the larger economic context as the main cause of their hardships, people in economic difficulties tend to see themselves or some set of immediate and unique circumstances as responsible for their position (e.g., Lane, 1962; Kinder and Kiewet, 1979; 1981). There is reason to suspect, in turn, that this orientation weakens the connection between personal economic woes and actions or judgments in the political sphere (e.g., Schlozman and Verba, 1979; Feldman, 1982; Kinder and Rhodebeck, 1982; but see Fiorina, 1978).

The latter conclusion, of course, need not indicate a general lack of concern with economic issues. Apparently, certain economic issues, including some of the social welfare concerns that were once so prominent, have recently declined in saliency, to be replaced by such considerations as crime, abortion, or drug use and by so-called 'post-materialist' problems like the environment, overpopulation, and nuclear war (e.g., Inglehart, 1981; Wayman and Stockton, 1983). Nevertheless, recent American evidence suggests that economic issues are still extremely important, so much so that a larger proportion of people than at any other time

5 C L A S S , C R I S I S , A N D POLITICAL I D E O L O G Y I N C A N A D A

since World War 11 rate the economy as the number one problem in society (Smith, 1980). This same study reveals, moreover, that popular concern over the economy correlates with changes over time in the consumer price index and other financial indicators. This finding implies that sometimes people’s attitudes are influenced significantly by economic fluctuations. In general, however, and especially with respect to public policy preferences, the long-term evidence indicates strong stability in the opinions of the population, even during economic shifts (cf. Page and Shapiro, 1982).

The foregoing review gives some indication of the extent to which economic changes influence public opinions. However, there is very little evidence in these studies and others which would confirm or refute the claim that class differences in political outlook are seriously affected by economic shifts. Only one of these analyses seems explicitly to have considered this question. Its results fail to reveal any class effect, at least to the extent that such class correlates as education and socioeconomic status have no appreciable influence on policy preferences (Chafetz and Ebaugh, 1983 : 291). A handful of other studies, all of them American, have addressed this question in some minor degree but have been equally inconclusive. For example, there is no clear tendency for those with low incomes to be more worried about economic problems than those with high incomes, even in bad times. Education has only a minor impact, and it is the highly educated, not the less educated, who reveal the greatest concern about economic matters (Smith, 1980: 176). A similar lack of strong association has been found in studies of the unemployed who, despite their economic plight, are only slightly more likely to differ from the employed on such questions as the value of government welfare programs. Using employment status as a proxy for class, writers in this area have concluded from such findings that ‘a low level of class consciousness’ exists in the American population with limited prospects for working-class ‘political mobilization‘ (Schlozman and Verba, 1979: 156; cf. Rosenstone, 1982: 44).

The lack of evidence to demonstrate that economic shifts have an impact on class differences in political attitudes might seem to dismiss the claim that economic crisis accentuates ideological divisions among classes. However, as we have seen, the existing research on this topic is rather sparse and, equally important, has centred almost entirely on the United States. An additional shortcoming of this work is its failure to address the conception and measurement of the class variable in any systematic fashion. The notion of class is alluded to in discussions of the findings, but what are really considered are various indirect indicators of class - education, income, employment status, or occupation -rather than class itself. This deficiency raises the possibility that any lack of strong evidence for class divisions in ideology is partly or wholly the result of an inefficient or inappropriate measurement of the class variable.

The current analysis is intended to fill some of these gaps in the existing literature: first, by adding to currently sparse research; second, by considering the question for the first time in the Canadian setting; and third, by employing an explicit measurement of class, one which is consistent with a growing body of recent work that has attempted to apply the classical Marxist conception of class to contemporary social analysis (cf. Wright and Perrone, 1977; Wright, 1979; Johnston and Ornstein, 1985). Using multiple indicator analysis of a three-wave

6 D O U G B A E R , E D W A R D G R A B B , A N D W I L L I A M A . J O H N S T O N

national survey, the paper explores the nature and extent of class differences in political ideology for the period between 1977 and 1981. The specific issue areas to be examined are particularly appropriate for an analysis of economic crisis effects, since they assess class attitudes concerning government policy changes to redistribute wealth to the disadvantaged and increase social welfare expenditures. If the economic downturn during this time has had an appreciable impact on ideology, these kinds of issues should be among the most likely to reveal significant shifts, given their focus on economic need and inequity. We shall now discuss the data and methods employed in the analysis. Following the presentation of results, certain similarities and differences between the recent period of economic decline and the Depression of the 1930s will be considered. Of particular concern are the possible links between attitude shifts and political mobilization and the role of organizational and institutional structures in facilitating or attenuating these linkages. While space precludes a more complete discussion, it is important to recognize that a historical and comparative dimension is essential to analyses of survey data. Our main contribution is to register this awareness and to provide some direction to a discussion of the politics of class in the late 1970s.

DATA A N D M E T H O D S

Sample The data in this paper result from an extensive study of Social Change in Canada undertaken at the Institute for Behavioural Research at York University. The main lines of this study involved national surveys of the Canadian population in 1977, 1979, and 1981.

In 1977 the sample was a multi-stage, stratified cluster sample where households were selected with equal probability. Eligible respondents were persons eighteen years of age or over. The sampling frame included all of Canada except the far north, remote areas, and residents of institutions. We examine results for that panel of subjects who participated in all three waves of the Quality of Life survey. We use as a criterion for inclusion membership in the work force (working at least twenty hours per week) in at least one of the three waves of the survey. Or the I 665 respondents who completed all three waves of the survey, 1090 are used in our analysis as a subsample of the class structure.

Class Debates over the nature and determinants of class position have been integral to sociological discourse for some time. Recently a number of Marxists have sought to reformulate the conception and measurement of class position (e.g., Wright, 1976; Carchedi, 1975; Poulantzas, 1968). From this debate, Wright’s approach has emerged as the most practicable and widely adopted, for various theoretical and empirical reasons. Hence, we have chosen to employ Wright’s definitions of class position throughout our analysis.’

