class, work, and politics

Class, work, and politics“

WILLIAM JOHNSTON and MICHAEL D. ORNSTEIN York University

Le texte explore la relation entre la classes sociale et l’idkologie politique en utilisant les risultats d’un sondage-kchantillon effect& parmi environ 1800 Canadiens au travail. En se basant sur les catigories de classe de Marx et les dkveloppements thkoriques de Carchedi, I’auteur fait ressortir des diffkrences modkrkes entre les classes pour ce qui a trait i l’appui du mouvement ouvrier, les depenses encourues pour le bien-Ctre social et pour une nouvelle ripartition des revenus. Une analyse de rkgression dkmontre qu‘une attitude plus gauchiste dans la classe ourviere reflete certains aspects des conditions de travail. En effet, la situation au travail joue un r81e considkrable dans l’attitude politique, en- dehors des diffirences associkes i la classe sociale. Par contre, la famille et l’instruction ne se rkvklent pas comme itant des facteurs importants pour expliquer la relation entre la classe et l’idiologie politique.

The paper explores the relation between social class and political ideology using data from a simple survey of approximately 1800 Canadians with jobs. Using Marxist class categories, following theoretical developments by Carchedi, moderate class differences are found to support the labour movement, social welfare expenditures, and redistribution of income. A regression analysis shows that the more left-wing position of the working class reflects a number of aspects of working conditions. Indeed, aspects of the work situation have a considerable impact on political attitudes beyond those differences associated

* This is a revision of a paper presented at the Third Annual Conference on Blue Collar Workers and their Communities in Windsor, Ontario, 5 May 1979. We gratefully acknowledge the support of the Social Sciences and Humanities Research Council of Canada (grant #s75-0332) and of York University for providing the funds to collect the data employed here and supporting the project staff. This paper forms a part of the research conducted under the Social Change in Canada project, directed by Tom Atkinson, Bernard R. Blishen, Michael D. Ornstein, and H. Michael Stevenson at the Institute for Behavioural Research, York University. The Institute was responsible for the collection of the survey data. Tom Atkinson designed the questionnaire items dealing with work that are used here; H. Michael Stevenson and Michael Ornstein designed the political attitude measures. Michael Ornstein was partially supported by a leave fellowship from the SSHRCC while this paper was written. We wish to thank Debbie Gibb and Shirley Wiley for typing drafts of this paper.

This paper was received in February, 1980, revised and accepted November, 1980.

Rev. canad. SOC. & AnthJCanad. Rev. SOC. & Anth. 19(2) 1982

197 C L A S S , WORK, A N D P O L I T I C S

with social class. Family background and education, on the other hand, are evidently not important factors in explaining the relation between class and political ideology.

The study of political attitudes among the Canadian general public raises two central issues for Marxist social science. The first question concerns whether an understanding of the class structure - defined on the basis of the relations of production - provides a meaningful insight into political consciousness. Second, presuming an affirmative answer to the first question, it is necessary to identify the experiences that provide the basis for that consciousness. For example, the discovery that working class people give stronger support to socialized medicine than other classes says something about the current struggle over cutbacks in medical services, but it does not really tell us why workers take the position they do. Excepting perhaps a small, politicized minority, people are not likely to take political positions according to some logical deductions from a knowledge of their class positions. Instead, the conditions of everyday life, perhaps personal needs for social services, experiences while growing up, in school, and at work, and the character of family and social relations provide an experiential basis for political judgements. The problem is then to identify what elements of that complex web of everyday experiences play a central role in forming political attitudes.

This paper focusses on the problem of identifying the experiential roots of class differences in political ideology, using data from a representative sample survey of Canadians. We employ a definition of the class structure of advanced capitalist societies elaborated by G . Carchedi in a recent series of articles (i975a, b, c). As a preliminary step, we establish the existence of class differences in Canadians' views of three related measures of political ideology, labelled support for the labour movement, support for redistribution of income, and support for social welfare measures. The analysis then considers the impact of family background and the characteristics of the work situation on the three measures. Multiple regression is used to separate the effects of class, background, and work situation.'

The problem of discovering the social roots of class differences in political attitudes has hardly been a central focus of political sociology. It is difficult enough to find any previous research employing a structural definition of social classes which addresses the character of political differences among them, let alone explorations locating the roots of those differences. Still, the two issues are deeply entwined in a series of critiques of classical Marxism which began in the 1950s. In his The Future of Socialism (1956), C.A.R. Crossland set out the argument that a convergence in the values of the workmg class and the middle class was undercutting the traditional lines of conflict between social classes. Baldly stated, this 'embourgeoisement' thesis saw structural changes in the economies of advanced capitalist nations blocking the development of working-class consciousness by altering the life style, values, and ideology of the working class in the direction of middle-class patterns. These new patterns would provide a basis for a homogenization of values and political ideology. Within the discipline of sociology, the important, empirical studies of Westergaard (1965) and Goldthorpe et al. (1968a, 1968b, 1969) speak to this argument. Although not resolving the problem definitively, these British studies revealed that class differences in social life, economic circumstances, and political attitudes had not vanished.

198 WILLIAM J O H N S T O N A N D M I C H A E L O R N S T E I N

Another attack on the idea that social class might be relevant to the understanding of political ideology emerges from the numerous post-war studies of voting. The Canadian variant of this school is pretty much indistinguishable from the American tradition, of which it is explicitly derivative.’ There is very strong evidence that class is not much of a factor in voting in federal elections (see, for example, Alford, 1963; Anderson, 1966; Meisel, 1975: Ch. I; Clarke et al., 1979; Williams et al., 1979). Combine these results with the assumptions that politics is synonymous with electoral politics and that social class is adequately measured by occupation and/or education and/or income and the conclusion is that class has little or no impact on Canadian politics. The problem of relating ideology to aspects of everyday life therefore vanishes.

