race and gender as bases of class fractions and their effects on earnings
TRANSCRIPT
Race and gender as bases of class fractions and their effects on earnings
PETER S. LI University of Saskatchewan*
Maints auteurs ont affirm6 que la race et le sexe representent des bases de fractions de classe en ce sen8 que leur construction sociale permet B des segments d’une classe de se constituer B partir de caracteristiques likes B la race et au sexe. Cependant, les btudes ernpiriques antbrieures tendent B considher seulernent la race et le sexe comme des facteurs de fractionnernent au sein de la classe ouvrihre seulement. A l’aide des donnees du recensement de 1986, cette communication analyse lea effets interactifs de la race et du sexe sur le revenu considkre comme facteur de fractionnement des classes. Les recherche8 de l’auteur indiquent que les effets produits par le groupement par race et par sexe sont inegaux en ce qui trait au revenu, les hommes gagnant davantage que les femmes tout en btant davantage ‘fractionn6s’ par le revenu que les femmes. Lorsqu’on fait abstraction des diffbrences de revenu entre classes, la race et le sexe demeurent des facteurs explicatifs importants du niveau de revenu, bien que leurs effets soient plus marques chez les cadres, les professionnels et les ouvriers. L’Ccart entre les sexes est plus prononc6, mais la race demeure importante comme facteur de fractionnement au niveau des revenus des homrnes, rnkrne aprbs correction pour tenir cornpte d’autres variables. L’auteur affirme que l’assimilation des relations de sexe et de race aux relations entre classes est insatisfaisante et qu’il existe de drieuses considerations theoriques pour considbrer que les causes des in6galitCs entre les sexes et les races se trouvent non seulement a l’intbrieur du domaine de la production, mais egalement a l’exterieur de celui-ci.
The literature has argued that race and gender are bases of class fractions in that their social construction enables segments of a class to be constituted on * This analysis is based on data from the 1986 Census of Canada, Public Use Microdata File
on Individuals, supplied by Statistics Canada. The data tapes were made available to the author through the University Library of the University of Saskatchewan as a member of a consortium of the Canadian Association of Research Libraries. The author would like to thank Andrew Hubbertz for initially obtaining the tapes, and Dawn Currie, Harley Dick- ineon, Teny Wothempoon, Vic Satzewich and the anonymous reviewers for their insight- hl comments. The author is solely responsible for the uae and interpretation of the cenaus data. A version of this paper waa presented at the 1991 annual conference of the Social Re- search Unit, Department of Scoiology, University of Saskatchewan, Saskatoon. This man- uscript was received in April, 1991 and accepted in May, 1992.
Canad. Rev. Soc. & Anth. I Rev. canad. SOC. & Anth. 29(4) 1992
489 EFFECTS O F RACE AND GENDER ON EARNINGS
racialized and gender-linked characteristics. However, previous empirical works tend to consider race or gender as fractionalizing the working class only. Using the 1986 Census data, this paper analyses the interactive effects of race and gender on earnings as grounds of fractionalizing classes. The findings indicate that race and gender groupings produce unequal effects on earnings: males have an income advantage over females, but race fractionalizes the earnings of men t o a greater extent than for women. When inter-class differences in earnings are removed, race and gender remain important in explaining earnings, although their effects are stronger among the managerial, the professsional, and the working class. The gender gap is more pronounced, but race remains important in fractionalizing the earnings of men even after controlling for other variables. The paper argues that it is insufficient to subsume gender and race relations under class relations, and that there are compelling theoretical grounds to consider gender and race inequality as having roots both within and outside the sphere of production.
Few studies in Canada have attempted to examine the joint effects of race and gender on income, despite many empirical findings which show that women earn less than men (Armstrong and Armstrong, 1984; Fox and Fox, 1983; 1986; Fillmore, 19901, and some studies which suggest that racial origins make a difference in income (Satzewich and Li, 1987; Li, 1988a). The purpose of this paper is to use the Microdata File on Individuals from the 1986 Census of Canada to analyse the joint effects of race and gender on in- come, in light of recent theoretical attempts to use the Mamian notion of ‘class fraction’ to incorporate race and gender into the concept of social class. 1
RACE AND GENDER AS BASES OF CLASS FRACTIONS
Karl Marx (1963) employs the notion of ‘fractions’ in analysing the plural- ity of the French bourgeoisie in The Eighteenth Brumaire of Louis Bona- parte when he refers to the industrial bourgeoisie, the financial aristocracy and the large landowners, and to their differences in political alliance and ideology. Thus, class fractions refer to groupings within the boundary of a class, which although sharing the same class relationships with other classes in terms of their relations to production, may differ in ideological commit- ments and political actions due to their locations within the class structure and their social and economic characteristics. Class fraction as a theoreti- cal concept is perhaps most systematically developed by Nicos Poulantzas (1978a: 77-85, 229-521, who interprets it as a ‘substratum of eventual so- cial forces’ (1978a: 841, and argues that its existence, like that of class, is de- termined by the relations of production and the place in the process of production, and reflected in political and ideological structures (Poulantzas 1978a: 78-79). Theorizing about petty bourgeois fractions, Poulantzas (197813: 316) argues that they ‘must be considered as defined by certain ten- dencies marked out by their class determinations and the current transfor- mations, and not by empirical and rigid boundaries according to some
490 PETERS. LI
"statistical" classification'. Poulantzas (197813: 317-9) also suggests that frac- tions of the petty bourgeoisie share differential 'distributive affinity' with the working class in terms of career and promotion, earnings, and inter- generational mobility. The writings of Mam and Poulantzas provide the an- alytical framework for studying class fractions, although the specific case where class fractions were analysed was the bourgeoisie for Mam, and the new petty bourgeoisie for Poulantzas.