For Wright, class position is defined by control over three processes. These are: I/ control over the means of production; z / control over labour power; and 3/ control over investments and resources. Through the employment of these criteria Wright constructs a class structure that consists of classes and contradictory


locations. There are three classes - the bourgeoisie, the working class and the petite bourgeoisie - and four contradictory locations - the small employers, managers, foremen, and semi-autonomous employees. (For an extended discussion of Wright, see Johnston and Ornstein, 1985.)

Our use of Wright’s typology was constrained by the information available in the 1977 questionnaire. In particular, lack of information on the size of firm made it impossible to distinguish small employers. Our analysis is thus limited to six categories: capitalists, petite bourgeoisie, managers, foremen, semi-autonomous employees, and workers.

Political Attitudes We employed two sets of indicators measuring respondents’ attitudes about the current uses and distribution of social wealth. One set of six indicators measured the degree to which respondents desired an expansion or contraction of social welfare programs. Respondents were given a 5-point rating scale, and asked if the government should spend ‘much more effort’ (coded I), ’more effort’ (coded z), ’about the same’ (coded 3), ’less effort’ (coded 4) or ’much less effort’ (coded 5). We reversed these scores so that a higher score indicated support for more government spending, while a lower score indicated support for cut-backs. The six items asked the respondent how much effort should be put into: I/ ‘health and medical care,’ z/ ‘helping the poor,’ 3/ ’eliminating discrimination against women,’ 4/ ‘protecting the rights of native people,’ 5 / ’workmen’s compensation,‘ and 6 / ’providing assistance to the unemployed.‘ We refer to these items as ’Social Welfare’ items.

The second set of three indicators measured attitudes towards policies to redistribute income and wealth. Three statements were given to respondents; for each respondents could ‘strongly agree’ (coded I), ’agree’ (coded 2 ) , ’neither agree nor disagree’ (coded 3)’ ‘disagree’ (coded 4), or ‘strongly disagree’ (coded 5). The three statements were as follows: I/ There is too much difference between rich and poor in this country. 2 / People with high incomes should pay a greater share of the total taxes than they do now. 3/ The government should provide jobs for Canadians who want to work but cannot find a job. These items are referred to as ‘Redistribution’ items. Like the ’Social Welfare‘ items, the scores were reversed so that a higher score indicated agreement with redistributive policies, a lower one represented disagreement with these policies.

To access the validity of our assumption that the nine indicators outlined above would form two latent variables - in factor analytic terms, the ’loadings’ on the redistribution variables would load on one ‘factor,’ while the six social welfare items would load on another - we employed a confirmatory factor analysis model for the 1977 wave (Joreskog and Sorbom, 1984). The hypothesized factor model fits the data relatively well, as indicated by the goodness of fit index (.980). The two latent variables - redistribution attitudes and social welfare attitudes - were relatively highly correlated (.769).3

The Longitudinal Panel Model Figure 1 provides a diagram of the three-wave panel model employed to evaluate


FIGURE 1

the impact of social class on changes in attitudes over time. It presents the model for the ’redistribution’ variable, measured at three points in time (q,, qz, q,) using three indicators (YI through Y3) at each time point.4 The p coefficients connecting the endogenous variables (q’s) represent the stability coefficients (cf. Wheaton et al., 1977)~ that is, the degree to which the various scores at time t can be predicted from scores at time t -1.5

The E terms in Figure I represent measurement errors. All things being equal, we would expect measurement errors for equivalent indicators to be correlated over timef6 an expectation verified by the discovery that the addition of correlated error terms substantially improved the ‘fit’ of our models. These correlated error terms represent many unmeasured variables (external to the model) which affect the measuryment of the latent indicators over time, and which confound the analysis if left unestimated.7

The exogenous variable in our model is social class. In fact, this ’variable’ is actually five different dummy variables, each treated as a fixed indicator.’ Thus, Y., in Figure I refers to five gamma parameters - Y,, through Y15. The terms connecting social class (or, more accurately, the five social class variables) with 1977 social welfare attitudes (q) represent class differences at the beginning of the 1977-81 time period. The terms connecting social class with 1979 and 1981 attitudes represent the impact of each social class variable on changes in attitudes, since previous attitudes have been controlled for.

R E S U L T S

As a backdrop to our consideration of the impact of social class on changes in attitudes over time, we present the individual item means for the variables employed in our analysis in Table 11. As mentioned earlier the original scores have been reversed so that a higher score implies support for redistribution or increased


TABLE 11

ltem

Mean Values

1977 1979 1981

Too much difference between rich and poor Government should provide jobs for those who High income people should pay more taxes Government expenditure: (a) Health and medical care (b) Protecting rights of native people (c) Providing assistance to the unemployed (d) Helping the poor (e) Eliminating discrimination against women ( f ) Workmen’s compensation

3.566

3.398 can’t find one 3.828

3.502 3.548 3.100

3,795 3.585 3.519

3.668 3.640 3.289

3.642

3.693

3.550 3.121

3.597 3.444

3.505 3.408 3.082 3.638 3.606 3.406

TABLE 111

REDISTRIBUTION (MEASUREMENT MODEL)

Lambda y Theta epsilon ltem 1977 1979 1981 1977 1979 1981

11 Too much difference rich, poor 1.00* 1.00* 1.00* ,705 .603 ,641 21Highincomesshouldpaymoretax ,760 ,909 1.039 ,787 ,656 ,716 3/ Government should provide jobs 1.098 1.066 1.187 ,948 ,846 ,814

Correlated error terms: [covariances] (e,) 1977-1979 1979-1981 1977-1 981

11 ,157 ,230 21 ,230 ,266 31 ,297 ,302

,210 , 2 3 2

.263

‘constrained to this value Chi-square = 70.77 df = 51 Goodness of fit index: ,991

government social welfare expenditures. The item means in Table II are for the entire sample (N = 1090).