In fact, a number of studies of non-electoral aspects of political ideology succeed in finding class differences (for example, see the work of Centers (1949)~ Hamilton (1972), and Form and Rytina (1969) for the United States, of Benney and Geiss (1950) for England, and of Hamilton (1967) for France). Although no full-scale study of the Canadian population seeks to examine and explain the relation between class and ideology,3 two important local studies deal with the issue. In a survey conducted in London, Ontario, Rinehart and Okraku (1974) employ both a blue-collar/white-collar dichotomy and a more elaborate eight-category occupational typology to examine differentials in ‘class’ identification and support for alternate models of political power. Stevenson’s (1977) analysis of a survey of the federal constituency of Winnipeg North shows that ‘left-wing radicalism‘ is strongly correlated with a variable labelled ’ownership of the means of production,’ and with occupational status, income, trade union membership, and a number of other variables. Unfortunately, the primary indicator of social class, ’ownership of the means of production,’ actually measures both ownership and occupational differences ! Stevenson establishes the impact of trade union membership and unemployment on political attitudes. Finally, previous analysis of data from the national sample survey used in this paper also provides evidence that class differences in ideology exist in Canada (see Johnston and Ornstein, 1979; Ornstein, Stevenson, and Williams, 1980). These studies, however, only establish that there are differences in political ideology worth examining. They do not show what aspects of social experience are associated with the differences in political ideology.

Two studies focus directly on the problem of interest here. Lipsitz (1964) finds that assembly line, repair, utility, and skilled maintenance workers have different levels of work satisfaction, attitudes towards their union, fatalism, political radicalism, and tolerance. He concludes that the technology and social setting of the work situation shape the development of these attitudes. Of the four types of workers, assembly line workers are least satisfied with their work, more fatalistic, more politically radical and less ’tolerant. ’ Mann and Blackburn (1975) inquire into the relations between specific job characteristics and political ideology among the non-skilled working class. The extent of contact with supervisors, the quality of the physical environment, and the amount of physical stress on the job were found to be associated with radicalism and dissatisfaction with work. Individuals with high levels of job skill were more likely to be politically conservative. Their cursory examination of the effect of class background did not establish any relation to ideology.

199 C L A S S , W O R K , A N D P O L I T I C S

Finally, before proceeding with our empirical analysis, it is worth describing a recent theoretical attack, from within the Marxist tradition, rejecting the idea of a necessary connection between class position and ideology. Cutler, Hindess, Hirst, and Hussain argue that ‘[tlhere is no necessity for political and ideological forces to be polarized around the memberships of the different classes’ (1977: vol. I, 242). Hirst (1977) further elaborates this challenge into the position that there is a necessary non-correspondence between classes and ideologies and that no particular area of everyday life (for example, the experience of work under capitalism) need make an important contribution to the formation of political ideology.

Thus, there are both conflicting theoretical positions and conflicting empirical evidence on the issue of the relation between class and ideology. From our point of view, however, the main weakness of the body of material reviewed is that very little of it speaks directly to what we see as a critical issue. The question of whether classes in modern capitalism have differing political ideologies has deflected attention from an examination of the role sf various aspects of everyday life on ideology. We approach the issue from a ’classical’ Marxist perspective that, in our interpretation, leads to the prediction that work experiences play a central role in mediating the relation between class and ideology.

The abundant evidence that family background and education play important roles in determining occupation and income4 suggests that they should influence both class destinations and political ideology. In keeping with the results of the major studies of status attainment, we predict that the impact of family background on political ideology will be mediated by class position - hence it should have no effect on political ideology, once class is held constant. On the other hand, we predict that thecontribution of education is only partly mediated by social class so that education exercises a direct effect on political ideology, with class held constant. Our argument is that members of classes have different experiences whose combined weight provides a basis for the elaboration of ideological differences. For instance, trade union members are likely to have experiences and involvements that shape their political ideology. Since trade union membership differs across social classes, differences in political ideology should result. We will explore the impact of a number of important aspects of family background and work. We take the variables describing the work situation to mediate the relation between class and political attitudes. Logically, class position - based on relations to the means of production - has consequences for job conditions, not the reverse. In the terminology of causal modelling, class causes job conditions.

DATA A N D M E T H O D

The sample The survey data discussed here were collected as part of the project on ‘Social Change in Canada’ conducted between May and July of 1977, measuring perceptions of the quality of life, personal values, political attitudes, and personal characteristics of 3288 Canadians. Data were obtained in face-to-face interviews with a probability sample of persons 18 years of age and over living in households in Canada, excluding the Yukon and Northwest Territories, residents of the far northern regions of Canada, public and private institutions and Indian reserva-

200 W I L L I A M J O H N S T O N A N D M I C H A E L O R N S T E I N

tions. A multi-stage sampling procedure was used to select the sample. This analysis is limited to the 1787 respondents employed for twenty or more hours per week at the time of the survey.5

Defining social class We employ a concept of class developed in the work of Carchedi (i975a, b, c), who proceeds from an examination of the capitalist production process to delineate capitalist relations of production which in turn form the basis for the ’economic identification’ of social classes. While recognizing the importance of the ideological and political dimensions of social class, Carchedi’s procedure is warranted by his view that classes are ultimately determined by the economic level of a mode of production facilities, control over newly accumulated capital, and control over the labour process.

a / class is regarded as a set of discrete categories, not as some continuum; b / individuals are assigned to classes on the basis of their objective characteristics, not their subjective identification; and c / the primary considerations are the ownership and control of existing production facilities, control over newly accumulated capital, and control over the labour process.

Carchedi, unlike Poulantzos (i975), does not differentiate workers engaged in productive labour (i.e., workers in production, transportation, and the provision of certain services) from unproductive workers (in commerce, the financial sector, and some services). Nor does he resort to defining categories of workers occupyin ’contradictory locations’ between distinct class categories, as does Wright (1976).