I t is difficult to observe any distinct pattern in these data. Certainly, there is no overwhelming trend towards support for ’conservative’ or ‘right-wing’ positions, since agreement with the item ’there is too much difference between rich and poor’ increased from 1977 to 1981. On the other hand, from 1977 to 1979 and 1979 to 1981 support declined for I/ the government providing jobs for those who cannot find them on their own, 2/ more government expenditures to help the poor, and 3 / more government expenditures on workmen’s compensation.

We shall now analyze the relationship between social class and redistributive and social welfare attitudes. Table 111 provides measurement model information for the dependent measure, support for redistribution. The lambda y matrix (A elements in Figure 1) represents the factor loadings for the indicators at the three different time points. As might be expected, the values for each of the three years

10 DOUG B A E R , EDWARD G R A B B , A N D W I L L I A M A. J O H N S T O N

are quite close to each other. The theta epsilon matrix represents the variances of the measurement errors (E’S in Figure I) and the covariances of these measurement errors (the lines connecting the E terms in Figure I). To conserve space in this table, the lambda y matrix, originally a 9 x 3 matrix, is reduced to a 3 x 3 matrix, with zero elements deleted. Likewise, zero elements in the theta epsilon matrix are deleted. As seen in the table, the parameters in the measurement model (linking the indicators with the latent variables) did not change substantially from year to year.9

Table IV presents the results for the redistribution variable, and Table 111 gives the measurement model for this. For purposes of our main question (the effect of social class on attitudes towards redistribution), the primary elements of interest are contained in the gamma matrix. Included in the analysis were dummy variables representing capitalists (N = xi), managers (N = 56), petite bourgeoisie ( N = 107)~ semi-autonomous workers ( N = 78), foremen ( N = 65) and those individuals for whom occupational data were not available in 1977 ( N = 239; this inclildes individuals outside the workforce in 1977 who later entered it). Parameter estimates associated with these variables indicate the influence of the category in question relative to the ‘reference’ category, which is ‘workers’ ( N = 534). The first row of the gamma matrix (‘Redistrib. ’77’) represents the impact of social class on initial (1977) attitudes. Since the latent variable, redistribution, is scaled according to the variance of the original indicators, the coefficients can be interpreted as movement along the five-point scale. Thus, capitalists score .693 lower on the support for redistribution scale than workers. This represents more than two-thirds of a unit along a scale which varies by only four units (and the variance of which is .329). Despite the small N for the ’capitalist’ category, the difference between capitalists and workers is statistically significant ( t = 2.921, p I .oi). All other class categories - managerial, semi-autonomous employees, petite bourgeoisie, and foremen/supervisors - are also less likely to support redistribution than workers.

The second row of the gamma matrix in Table IV (’Redistrib. 79’) represents the impact of social class on 1979 attitudes, controlling for 1977 attitudes. In other words, these parameter estimates represent the influence of social class on changes in attitudes from 1977 to 1979. Here capitalists, managers, and foremen/ supervisors, already less in favour of redistribution than workers, appear to have moved even further to the right in relation to workers, given the fact that these class groupings already scored higher on the dependent measure than workers in 1977. It should be noted, however, that the parameter estimate for capitalists is not significant at p 5 .05.

The third row of the gamma matrix in Table IV (‘Redistrib ’81‘) represents the impact of social class on 1981 attitudes, controlling for both 1979 and 1977 attitudes. Because the available evidence suggests a second-order process (time 3 attitudes being predicted not only by time 2 attitudes but also time I attitudes), this row may provide a stronger test for the hypothesis that polarization has occurred. As can be seen, the coefficients are quite weak and do not point clearly to a polarization effect.

A global test was performed to ascertain whether, in general, the effects of the social class dummy variables on attitudes towards redistribution were statistically

11 CLASS, CRISIS, A N D POLITICAL IDEOLOGY IN CANADA

TABLE N STRUCTURAL EOUATION MODEL - REDISTRIBUTION

Beta Matrix (p) ('Stability Coefficients') Redistribution

1977 1979

1979 ,813

Gamma Matrix (t-values in brackets) (y elements in Figure 1)

1981 ,331 ,440

Petite Semi- Capitalists Managers Bourgeoisie Autonomous Foremen Missing

Redistribution '77 - ,693 -. 215 -.371 -.499 -.235 -.lo8 (-2.921) (-1,976) (-4.429) (-5.152) (-2.311) (-1.792

Redistribution '79 - . z j i --.218 +.007 -.077 ,234 - .022

(--1.~22) (-2.521) (+.098) (-,981) (-2.870) (.471) Redistribution '81 -.iz8 -.074 +.063 +.040 +.032 +.007

(.799) (.930) (+1.112) (+.611) (+.424) (+.181)

Psi MatrixandR'values*

R' ~

Redistribution '77 301 083 Redistribution '79 064 784 Redistribution '81 072 704 ('psi matrix elements presented here [diagonal elements] represent vanances of { elements in Figure I)

Model with Gamma elements constrained to equality for change terms (t-values in brackets) ~ ~~~ ~

Petite Semi- Capitalists Managers Bourgeoisie Autonomous Foremen

Redistribution '79 -.113* -.113* -,113* -.113* -.113*

Redistribution '81 +.022** +.on** +.022** + .022* +.022** (-2.354) (-2.354) (-2.354) ( - 2.354) (-2.3541