Carchedi’s ‘economic identification’ of social classes rests on three dichotomous criteria: ownership of the means of production, performance of surplus labour, and the distinction between the functions of the ‘collective worker’ and the ‘global capitalist’ (i975b: 362-4). Of these criteria, ownership of the means of production plays the fundamental role, but all three combine to define relations of production. 7

The bourgeoisie and the proletariat occupy the two class positions characterized by a correspondence among the three class criteria. Members of the bourgeoisie exercise real ownership of the means of production, appropriate surplus labour from workers, and carry out the controlling and supervisory functions of the capitalist (Carchedi, i975b: 369). Members of the new middle class occupy work situations in which the three criteria do not coincide. Like the bourgeoisie, they perform supervisory and control roles in production but, like the working class, they do not own the means of production (Carchedi, i975b: 373). The petty bourgeoisie supports itself primarily by its own labour, which is often supplemen- ted by hired employees. The petty bourgeoisie is not identifiable within the pure capitalist mode of production; its place is in independent commodity production.

In operationalizing Carchedi’s categories, individuals are assigned to social classes on the basis of their responses to two questionnaire items that ascertained their occupations and whether or not they were self-employed. High and middle managers are included in the bourgeoisie, because they exercise real ownership of the means of production, obtain their salaries from revenue (i.e., from surplus

%

201 CLASS, W O R K , A N D POLITICS

value), and perform the function of the global capitalist. The new middle class includes individuals who control and maintain surveillance of the labour process. The petty bourgeoisie contains all the self-employed. The working class includes all employed workers who do not engage in the surveillance or supervision of production. The result is a class structure divided in the following proportions: bourgeoisie 7.7 per cent; petty bourgeoisie 13.8 per cent; middle class 19.6 per cent; and working class 59.9 per cent.

Political ideology In comparison with class, our measurement of political ideology is crude. Outside the specialized realm of voting and the perception of government institutions, there is certainly no consensus regarding the definition of the basic dimensions of political ideology. There are theoretical objections to two common solutions to the problem. The first of these, followed by Stevenson (1977) in the creation of his left-wing radicalism scale, is simply to factor analyze a large battery of candidate items retaining all items that load on the strongest factor. The problem is that the resulting scale is likely to represent some amalgam of a number of underlying traits that are correlated but might bear on different aspects of political ideology. A second alternative, employed by Form and Rytina (1969) and by Rinehart and Okraku ( ~ 9 7 4 ) ~ is to ask respondents to choose among two or more short statements of holistic political positions on some issue. But the statements presented need not correspond to views of any significant proportion of the population so the validity of the responses is in question. Our previous paper (1979) em- ploys a number of single item measures of political ideology. In the present instance, however, we felt that the unreliability of individual items would weaken the multiple regression analysis. We therefore chose to work with three short scales measuring support for the labour movement, support for redistribution of income, and support for social welfare measures.* We were only partly successful in realizing this aim - the inter-item correlations averaged only about .25. This low value is not encouraging and certainly reflects a lack of ‘constraint’ in popular political attitudes. Still, the scales are more reliable than the individual items - so we proceeded. We were guided in our choice of these scales by our previous research (Johnston and Ornstein, ~ 9 7 9 ) ~ which suggested that they would be likely to locate existing class differences in political ideology and by our assessment that they related to critical political issues.

The three measures are fairly strongly related: the correlation between support for redistribution and for social welfare is ,49; and those two scales have correlations with support for labour of .36 and .37, respectively. The strength of these relations suggests that there is some coherence in the political attitudes of Canadians. The stronger relation between support for redistribution and for social welfare is presumably a function of their both focussing on distributional issues in the sphere of consumption; support for labour relates to the sphere of production.

The labour process and the conditions of labour A total of eight measures describing the respondents‘ work situation are included in the analysis. They are intended to provide a general picture of the objective nature and subjective perception of jobs. They measure the following:9


a / the dichotomy between manual and non-manual workers, based on the Pineo, Porter and McRoberts (1977) classification of ’unit groups’; b / the dichotomy between a category of skilled workers and the remaining semi-, unskilled work force, also based on the Pineo, Porter, McRoberts categories; c / trade union membership, a dichotomy; d / the perceived quality of the work environment (using a thirteen-item scale); e / A four-point scale measuring difficulties with hours of work; f / whether the respondent often feels he or she is ’very tired physically’ when finished work, a dichotomy; g / whether the respondent is often ‘very tired mentally’ after work, a dichotomy; h / the rate of pay (measured as the logarithm of annual pay, in thousands of dollars to reduce heteroscedasticity).

In addition, a ninth variable, family income (scored as for the rate of pay) is included in this group of variables.

Background variables We first selected three conventional measures of characteristics of respondents’ families of origin - the education of the respondent’s parents and the occupation of his father. Educational attainment was measured in years of schooling and occupation with a set of sixteen occupational categories (from Pineo, Porter and McRoberts, 1977) that divides occupations according to levels of skill, the manual versus non-manual distinction, and industry. This detailed occupational categori- zation had no effect on political ideology, net of the impact of parents’ education, and was therefore dropped from the analysis. Following our previous arguments, it would have been better to employ a typology for father’s occupation also based on Marxist theory, and it is possible that, measured in this way, father’s occupation would have some effect. Unfortunately the present data do not allow us to test this possibility.

The respondent’s educational attainment, also measured in years, and family income (actually, the logarithm of its value in thousands of dollars, the conventional transformation to reduce heteroscedasticity) were also included in the analysis. Our previous analysis revealed that education affected political attitudes, even with social class held constant. Although we did not find much in the way of income effects, a number of other researchers, including Form and Rytina (1969), Free and Cantril(1967), and Hamilton (1972) do find income effects in other populations. The measure of the ability to buy consumer goods provided by income makes it a logical choice for inclusion in any analysis of class differentiation.