(+.536) (+.536) (+ ,536) (+.536) (+.536) ~~

' 1979 elements constrained to equality (representing change 1977-79) * * 1981 elements constrained to equality (representingchange 1979-81)

significant. This test consisted of constraining all of the gamma matrix elements to zero, and comparing the chi-square values of this model with the model presented in Table IV. The chi-square value for this test was 57.59 ( d j = is), which was significant at p = I .ooi. An examination of Table IV would suggest, however, that much of this is attributable to the impact of social class on initial attitudes, and not on changes in attitudes. The global test of the twelve 'change' parameters suggested that, in general, these terms were not statistically significant (chi-square = 18.46, df = 12, p I .IT). Because of the small Ns in some of the class categories, it might be argued that only an extremely large effect of class on change in attitudes could produce significant results for such a global test. We undertook a


test to ascertain whether or not there was a significant difference in the change in attitudes between workers and all other classes, since most of the literature dealing with 'polarization' speaks of polarization between workers and other classes. In this analysis, the 'gamma' parameters associated with capitalists, petite bourgeoisie, semi-autonomous employees, managers, and foremen were constrained to equality in each year (1979 and 1981) but not across years. This model is shown in Table IV, under the heading 'Gamma Matrix - Parameters Constrained to Equality.' Under this model, a significant difference in change in attitudes was found between 1977 and 1979 (gamma = -.113, t = -2.354). Briefly, there was an increase in the tendency for non-workers to favour redistribution less than workers. With fewer degrees of freedom, this test may be more sensitive to effects, if the assumption that capitalists, petite bourgeoisie, semi-autonomous employees, foremen, and managers behave alike holds.

While there is some evidence of polarization between 1977 and 1979, this finding is not replicated for the period 1979 to 1981 (gamma = +.ozz, t = .534). The analysis of change during this second time period provides a stronger test of a 'polarization' hypothesis, since the effects, not only of time t - i attitudes but also of time t-z attitudes, have been considered.

Very similar results were obtained with the social welfare variable. For the sake of brevity, these are not reported in a table, but key coefficients will be mentioned. The effects in this instance are not as dramatic as those with the 'redistribution' variable, but 1977 workers more likely favoured increased social welfare expenditures by the state than other classes. Most opposed to government social welfare expenditures was the petite bourgeoisie ( - . 2 5 2 ) , followed by foremen/ supervisors (- .z03), managers (-. 184), and then semi-autonomous employees (-.163).'" Except for capitalists, all groups differed significantly from workers a t p 5 .05. Social class seems to have influenced attitudes towards social welfare less than attitudes towards redistribution, since these coefficients are lower than those shown in Table rv for the 'redistribution' variable (both dependent measures have indicators which have 5-point response scales). But, with the exception of the findings for capitalists, the pattern was almost the same as the pattern for the 'redistribution' variable.

From 1977 to 1979, capitalists, who in 1977 held attitudes quite similar to those of workers, moved markedly to the right (-.303). Petite bourgeoisie members, who already represented the strongest opposition to government social welfare efforts, moved even further in that direction from 1977 to 1979 (an additional - . i i z ) . Finally, semi-autonomous employees, who in 1977 were somewhat more opposed to social welfare expenditures (-.163), moved much further in that direction (an additional -.151). Again, one can see a sort of polarization between workers and all other groups which, if anything, is accentuated in the time period 1977-79.

In the analysis reported in Table IV, social class was treated as an independent variable in the examination of change over time; we did not include controls for education or income. To a considerable extent, educational differences, as differences in credentialization and as differences in skills and resources, are built into Wright's model of classes. This is done in two ways: implicitly in Wright's initial formulation of the distinction between semi-autonomous employees and

13 C L A S S , C R I S I S , A N D P O L I T I C A L I D E O L O G Y I N C A N A D A

other workers (Wright, 1978); and explicitly in Wright’s (1985) more recent formulation that distinguishes ’credentialized‘ workers from those who are uncredentialized. To some extent, then, the addition of education in an analysis would represent over-partialling. On the other hand, income is an intervening variable in the connection between social class and political attitudes, since income differences are heavily determined by class differences (Wright, 1985 : 147-8, 165ff.). To control for income would, in this sense, understate the causal influence of social class on political attitudes, since social class is causally prior to income.

I t might be argued, though, that other measures of socioeconomic status could provide more parsimonious descriptions of status-related differences in attitudes or in changes in attitudes. We therefore replicated the analysis shown in Table IV with an analysis controlling for education and income” and one controlling for education, income, region, and age.’j

With the control variables added, very few changes emerged. None of the significant findings regarding differences in change over time was affected by the introduction of the control variables. If anything, the class polarization effects noted earlier became slightly stronger: with controls for education, income, region, and age, the global test of the twelve ’change’ parameters reported earlier as not significant (chi-square = 18.46, df = 12, p 5 .TI, became significant at p 5 .05 (chi-square = 21.16, df = 12). While none of the results concerning changes in attitudes differed, there were two slight differences in the initial position of the classes relative to each other in 1977. The difference between managers and the working class in 1977, which was significant without controls (coefficient of -.215, t = -1.976), became trivial and insignificant with controls (coefficient of -.038, t = -.362). And foremen, who differed significantly from the working class in 1977 without controls (-.235, t = -2.311), did not differ significantly from the working class with controls for education and income (-.169, t = -1.757). But the tendency for both managers and foremen to ’polarize’ with respect to the working class, and to exhibit an increased level of opposition to redistribution in relation to the latter from 1977 to 1979, remained and, if anything, increased slightly with the introduction of controls. ‘4

Thus far, our analysis has been based on the use of an independent variable for social class (actually, a set of dummy variables in the analysis), based on codings obtained from respondents in 1977. As indicated earlier, 239 of the I 090 respondents could not be coded into a class in 1977, as they did not supply enough information for coding. By 1979, 126 of these 239 respondents provided enough information to be coded as ‘working class.’ Although in general the ’missing’ group was quite close to the ‘working class‘ for both dependent measures, we evaluated this group of 126 to determine whether the large number of missing values in 1977 affected any of our results. Although not shown here, this analysis indicates that the scores on the two dependent measures of those who were ’missing’ in 1977 but ’working class’ in 1979 were almost identical to the scores of those consistently coded ’working class’ in 1977 and again in 1979.