DATA A N A L Y S I S

We begin with an examination of the extent of class differences in the three measures of political attitudes, then proceed to the discussion of what specific factors might account for those differences. Table I gives the mean scores in each of the four class categories. On the scales of support for labour and support for

203 CLASS, WORK, A N D POLITICS

TABLE I POLITICAL AlTITUDES BY SOCIAL CLASS

Variable

Social Class

Working Middle Petty Class Class BourReoisie BourReoisie Total

Support for labour Mean 9.80 9.00 8.40 8.30 9.40 Standard Deviation 1.80 2.00 2.10 2.20 2.00 Difference from mean

in standard deviations .23 -.17 -.48 -.52 .oo Support for redistribution Mean 9.30 8.50 8.10 7.60 8.80 Standard Deviation 2.10 2.30 2.40 2.50 2.20 Difference from mean

in standard deviations .19 -.16 -.32 -.52 . 00

Support for social welfare Mean 23.20 21.50 21.00 20.10 22.30 Standard Deviation 4.20 4.10 4.20 4.10 4.30 Difference from mean

in standard deviations .21 -.20 -.31 - .52 .OO (1052) (350) (246) (139) (1787)

NOTE: N is in parentheses. Missing values are for: support for labour 24; support for redistribution 6; support for social welfare 3.

redistribution, a score of 9 is given to respondents taking the neutral position on all three items; a score of 12 is given to respondents who agree and of 15 to those who strongly agree with every statement; similarly scores of 3 and 6, respectively, would be given to respondents who strongly disagree and disagree with all the statements.

As Table I indicates, on the scale of support for labour the working class averages 9.8, about a point above the middle of the scale; the petty bourgeoisie and bourgeoisie have almost identical mean scores of 8.4 and 8.3, respectively; and the middle class is right at the centre, with an average score of precisely 9.0. The differences are in the expected direction, but there is nothing like complete polarization on these issues. Class explains 8.8 per cent of the variance in support for labour (see Table II, top row). A similar pattern is manifest for support for redistribution, but the overall mean is six-tenths of a point lower - only the working class has an average score above the neutral point. There is another more important difference: the petty bourgeoisie proves to be less conservative than the bourgeoisie on this issue, though still measurably more conservative than the middle class and considerably to the right of the working class. Overall, there is somewhat less class cleavage than was found in support for labour, and class accounts for 5.3 per cent of the variance in support for redistribution. The pattern of support for social welfare is almost identical to that for redistribution. All four mean scores fall between the mid-point on the scale (a score of 17) that would be given to respondents who wanted, on average, ’about the same effort’ in the eight social welfare areas and the score (25) they would obtain by wanting ‘more effort’


TABLE 11 VARIANCE IN SUPPORT FOR LABOUR, REDISTRIBUTION, A N D SOCIAL WELFARE EXPLAINED BY SOCIAL CLASS, JOB CHARACTERISTICS AND BACKGROUND (PERCENTAGES)

Variance Explained (adjusted for degrees offreedom)

Independent Holding Support for Support for Support for Variable Constant Labour Redistribution Social Welfare

Social Class Alone Job and Income Family Background Both

Social Class Job and Income Both

Family Background Social Class Both

Family Background Alone

Job and Income Alone

All

8.8 3.4 8.1 2.4 2.8 2.1 0.6* 1.0

10.8 8.6 5.0 3.9

14.8

5.3 0.7‘ 4.2 1.0 5.5 4.2 1.1 1.4

11.1 7.3 7.1 4.2

13.8

6.6 1.7 5.6 2.7 2.7 2.0 0.1* 0.4* 9.4 6.9 4.3 3.1

11.3

‘not significant at .05

in each area. Class explains 6.6 per cent of the variance for this variable. Note that there are no important differences in the standard deviations of the three political variables among the class categories.

The similarity of these findings is hardly surprising, given the strong correlations among the scales. The patterns correspond to the classical conceptions of class interest: the working class and bourgeoisie occupy, respectively, the left and right extremes; the petty bourgeoisie is more favourable to redistribution and social welfare than the bourgeoisie, but equally opposes the extension of trade union rights. The middle class is divided on these issues. It is difficult to say whether the class differences are ‘large’ or ’small.’ There is nothing like total polarization; but, on the other hand, if these results are compared to other studies of the relation between social structure and political attitudes and voting, the differences are substantial, especially in light of the relatively low reliability of the three scales.

Labour Process and Conditions of Work The introduction of the measures of working conditions produces a dramatic decline in the impact of social class on support for labour, indicating that we have succeeded in identifying underlying variables that serve to ’transmit’ just over half the total effect of class. Class alone explains 8.8 per cent of the variance in support for labour, but it raises the total explained variance by only 3.4 per cent when the nine measures of working conditions are first entered in the equation (all the variance estimates are in the first column of Table 11). The job variables alone explain 10.8 per cent of the variance in support for labour, 5.0 per cent above and beyond the effect of social class. So, job characteristics not only serve as

205 C L A S S , W O R K , A N D P O L I T I C S

intervening variables, they have independent effects that cannot be attributed to social class.

An examination of the regression coefficients, in Table 111 , shows the directions of these effects. First, comparing the coefficients, for the class variables (top four lines, first and fifth columns), the difference between the working class and the bourgeoisie falls from .75 to .46 standard deviations in support for labour. When working conditions are held constant (in column 3 ) , the petty bourgeoisie moves to the right of the bourgeoisie - but the shift is of numerically small magnitude. The relative positions of the four classes do not change substantially from the raw means in Table I.

Taking the effects of occupational characteristics alone, the regression coefficients (Table 111, third column, bottom) indicate that support for labour is stronger among workers who

a / hold unskilled and semi-skilled rather than skilled jobs; b / are members of trade unions - a very strong effect; c / report being frequently physically tired at the end of a day's work; d / do not report being mentally tired after work; and e / earn less pay on the job.