This analysis of missing cases also makes possible an evaluation of the impact of changes in class position on attitudes, by coding a series of dummy variables representing respondents who moved from one class to another between 1977 and 1979. Of course, small Ns for each of the ‘change’ variables mitigated against any

14 D O U G B A E R , E D W A R D G R A E B , A N D W I L L I A M A . J O H N S T O N

significant findings, since most of the workforce remained in the same social class from 1977 to 1979; hence, our results must be considered speculative at best.'5 We analyzed both the immediate effect of a change in status (impact on 1977-79 attitude change) and slightly longer-term effects (impact on 1979-81 attitude change). Since we do not know when changes in status occurred, the 'immediate' effect could represent a time period varying from one month to two years. The Ns were not large enough for separate analysis of different origin classes, but a move from any class to the capitalist class was associated with increased opposition to redistribution (-.438, p 5 .ox). Those who moved from the semi-autonomous category to the working class underwent an 'immediate' attitude change toward support for redistribution (+.384, p I . o i ) . No other individual results were significant.

Earlier, we described how social class did not significantly effect changes in attitudes from 1977 to 1979 and 1979 to 1981 when 'global' tests of significance were undertaken. Actually, as might be recalled, the global tests yielded results which were almost, but not quite, significant a t p 5 .05 ( p I .IT for the redistribution variable and p 5 .07 for the social welfare variable). Using a slightly more sensible test which pooled all classes other than the working class, we did obtain results which were statistically significant, although almost all of this was attributable to changes from 1977 to 1979. When the 'global' tests outlined above were performed again on models which included the dummy variables representing changes in class status as control variables, we found that the significance levels for the global tests - hitherto almost significant - now became significant at the p I .05 level.I6

In summary, we find relatively strong class differences at the onset of our period of study (1977). These differences are maintained throughout the study period (1977-81) with a slight polarization between the working class and all other classes. Most of this polarization occurs between 1977 and 1979; there is almost no evidence of any change between 1979 and 1981.

D I S C U S S I O N

At first glance, our findings indicate considerable stability in political attitudes during the period under study, both for the population in general and for each class delineated. Neverthe1ess;there are some attitude shifts and class differences that merit consideration, especially when we compare our results with the generally inconclusive American studies reviewed at the beginning of the paper. For example, we have found as expected that, more than all other classes, workers are significantly more in favour of greater redistribution of wealth and increased support for social welfare programs. These differences are sustained, moreover, in spite of very small sample sizes for several of the class categories, most particularly the capitalists. What is more central to our main research problem is that findings also indicate at least some evidence of polarization over time. This is true to the extent that, during the 1977-9 period, non-workers generally became even less supportive of redistribution and social welfare, relative to the workers in the study. While this polarization does not seem to carry forward to 1981, it does lend some support to the view that deepening economic problems may be associated with


divergent class ideology, specifically around issues of resource distribution and access to social services. That we have found even this minor shift is notable when we consider that the limited sizes of certain class subsamples and the somewhat restricted variances of some measures must diminish the ability to detect ideological changes.

One matter likely to contribute to the lack of stronger ideological shifts is the short time period to which our analysis has been restricted. On the one hand, because the period from 1977 to 1981 involves just four years, it could be argued that this is an insufficient time span for fundamental attitude changes to have taken root in the popular consciousness. On the other hand, various analysts in the fields of collective behaviour and social movements have contended that the opposite argument is in order. It has been suggested, for example, that movement toward revolutionary upheavals will likely occur during the earlier, not the later, stages of a sharp economic reversal, particularly if the downturn follows a prolonged period of objective economic and social development (Davies, 1962; 1969; cf. Gurr, 1970: 46-56). Because of the similarities between this pattern and the historical context of our analysis, the four-year period examined might have been sufficiently long to reveal any serious ideological shifts should they have taken place.

Nevertheless, there is reason to ask whether the modest influence of economic decline on Canadian policy preferences, as well as its limited impact on class differences in political ideology, may undermine notions of a strong relationship between economic crisis and political ideology. The rather weak evidence uncovered for polarization during this difficult economic period constitutes a warning about the danger of reductionism: the assumption, in this case, of an automatic relationship between economic context and political ideology. Polariza- tion appears to be a contingent outcome, arising not simply from economic circumstances but from a wider array of cultural, political, and social factors. Therefore, it is instructive to direct our attention to the wider array of factors that may influence a connection between economics and politics.

In seeking to uncover other factors which may mediate the relationship between economics and ideology, we could consider evidence from the last major period of economic crisis, the Depression of the 1930s. Social analysts largely recognize that the Depression of the 1930s radicalized political life in North America. While an absence of the 'hard' evidence provided by national surveys probably renders the estimate of the extent of the radicalization somewhat imprecise, much indirect data suggest that the Great Depression had a much more pronounced effect on political opinion and discourse than we have established for the crisis of the 1970s.

In Canada, the emergence of protest movements and radical political parties, such as Social Credit in Alberta, the CCF, and the Union Nationale in QuObec, suggests an altered political environment and, more importantly, a transformed audience. Moreover, the clear radicalization of the political platforms and traditional political discourse, including Bennett's New Deal and Hepburn's rejuvenated Liberalism, indicates substantial shifts in popular consciousness at that time (Young, 1978; Creighton, 1970; McNaught, 1969). While u. s. radicalization remained largely within the traditional parties, students of electoral behaviour have identified the I ~ ~ O S , along with the 1890s, as a rare period of


partisan realignment. As McSeveny notes for the 1890s: ’The voting shifts of 1893-6 testified to the political impact of the depression and to the skill with which the Republican party exploited the economic issues created by that depression’ (1972: 224). Similarly for the I ~ ~ O S , a monograph on voting in the state of Washington found that the 1930s witnessed a classical critical realignment involving both switching and increased participation (Allen and Austin, 1979). Evidence from Pennsylvania and Wisconsin confirms this critical realignment but emphasizes the importance of political mobilization rather than switching (Prindle, 1979). Not only was there an immense national party turnover, but a clear radicalization of the Democratic Party. As Holt (1975: 41) observes of the 1932 and 1936 Democratic Party campaigns: ‘In one obvious sense, the switch from cooperation and national unity to assaults on economic royalists and demands for social justice represented a turn toward a more radical appeal.’ And this was not only true of political rhetoric but of the content of public policy. Sinclair (1977) has demonstrated that the New Deal offered both a new policy agenda and radical remedies for the nation’s problems, all of which greatly enhanced the ideological content of American politics. Generally, then, there can be little doubt that the I ~ ~ O S , unlike the I ~ ~ O S , was a period in which an economic crisis engendered a dramatic transformation of popular consciousness.