It is interesting that the measure of the perceived quality of the work situation, a very reliable thirteen-item scale, has no independent effect at all, suggesting that the underlying conditions of work rather than perceptions of satisfaction shape attitudes towards the rights of labour and management. Perhaps perceptions of work are coloured by workers' expectations, which are a function of those underlying physical conditions. Our finding about the effects of physical and mental tiredness is interesting - these two variables obviously contain a mixture of objective and perceptual factors, but we cannot say in what proportions. It is also notable (Table 111, bottom row) that family income has no effect on political attitudes, once job conditions are taken into account; nor do problems concerning hours of work have any impact.

The fifth column of Table III shows what happens to the effects of job conditions, when social class is held constant. There are two dramatic changes - the effect of trade union membership is reduced from .52 to .19 (standard deviations between union and non-union members) and the effect of level of skill disappears. The impact of trade union membership on political attitudes is thus, in large measure, a reflection of class differences. Support for labour cannot be reduced to a simple calculus whereby members support their unions and non-members do not. Finally, a comparison of columns 5 and 6 of Table 111 shows that holding social background constant does not alter the impact of job characteristics.

Because of the strong similarity between the measures of support for redistribution and social welfare, they may be discussed together. But first, consider the difference between these two scales and for support for labour. Support for redistribution and social welfare emphasizes differences in levels of consumption and social relations outside the workplace; they do not refer directly to the social relations of production. Therefore, they should be less tied to job characteristics than support for labour and more attached to indicators of social relations and consumption - in this case measured by education and family income

TABL

E m

lE

GR

ESS

lON

OF SUP

PORT

FOP U

BO

UR

, RE

DIST

ILIB

UTIO

N, A

N0

SO

CIA

L W

ELFA

RE (A

LL S

TAND

ARDK

ED] O

N S

OC

IAL

CU

SS

. B

AC

KG

RO

UN

D A

ND

IO

B CH

AIlA

ACT

ERlS

TLCS

A

ND

lN

CO

ME

t

Regr

essio

n co

effic

ient

Re

gres

sion

Coef

ficie

nt

Regr

essio

n Co

effic

ient

fo

r Su

ppor

t for

Lab

our

for S

uppo

rt fo

r So

cial

Wel

fare

fo

r Sup

port

for

Redi

strib

utio

n In

depe

nden

t

Soci

nl C

lass

w

orla

ng ci

a5

5

.23

.22

.16

.17

.19

.16

.07

.08

.21

.19

.I2

.13

Mid

dle d

us

- .17

' -.1

3'

-.08'

-.08'

-.16'

-.1

9 -.0

1 -.0

1 -.2

0'

-.17'

-.0

9'

-.0

9*

-.35'

-.2

4'

-.26'

P

my

bour

geoi

sie

-.48'

-.5

1*

-.39'

-.4

3'

-.32'

-.3

7 -.2

3*

-.28*

-.3

1'

-.44'

-.23'

-.2

8*

Bour

geoi

sie

-.52'

-.4

5'

-.30-

-.30'

-.52*

-.3

2 -.

06

-.06

-.5

2'

Back

grou

nd

Fath

er's

educ

anon

s .w

-.0

1

.w

-.w

-.w

-.01

-.04

-.05

-.04

Mot

her's

edu

ca.an

ons

-.11'

-.11

' -.l

o' -.1

1'

-.lo'

-.lo'

-.06

-.

06

* -.04

Resp

onde

nt's

educ

atio

ns

-.09'

70

6'

-.03

-.17'

- .1

4'

-.06'

-.1

1 - .0

7'

.w

Vnr

inbl

e (1)

(21

(3)

(4)

(51

(61

(7)

(81

(9)

(10)

(11)

(1

21

(131

(1

4)

(15)

(16)

(1

7)

(18)

lob

and

Inco

me

Man

ual o

ccup

atio

n .0

6 -.0

1 -.0

3 .1

4'

.11*

.0

4 .1

6' .10

.03

SH

ed oc

cupa

tion

-.11'

-.01

.w

- 17

' -.2

4'

-.15'

- 0

7' -.

l3*

-.11

Tr

ade

wo

n m

embe

rshi

p .5

2' .1

9'

.18*

.2

8' .2

2'

.20'

.2

7'

-.19'

-

11'

Pem

rvcd

qua

lity

of w

ork'

.w

.w

.w

.w

03

.03

.17'

-.0

9'

-09'

D

iffic

ulty

with

wor

khou

rs'

-.03

-04

-.04

.06'

.0

6'

.05*

.0

5'

.04

.04

Rate

of p

ay'

-.12'

-.

08'

- 08'

-.15'

-.1

4'

-.13'

-.1

1'

-.08

-.08'

Fam

ily in

mm

e' - .0

5 -.0

5 -.

03

-.09'

-.l

o' -.0

8'

-.06'

-.

06

-.0

4

Phys

ical

ly ti

red

by jo

b .2

0' .2

2 .1

9*

.20'

22

. .1

9' .0

9'

.IS'

.17'

M

enta

lly n

red

by jo

b - .1

4*

-.15

- 12

. - .18'

-.la'

-.15'

-.0

9'

-.18*

-.16'

'Significantly

diff

eren

t fro

m z

ero

at th

e .0

5 le

vel,

for s

ocia

l das

s, c

ompa

rison

of

othe

r thr

ee d

arsc

s to

the

wor

lung

cla

ss

)Except

for t

he fo

ur c

lus v

aria

bles

, whi

ch m

easu

re d

iffer

ence

s fr

om th

e m

ean

'sta

ndar

dize

d va

riabl

e

207 C L A S S , W O R K , A N D POLITICS

(as distinct from rate of pay). By this reasoning they should be less tied to social class itself than was support for labour, because there is considerable overlap in the educational attainment and income levels of the four classes.

The results do little to support these hypotheses, First, jobs play an even stronger role in mediating the relation between social class and support for redistribution and social welfare, than they did in the case of support for labour. Social class alone explains 5.3 per cent of the variance in support for redistribution, but its effect net of job characteristics is only . 7 per cent; for support for social welfare the corresponding figures are 6.6 per cent and 1 .7 per cent (see Table 11,

columns z and 3). Furthermore, working conditions (including a rather small effect of family income) have strong effects on the two scores explaining 11.7 per cent and 9.4 per cent of the variance, respectively, with no other factors considered. Trade union membership has a stronger effect on support for labour than on support for redistribution and social welfare, but when social class is held constant, union membership has approximately equal effects on all three variables.

Holding social class constant, the effects of a number of features of the job are unaltered. Support for redistribution is a function of the differences between manual and non-manual workers and between skilled and the combination of semi- and unshlled workers, trade union membership, difficulty with work hours, phyical and mental tiredness, and rate of pay. All the effects are in the directions observed in the previous analysis of support for labour, except for the effect of mental fatigue. Workers faced with more unpleasant and less remunerative jobs take more left-wing positions.

When job conditions are held constant, a considerable alteration in the positions of the four social classes takes place. On the scale of support for redistribution, there is a remarkable decline in the difference between the two extreme classes, from .71 standard deviations to only .30. Originally the classes are spread out on a continuum from the working class on the left, through the middle class and petty bourgeoisie, to the bourgeoisie on the right. When job conditions are held constant, there is little difference among the worhng class (which is still on the left), the middle class, and the bourgeoisie. The petty bourgeoisie is farthest on the right, perhaps suggesting their greater fear of being levelled to the condition of the working class - a threat that members of the bourgeoisie are confident they have the political strength to combat. The class differences in support for social welfare strongly resemble those for support for labour. Once more, the insertion of the three background variables does little to change the patterns just described.

Background effects The fact that the introduction of statistical controls for class background has little impact on the relation among social class, job characteristics, and political attitudes, means that a political socialization model is not likely to be of much use in explaining the relation between class and ideology.'" The impact of family background is generally not as strong as that of either social class or jobs. With no other variables held constant, it explains 2 . 8 per cent of the variance in support for labour, 5.5 per cent in support for redistribution, and 2 . 7 per cent in support for social welfare (in Table 11). These effects are reduced by about one-quarter when social class is held constant, by at least three-quarters when job conditions are


added. With class and job held constant, the values of the explained variance are reduced, respectively, to 1.0 per cent, 1.4 per cent and 0.4 per cent.

The most striking result of an examination of the regression coefficients (in Table 111, columns 6, 12, 18) is that educational attainment, of respondent and mother, is negatively related to support for left-wing political positions. Father’s education has almost no effect, when the other two variables are held constant. The introduction of controls for job characteristics serves to change the relative importance of mother’s and respondent’s education: the insertion of job as a control dramatically lowers the effect of respondent’s education, but does little to alter the impact of mother’s education. This makes sense, for the studies of occupational attainment all reveal much stronger relations between respondent’s jobs and their own levels of education than between jobs and their parents’ education and occupations. Job conditions mediate the effect of education, as well as that of class, on political ideology.

A comparison of class differences, with and without education held constant, shows that the controls for education, as for job conditions, alter the positions of the petty bourgeoisie and bourgeoisie - the former holding a more liberal position when job and background are held constant (compare columns I and 4,7 and io,13 and 16). Thus much more of the evident conservatism of the bourgeoisie can be laid down to other factors - higher levels of education, non-union jobs, higher pay and family income, and so on- than is true for the petty bourgeoisie. This suggests that the class condition of the petty bourgeoisie is an independently conservatizing element, that is absent for other classes, including the bourgeoisie. Still, these differences are of relatively small magnitude. Also, the bourgeois category includes both high and middle level managers, certainly it does not represent the big bourgeoisie, since it includes almost 8 per cent of the working population.

C O N C L U S I O N

In our view, the findings support the arguments that class can be successfully defined on the basis of Marxist categories, that modest class differences in political ideology do exist and are elaborated on the basis of class interest, and that differences in the work situation of the various classes serve to mediate the effect of class on political ideology. The attempt to consider the prior role of social background as a common cause of class and political attitudes was not very fruitful. The implication is that models of political socialization, at least insofar as they are based on parental education and occupation, do not provide adequate explanations of political ideology. The immediacy of the work environment clearly plays a more critical role. The results also cast doubt on the theoretical arguments of Hindess and Hirst and their collaborators that there is a ’necessary non-correspondence’ between the economic and the political. Still, the extent of the correspondence is quite small.

The more left-wing position of the working class reflects a number of aspects of the jobs they hold, including the distinctions between manual and non-manual work, between skilled and semi- and unskilled jobs, rates of pay, physical and mental tiredness, and trade union membership. In a certain sense, however, our original goal of locating the experiential basis of class differences in political

209 CLASS, W O R K , A N D P O L I T I C S

attitudes is retreating as we advance - for these findings pose a new set of questions about the roots of political ideology. If we began by asking what it is about class that leads to differences in political attitudes, we now may wonder what it is about trade union membership, and physical and mental tiredness that causes them to affect ideology. The items dealing with tiredness raise the most serious problem - since it is not clear to what extent they represent the objective conditions of jobs and to what extent they are measures of individual differences in perceptions of jobs. We have made progress, but the initial problem is only partly solved.

An interesting by-product of this analysis is the finding that there are aspects of jobs whose effects on political attitudes cannot be traced to underlying class differences. These effects are certainly comparable in magnitude to the class differences. These findings suggest that we inquire into the nature of political differences within social classes, rather than account for differences between classes. The question then is whether different factors might operate in different classes, or, to put it another way, whether there is interaction between class and the mediating variables.