Our problem is to explain why broadly similar economic contexts generated quite different outcomes in the two periods. To do this we will combine the findings from two bodies of literature. The first is concerned with political mobilization, the second with social psychological factors that may be involved in the process of group attitudinal change. Our core speculation is that the relative stability of attitudes in the 1970s resulted from an absence of collective action and mobilization around class issues like redistribution and social welfare. Applying Tilly’s collective action approach to social movements we can briefly indicate what changes might have served to restrict collective action and, as we will argue, therefore restrict attitudinal change as well.

The collective action perspective on social movements employs four key concepts: interests, organization, mobilization, and opportunity. The first three are articulated within a mobilization model and pertain to the internal attributes of a collective actor. It is to these, particularly that of the working class, that we direct our attention. Interests are the shared advantages or disadvantages accruing from interaction with other collective actors and, together with the level of organization, common identity, and unifying structure within a group, establish the internal determinants of the capacity to act or mobilize. Mobilization essentially is the process by which actors move from being a passive aggregation to a collective force in public life (Tilly, 1978).

What is striking about the period of the late 1970s is the relative absence of mobilization around working-class economic issues. Whatever mobilization did occur tended to limit mass involvement dramatically and was, therefore, limited in its influence on attitudes. Unlike the Depression of the igjos, when grassroots organizations and collective action proliferated, the social movements of the 1970s either revolved around non-class axes or took forms which seriously restricted involvement.

Now, the collective action approach asks us to explain differences in mobilization


levels by inquiring into changes in the interests and organization of the collective actors. In terms of interests it may simply be that the broad safety net of social services implemented since the 1930s has reduced interests in collective action related to economic problems. Prior to the installation of unemployment insurance benefits, the impetus for unemployed workers to take action was undoubtedly much more powerful than it is now (cf. Rosenstone, 1982: 27). The level of organization is more difficult to assess. O n the one hand, an appreciable increase in the institutional completeness of working-class economic organizations is apparent. Here we refer to the major legacy of the I ~ ~ O S , industrial unionism. However, the development of such formal union organizations could have increased the separation between members and representatives, and hence decreased levels of membership involvement (McCarthy and Zald, 1979). Indeed, we suggest that it was precisely in the early stages of mass industrial unionism that inclusiveness, the extent to which a group absorbs its members’ whole lives, was highest. Correspondingly, netness, which in collective action terms is the extent that members of a group are linked by a specific kind of interpersonal bond, would be highest. Anyone familiar with the struggles to organize the unorganized is aware that the process profoundly influenced the social networks and consciousness of hundreds of thousands of ordinary people. Reduction in netness or inclusiveness due to the bureaucratization of working-class organization would more than offset any increased efficiency and, when combined with the decreased interest engendered by the welfare safety net, would explain lower levels of popular mobilization.

We think that the social psychological literature on the process of attitude change permits an understanding of how the relative absence of mobilization limited attitudinal change. In particular, it appears that for attitudinal change to be significant the need for change must be supported by surrounding events, others must appear to support the change, and the methods for and the obstacles against achieving change must be clearly demonstrated (Halloran, 1967: 59; cf. Cohen, 1964; Zimbardo and Ebbesen, 1969). Especially with the existing social services and the bureaucratization of union organizations to administer and institutionalize class action, it is possible that none of these preconditions for attitude change suffices to generate an ideological shift. Moreover, most research into the impact of social change on attitudes shows rather consistently that actions are more likely to shape attitudes than the opposite. ‘People do something first, then bring their attitudes in line with their behaviour’ (Triandis, 1972: 127). Hence, without effective pressures for action and mobilization by workers or other classes, it is uncertain whether any greater attitude shifts than those reported here will occur in the near future.

N O T E S

I With the publication of Classes (1985) Wright has recast his theory of class. Implicitly the notion of contradictory relations relied upon a triptych of position, interests, and consciousness. In other words the process of class formation involved objective structures generating subjective consciousness through the mediation of common interests. Roemer‘s (1982) elaboration of a theory of exploitation allowed Wright to


ground the notion of interests in a concept of exploitation. This sharpening led to a realization that the assignment of agents to positions should be exploitation based rather than domination based. Correspondingly, the class boundaries were redrawn. Since we are not directly testing the adequacy of Wright’s new class structure and because the empirical differences between the old and the new are minimal, we do not, and could not given our data, generate a class structure based on Wright’s new conception. To give readers a sense of the marginal empirical difference, Wright (1985: 187) notes that 96 per cent of the workers in his original conception are workers or marginal workers in his new conception.

z The operationalization of Wright proceeded from the creation of twenty-three occupational groups. First, all owners were withdrawn from these groups and placed in either the bourgeoisie or petite bourgeoisie depending upon their occupational title. For example, an individual who owned a firm and was classified as a financial manager would be assigned to the bourgeoisie. Second, the non-owning classes were assigned positions according to their classification, e.g., manager, foremen, or our estimates of the degree of autonomy of the occupation. €or example, a university professor would be assigned to the semi-autonomous category while an elementary school teacher would be assigned to the working class.

3 However, there is empirical justification for treating these two variables as distinct constructs. We employed a formal test to compare our two-factor model to a single- factor model (that is, a model in which the correlation between the two factors was fixed at 1.0). Under a single-factor model, the chi-square statistic rose from 91.13 (df = 26) to 125.06 (d f = 271, a difference which is significant at p 5 .ooi. This result provides a statistical justification for treating the two latent variables as distinct constructs.

4 Normally, each indicator would be given a separate label, viz., Yi through Y9. We chose to reuse the same labels in Figure I to emphasize the fact that the same indicators were being employed at three different time points.