N O T E S

I A related issue that could be raised in this analysis concerns whether the relations between political attitudes and family background and job conditions vary ucross classes. To put it another way, the question is whether there is an interaction between social class and either or both of family background and job conditions. One of the reviewers of this paper argued that an exploration of this issue is essential to our analysis of whether differences in family background and job conditions account for class differences in political attitudes. We do not agree. The two questions can be theoretically separated. However, in a note below, we shall indicate in a general way, the extent and direction of interaction effects.

2 See, for example, the acknowledgements in the preface to Schwartz (1974). 3 We should note that there have been a number of studies in Canada of the develop-

ment of ideology within particular social classes or fractions of classes: Lipset (1950), Macpherson (1953)~ and Sinclair (1976) consider the petty bourgeoisie; Grabb (1975) examines the lower middle class; and Bliss (i97.+), Clement (1975). and Naylor (1975) deal with the bourgeoisie.

4 For Canadian evidence, see Turrittin (1974). 5 Non-response correction factors (in the form of weights) are applied to the data to

assure the regional and urban-rural distributions of the sample match the population. Analysis of the correspondence between the demographic characteristics of the sample and the adult Canadian population indicated that the sample was representative. It follows that the sub-sample used in the analysis is also representative. For the sub-sample, the mean age is 37.5 years and the mean for sex is 1.30 where I is male and 2 is female.

6 For a comparative assessment of the political attitudes of the class categories defined by Carchedi, Poulantzas, and Wright, see Johnston and Ornstein (1979). Our use of Carchedi’s categories reflects a theoretical preference and the desire to limit the length of this paper. The earlier analysis of Poulantzas demonstrates that his combination of the new and old petty bourgeoisies obscures important differences between

210 WILLIAM J O H N S T O N A N D MICHAEL ORNSTEIN

them. Wright’s assignment of workers who control their own labour process to the ’contradictory location’ of semi-autonomous workers removes an important sector from the working class, even though they perform wage labour and neither own their means of production nor control capital investment. Some of the differences that are here attributed to the work situation would therefore appear as class differences between the working class and semi-autonomous workers, as defined by Wright.

7 The distinction between the new middle class and the bourgeoisie rests on the differences between ownership and non-ownership. Managers are included in the bourgeoisie because they exercise real rather than legal ownership, where real ownership is taken by Carchedi as the power to dispose of the means of production and labour power. While there are real ‘boundary problems’ in making an empirical separation between the classes, at the theoretical level the presence or absence of ownership is a categorical distinction.

a 1 Workers should have the right to refuse to work in conditions which they consider to be unsafe, until a government inspector assures them that conditions are safe. b I During a strike, management should be prohibited by law from hiring workers to take the place of strikers. c / Workers should have positions on the board of directors of the organization for which they work.

The scale of support for redistribution was created by summing the responses to the following statements:

a / There is too much difference between rich and poor in this country. b I The government should provide jobs for Canadians who want to work but cannot find a job. c I People with high incomes should pay a greater share of the total taxes than they do now.

For the two scales above, a card was presented to the respondents containing the following alternatives: strongly agree (scored 5 ) ; agree (4); neither agree nor disagree (3); disagree (2); strongly disagree (I); and no opinion (3). Respondents giving more than one ‘no opinion’ response were not scored on the scale. The mean inter- item correlation was .27 for support for redistribution, and .26 for support for labour -when computed for the entire cross section of the Canadian population (not just the subset of workers analyzed here). Support for social welfare was measured by combining the responses to eight items, as measured in the following larger question:

We would like to know how much effort you think government should put into a number of activities. Please choose the answer on this card which comes closest to your opinion about the effort that should be made in each area. Remember that put- ting more effort into one of these areas would require a shift of money from other areas or an increase in taxes. How much effort should be put into

a I health and medical care b / providing assistance to the unemployed

8 Support for labour was measured by summing the responses to the statements:

211 CLASS, W O R K , A N D P O L I T I C S

c / creating more jobs d / helping the poor e / building public housing f / education g / helping retired people h / workman's compensation The card contained the following alternative responses: much more effort (scored 5); more effort (4); about the same effort (3); less effort (2); much less effort (I). The mean correlation among the eight items is .27 for the entire population.

b / Supervisory, semi-professional, professional, technical, managerial occupations are all classified in the 'skilled' category. d / This measure is derived from summing the responses to a question that reads: 'I am going to read you some statements about jobs. Please tell me how true each one is of your job using one of these answers.' [The respondent is handed a card with four responses: very true (scored 4); somewhat true (3); not very true (2); and not at all true (I).] I / The people you work with are competent and helpful; z / The pay is good; 3/ The physical surroundings are pleasant; 4 / The work is interesting; 5 / The job security is good; 6 / The chances for getting ahead are good; 7 / There is an opportunity to develop your skills and abilities; 8 / Your supervisor is good at hidher job; 9 / The job does not interfere with your personal life and leisure activities; 10 / There is the recognition you deserve for your work; 11 / You can influence important decisions that are made by your supervisor; IZ / There is a great deal of freedom to decide how to do your work; 13 / The work you do makes a real contribution to other people. e / This variable is defined by combining the responses to two items that read as follows: I / Do you have any problems or difficulties concerning the hours you work, your work schedule or overtime? (a) yes (continue to next question); (b) no (scored zero, next question skipped). z / How serious are these problems for you? (a) very serious (scored 3); (b) fairly serious (scored 2); (c) not very serious (scored I).

serious multicolinearity problems in the regression. An inspection (employing analysis of variance and tasks for non-linear effects) of

the relations between the measures of political attitudes and all the continuous variables in the regression failed to turn up any important curvilinear relations. A comparison of the correlations between the three measures of political attitudes and the independent variables (except for class) within class categories indicates some interaction, but not enough to suggest that there are different relations between politics and family background and job conditions for each of the classes. The most important difference is that educational attainment of the respondent and family income and rate of pay have larger negative effects on all three dependent variables for the middle class, petty bourgeoisie, and bourgeoisie than for the working class. Otherwise the differences are small and unsystematic. Unfortunately, the detection of interactions involves comparisons based on the correlation coefficients for sub-

9 The measurement of the variables is self-explanatory, except for listed below.

10 The correlations among the independent variables are not large enough to cause

212 WILLIAM J O H N S T O N A N D MICHAEL O R N S T E I N

groups of the total sample, numbering much less than total sample size of 1787. The errors in these comparisons are therefore quite large.