5 The model shown in Figure I involves a second-order process: scores for q3 (time t ) are determined not only by scores at time t-1 but also by scores at time t -2 . In time- series terminology, the process has a ’memory’ which is longer than one time period. For the redistribution variable, the addition of a second-order term (p,) significantly (and substantially) improved the ’fit’ of the model, and was thus included.

6 The term ’autocorrelation’ is sometimes used in this regard (cf. Kessler and Greenberg, 1981: 15off.)

7 For both of the dependent measures (social welfare and redistribution), the addition of second-order autocorrelation terms improved the fit of the model substantially (in time series terminology, we estimated A R ~ models).

8 For a justification of the use of fixed-indicator exogenous dummy variables in Lisrel models, cf. Joreskog and Sorbom, 1v.1-1v.2.

9 Indeed, a formal statistical test of the measurement model in Table 111 against a model with corresponding lambda parameters across the three years constrained to equality revealed a non-significant difference (chi-square = 7.74, df = 42).

expenditure. The coefficients can be interpreted in terms of the metric of the original variables employed - that is, as movement along as five-point scale, from ’much more’ effort to ‘much less’ effort.

11 A global test to ascertain whether the effects of social class on attitudes towards social

10 Once again, a high score indicates a preference for less government attention and


welfare were significant yielded a chi-square value of 53.97 (df = 18), which is significant at p = ,001. The global test for the twelve ‘change’ parameters suggested that, in general, these terms were not statistically significant (chi-square = 20.81, df = 12, p < .07). However, a test of the proposition that capitalists, petite bourgeoisie, managers, foremen, and semi-autonomous employees all moved more to the right than did workers from 1977 to 1979 and again from 1979 to 1981 (all ‘gamma’ matrix elements in each year constrained to equality) yielded significant results from 1977 to 1979 (gamma = -.095, t = -3.102) but not from 1979 to 1981 (gamma = (.028, t = +1.003).

12 Education was coded with ten possible response categories from ’no schooling’ to ’professional degree or doctorate. ’ Income was coded with eighteen possible response categories.

(Newfoundland, Nova Scotia, New Brunswick, Prince Edward Island), QuPbec, the Prairies (Manitoba, Saskatchewan, Alberta) and British Columbia, with Ontario serving as a reference category. Age was coded in years.

14 For foremen, the coefficient increased in magnitude from -.218 ( t = -2.521) without controls to - .237 ( t = -3.050) with controls for education and income and - .246 ( t = -3.536) with controls for education, income, region, and age. For managers, the coefficient went from -.218 ( t = -2.521) without controls, to -.190 ( t = -2.264) with controls for education and income to -.182 ( t = -2.438) with controls for education, income, region, and age.

15 Such variables were coded: I/ working class to managerial, z/ working class to foremen, 31 working class to petite bourgeoisie, 4/ working class to semi-autonomous, and, 5/ semi-autonomous to working class, 6/ foremen to working class, 7/ any class to capitalist. In all other cases, N s were less than 10.

16 The control variables included the seven dummy variables for change in status and five dummy variables representing the 1979 class location of respondents coded ‘missing’ in 1977 but for whom class location information was available in 1979.

13 Region was coded as a set of four dummy variables for the Atlantic provinces

REFERENCES

Allen, H. and E. Austin 1979 ’From the Populist Era to the New Deal.’ Social Science History 3(2): 115-43 Anderson, P.A. 1980 Arguments Within English Marxism. London : Verso Calvert, John 1984 Government, Limited. Ottawa: Canadian Centre for Policy Alternatives Carchedi, G. 1975 ’Reproduction of Classes a t the Level of Production Relations.’ Economy and Society

Chafetz, J. and H. Ebaugh 1983 ’Growing Conservatism in the United States.’ Sociological Perspectives 26(3):

Cohen, A. 1964 Attitude Change and Social Influence. New York: Basic Books Creighton, D. 1970 Canada’s First Century. Toronto: Macmillan of Canada

4 : 361-417

275-98

20 DOUG B A E R , E D W A R D G R A E B , A N D W I L L I A M A . J O H N S T O N

Crosland, C.A.R. 1957 The Future of Socialism. New York: Schocken Dahrendorf, R. 1959 Class and Class Conflict in Industrial Society. Stanford: Stanford University Press 1979 Life Chances. London: Weidenfeld and Nicolson Davies, J. 1962 ’Toward a Theory of Revolution.’ American Sociological Review 27(1): 5-19 1969 ’The J-curve of Rising and Declining Satisfactions as a Cause of Some Great

Revolutions and a Contained Rebellion. Pp. 690-760 in H. Graham and T. Gurr (eds.), The History of Violence in America. New York: Bantam Books

Economic Council of Canada 1984 Steering the Course. Twenty-First Annual Review. Ottawa: Ministry of Supply and

Engels, F. 1895 ’Introduction to “The Class Struggles in France”.’ In Karl Marx: Selected Works

Feldman, S. 1982 ’Economic Self-Interest and Political Behavior. American Journal of Political

Fiorina, M. 1978 ’Economic Retrospective Voting in American National Elections. ’ American Journal

Galbraith, J .K 1968 The New Industrial State. New York: Signet Gamble, A. and P. Walton 1976 Capitalism in Crisis. London: Macmillan Gonick, C. 1978 Out of Work. Toronto: James Lorimer Gurr, T. R. 1970 Why Men Rebel. Princeton: Princeton University Press Halloran, J . 1967 Attitude Formation and Change. Westport: Greenwood Press Holt, J. 1967 ’The New Deal and the American Anti-Statist Tradition.’ Pp. 27-50 in J. Braeman

Hughey, M. 1982 ’The New Conservatism.’ Social Research 49(3): 781-829 Inglehart, R. 1981 ‘Post-Materialism in an Environment of Insecurity.’ American Political Science

Johnston, W. and M. Ornstein 1985 ’Social Class and Political Ideology in Canada.’ Canadian Review of Sociology and

Joreskog, K. and D. Sorbom 1984 LISREL VI: Analysis of Linear Structural Relationships by Maximum Likelihood and

Kaliski, Stephan F. 1985 ‘Trends, Changes and Imbalances: A Survey of the Canadian Labour Market.’

Services

(vol. I ) . New York: International Publishers

Science 26(3): 446-66

of Political Science z z ( z ) : 426-43

(ed.), The New Deal. ‘Columbus: Ohio State University Press

Review 75(4): 880-900

Anthropology 2 2 ( 3 ) : 369-93

Least Squares Methods. Chicago: National Educational Resources


Pp. 77-140 in W. Craig Riddell (ed.), Work and Pay: The Canadian Labour Market. Toronto: University of Toronto Press

Kessler, Ronald C. and David F. Greenberg 1981 Linear Panel Analysis: Models of Quantitative Change. New York: Academic Press Kinder, D. and D. Kiewet 1979 ’Economic Discontent and Political Behavior. ‘ American Journal of Political Science

1981 ’Sociotropic Politics: The American Case.’ British Journal of Political Science 11 :

Kinder, D. and L. Rhodebeck 1982 ‘Continuities in Support for Racial Equality, 1972-1976.’ Public Opinion Quarterly

Lane, R. 1962 Political Ideology. New York: The Free Press Lipset, S.M. 1960 Political Man. Garden City: Doubleday 1981a ’Whatever Happened to the Proletariat?’ Encounter 56: 18-34 1981b Political Man (2nd ed.). Baltimore: Johns Hopkins University Press Mandel, E. 1978 The Second Slump. London: New Left Books Marx, K. 1844 Economic and Philosophic Manuscripts of 1844 (1975 ed.). In Marx Engels Collected

1867 Capital (vol. I). New York: International Publishers Marx, K. and F. Engels 1848 The Communist Manifesto. New York: Washington Square Press

1979 The Dynamics of Social Movements. Cambridge, Mass. : Winthrop McNaught, K. 1969 The Pelican History of Canada. Middlesex: Penguin McSeveny, S.T. 1972 The Politics of Depression. New York: Oxford Press Nasser, A. 1975 ’The Twilight of Capitalism.’ Insurgent Sociologist 6(2): 5-28 Owens, J . and E. Colson 1980 ‘Economic Fluctuations and Congressional Elections.’ American Journal of Political

Page, B. and R. Shapiro 1982 ‘Changes in American Policy Preferences, 1955-79. ‘ Public Opinion Quarterly

Poulantzas, N. 1968 ’On Social Classes.’ New Left Review 78: 27-54 1978 State, Power, Socialism. London: New Left Books Prindle, D. 1979 ’Voter Turnout, Critical Elections and the New Deal Realignment.’ Social Science

Riddell, W. Craig 1985 ’Work and Pay: the Canadian Labour Market: an Overview.’ Pp. 1-75 in W. Craig

23(3): 495-527

129-61

46(2): 195-215

Works vol. 3). New York: International Publishers

McCarthy, J. and M. Zald

Science 24(3): 469-91

46(1): 24-42

History 3(2): 1 4 - 7 0

22 DOUG BAER, E D W A R D G R A B B , A N D WILLIAM A . JOHNSTON

Riddell (ed.), Work and Pay: The Canadian Labour Market. Toronto: University of Toronto Press

Rielly, 1. 1983 ‘American Opinion.’ Foreign Policy 50: 86-104 Roemer , J . 1982 A General Theory of Exploitation and Class. Cambridge: Harvard University Press Rosenstone, S. 1982 ’Economic Adversity and Voter Turnout. ‘ American Journal of Political Science

Schlozman, K. and S. Verba 1979 Injury to Insult. Cambridge: Harvard University Press Sinclair, B. 1977 ‘Party Realignment and the Transformation of the Political Agenda.’ American

Smith, T. 1980 ‘America’s Most Important Problem.’ Political Opinion Quarterly 44(2) : 164-80 Tilly, C. 1978 From Mobilization to Revolution. Reading, Mass: Addison- Wesley Triandis, H. 1972 Attitude and Attitude Change. New York: Wiley Verba, S. and K. Schlozman 1977 ’Unemployment, Class Consciousness and Radical Politics. ‘ Journal of Politics 39(2) :

Wayman, F. and R. Stockton 1983 ‘The Structure and Stability of Political Attitudes.’ Public Opinion Quarterly 47(3):

Wheaton, Blair, B. Muthen, D.F. Alwin and G.F. Summers 1977 ’Assessing Reliability and Stability in Panel Models.’ Pp. 84-136 in D.R. Heise

Willox, P. 1980 The Capital Crisis. Stockholm: Alinquist and Wiksell Wolfson, Michael C. 1986 ‘Stasis and Change - Income Inequality in Canada 1965-1983.’ Canadian Statistical

Wright, E.O. 1975 ’Alternative Perspectives in the Marxist Theory of Accumulation and Crisis.’

1976 ‘Class Boundaries in Advanced Capitalist Societies.‘ New Left Review 98: 3-42 1978 Class, Crisis, and the State. London: New Left Books 1979 Class Structure and Income Determination. New York: Academic Press 1985 Classes. London: Verso Wright, E.O. and L. Perrone 1977 ’Marxist Class Categories and Income Inequality. ‘ American Sociological Review

Young, W. 1978 Democracy and Discontent. Toronto: McGraw-Hill Ryerson Zimbardo, P. and E. Ebbesen 1969 Influencing Attitudes and Changing Behavior. Mento Park: Addison-Wesley

26(1): 25-46

Political Science Review 71(3): 940-53

291-323

329-46

(ed.),. Sociological Methodology 1977. San Francisco: Jossey-Bass

Review (February) : vi-xxvii

Insurgent Sociologist 6(1) : 5-40

441): 32-55

Documents

Class, crisis, and political ideology in Canada: recent trends