R E F E R E N C E S

Alford, Robert 1963 Party and Society. Chicago: Rand McNally Anderson, Grace M. 1966 ’Voting behavior and the ethnic-religious variable. ’ Canadian Journal of Economics

Benney, Mark, and Phyllis Geiss 1950 ‘Social class and politics in Greenwich.’ British Journal of Sociology I: 310-27 Bliss, Michael 1974 A Living Profit. Toronto: McClelland and Stewart Carchedi, Guglielmo i975a ‘Reproduction of social classes at the level of productive relations.’ Economy and

i975b ‘The economic identification of the state employees.‘ Social Praxis 3(1): 93-120 1 9 7 5 ~ ‘On the economic identification of the new middle class.’ Economy and Society

4(2): 1-85 Centers, Richard 1949 The Psychology of Social Classes. Princeton, N. J. : Princeton University Press Clarke, Harold, Lawrence Leduc, Brigit Jenson, and Jon Pammett 1979 Political Choice in Canada. Toronto: McGraw-Hill Ryerson Clement, Wallace 1975 The Canadian Corporate Elite: An Analysis of Economic Power. Toronto:

Crossland, C.A.R. 1956 The Future of Socialism. London: Jonathan Cape Cutler, Anthony, Barry Hindness, Paul Hirst, and Althar Hussain

and Political Science 32(1): 27-37

Society 4(4): 361-417

McClelland and Stewart

1977 Marx‘s Capital and Capitalism Today (2 vols). London: Routledge and Kegan Paul

Form, William H., and Joan Rytina 1969 ‘Ideological beliefs OR the distribution of power in the United States. ‘ American

Free, Lloyd A. and Hadley Cantril 1967 The Political Beliefs of Americans: A Study of Public Opinion. New Brunswick,

Goldthorpe, John, David Lockwood, Frank Bechhofer, and Jennifer Platt 1968a The Affluent Worker: Industrial Attitudes and Behaviour. Cambridge: Cambridge

University Press 1968b The Affluent Worker: Political Attitudes and Behaviour. Cambridge: Cambridge

University Press 1969 The Affluent Worker in the Class Structure. Cambridge: Cambridge University

Press. Grabb, E.G. 1975 ‘Canada‘s lower middle class.’ Canadian Journal of Sociology 1(3): 205-312

Sociological Review 34: 19-31

N. J. : Rutgers University Press

213 C L A S S , WORK, A N D P O L I T I C S

Hamilton, Richard F. 1967 Affluence and the French Worker. Princeton: Princeton University Press 1972 Class and Politics in the United States. New York: Wiley Hirst, Paul Q. 1977 ’Economic classes and politics.’ Pp. 125-54 in Alan Hunt, Class and Class Structure.

London: Lawrence and Wishart Johnston, William, and Michael D. Ornstein 1979 ’Social class and political ideology in Canada.’ Paper presented at the Annual

Lipset, Seymour M. 1971 Agrarian Socialism. Berkeley: University of California Press Lipsitz, Lewis 1964 ’Work life and political attitudes: a study of manual workers.’ American Political

Macpherson, C.B. 1953 Democracy in Alberta. Toronto: University of Toronto Press Mann, Michael and R.M. Blackburn 1975 ’Ideology in the non-skilled working class.’ Pp. 131-160 in M. Bulmer, Working

Meisel, John 1975 Working Papers on Canadian Politics (2nd. ed.). Montreal: McGill-Queen’s

Naylor, Tom 1975 The History of Canadian Business (2 vols). Toronto: Lorimer Ornstein, Michael D., H. Michael Stevenson, and H. Paul M. Williams 1980 ’Region, class and political culture in Canada.’ Canadian Journal of Political Science

Pineo, Peter C., John Porter, and Hugh A. McRoberts 1977 ‘The 1971 census and the socioeconomic classification of occupations.‘ Canadian

Review of Sociology and Anthropology 14(1): 91-102 Poulantzas, Nicos 1975 Classes in Contemporary Society. London: New Left Books Rinehart, James and Ismael Okraku 1974 ‘A study of class consciousness.‘ Canadian Review of Sociology and Anthropology

Schwartz, Mildred 1974 Politics and Territory: The Sociology and Regional Persistence in Canada.

Sinclair, Peter 1976 ‘Class structure and populist protest: the case of Western Canada.’ Canadian Journal

Stevenson, Paul 1977 ’Class and left-wing radicalism. ’ Canadian Review of Sociology and Anthro-

Turrittin, Anton H. 1974 ‘Social mobility in Canada: A comparison of three provincial studies and some

Meeting of the F e r i c a n Sociological Association, Boston

Science Review 58: 951-62

Class Images of Society. London: Routledge and Kegan Paul

University Press

XIII, ~ (June) : 227-71

l l (3): 197-213

Montreal: McGill-Queen’s University Press

of Sociology ~(i): 1-17

pology 14 (3): 269-84


methodological questions. ’ Canadian Review of Sociology and Anthropology Special Issue: 163-86

Westergaard, John 1965 ’The withering away of class.’ In P. Anderson, Towards Socialism. London: Fontana Williams, A. Paul M., David Bates, Michael D. Ornstein, and H. Michael Stevenson 1979 ‘Class, ideology and partisanship in Canada.’ Paper presented at the annual meeting

Wright, Erik Olin 1976 ‘Class boundaries in advanced capitalist societies.’ New Left Review 98: 3-41

of the Canadian Political Science Association, Saskatoon

class, work, and politics

Documents