Recent works have broadened the concept of 'class fractions' to struc- tural divisions within a class, based on social relationships f occupational groups, gender and race (Freedman, 1975; Barrera, 1979). For example, Barrera defines what he calls 'an ascriptive class segment' as 'a portion of a class which is set off from the rest of the class by some readily identifia- ble and relatively stable characteristics of the persons assigned to that seg- ment, such as race, ethnicity, or sex, where the relationship of the members to the means and process of production is affected by that demarcation' (Barrera, 1979: 212). The notion of 'class fractions' is also adopted in recent work by Anne Phizacklea and Robert Miles, who refer to it as 'an objective position within a class boundary which is, in turn, determined by both economic and politico-ideological relations' (1980: 6). They argue, in essence, that sexism is an ideological force which sets women apart, in the same way that racism does to racialized groups, to render them a distinct position in economic, political and ideological relations; hence women and racialized groups constitute class fractions (Phizacklea and Miles, 1980; Miles, 1982).
Despite the theoretical advances to incorporate gender and race as bases of class fractions, much empirical work in this area tends to focus on frac- tions of the working class only. This is most evident in studies of race and class in which race is often seen as fractionalizing the working class in two ways (Cox, 1948; Miles, 1982; Castles and Kosack, 1985). First, race divides the working class into white and nonwhite workers, and racism benefits capitalists in weakening working class solidarity. Second, race splits the labour market for differential wages, thus enabling.employers to economize on the cost of production, and to obfuscate class contradictions as racial an- tagonism (Bonacich, 1972; 1975; 1976).
In Canada, there is strong historical evidence to show that some groups were indeed racialized to produce the effects noted above (Muszynski, 1988; Creese, 1988-89; Makabe, 1981; Li, 1988b; Bolaria and Li, 19881, but the pre- dominance of racial groups as the proletariat in the specific historical peri- ods being examined makes it redundant to analyse racialized class fractions beyond the working class. Several studies of Western Europe in the post World War I1 era also use the analytical framework of class fractions as con- stituted by race, gender and immigrant status, but the analysis is not ex- tended to the capitalists or the petty bourgoisie. For example, in their classic study of immigrant workers and class structure in Western Europe, Castles and Kosack (1985) focus their analysis on the division of immigrant work- ers and indigenous workers within the same working class. Studies of Eng- land by Phizacklea and Miles (19801, and Miles (1982) also do not extend beyond the gender fractions and the racialized fractions of migrant and Eng-
B
491 EFFECTS OF RACE AND GENDER ON EARNINGS
lish workers. While these works have demonstrated the utility of the con- cept of ‘class fractions’ in understanding how race divides up the working class along economic, political and ideological grounds, the rationale of racialization seems to apply best to the working class. The basic logic of working class fractionalization based on race is that such cleavages are re- sults of recruitment of certain forms of labour much needed by advanced capitalism, and that the resultant configuration of working class fractions is produced by racial antagonism among labour and reinforced by state policy (Castles and Kosack, 1985; Phizacklea and Miles, 1980; Li, 1988b).
The literature is less explicit about using gender as a basis of class frac- tion, although many studies in Canada have presented strong grounds for doing so (Muszynski, 1988; Creese, 1988-89; Lowe, 1980; Cuneo, 1985; Car- roll, 1987). Several studies in Canada, dealing with the growth bf women in the paid labour force and its effects on jobs and workers, have produced im- portant findings which support the notion of gender fractionalizing the work force. For example, Graham S. Lowe (1980) shows that the bureaucratiza- tion of corporate capitalism creates the demand for routinized office jobs which become gender-linked in terms of job characteristics, and feminized in terms of their occupants. In this way, office jobs are being segmented into the high-paying managerial positions and low-paying ones confined to routine tasks; gender becomes the convenient basis for segregating the oc- cupants of these jobs. More recently, Carl J. Cuneo (1985) argues that despite the transformation of the class sturcture and the increased partici- pation of women in the paid labour force between 1931 and 1984, women have become more proletarianized than men. This finding is confirmed by William K. Carroll (1987), who further argues that the differential position of men and women in the class structure is linked to gender segregation in occupations. Taken together, these findings clearly suggest that gender frac- tionalizes class in two important senses. First, as the occupational structure changes under corporate capitalism, gender provides the ground for segre- gating occupations into men’s jobs and women’s jobs; sex-typing is in terms of the tasks required of jobs and the occupants recruited for them. Second, women have become more proletarianized than men in the labour force, and as a result, have produced what Poulantzas calls an ‘objectively proletarian polarization’.
Thus far, the literature has provided strong theoretical grounds for asserting race and gender as two separate bases for fractionalizing the class structure, but empirical works in this regard have tended to focus on the working class only. However, the interactive effects of race and gender as class fractions have not been systematically examined, although several case studies have highlighted the importance of considering race and gender jointly. For example, the historical account of salmon cannery workers in British Columbia by Alicja Muszynski (19881, and that of the working class in Vancouver by Gillian Creese (1988; 1988-89) show that both race and gender were used as bases in segmenting the working class; the segmenta- tion was apparent not only in relationships of production and differential wages, but also permeated ideological and political relationships.
492 PETERS. Ll
EFFECTS OF RACE AND GENDER ON EARNINGS
Treating race and gender as bases of class fractions provides sound theoreti- cal grounds for interpreting income inequality, especially within the work- ing class. Since race and gender have been the social bases for recruiting labour, and since the abundant supply of these forms of labour has enabled the restructuring of the work force in terms of job types and the selection of job incumbents, it follows that race and gender would be relevant sources of income inequality. Indeed, the findings thus far on gender inequality have been unequivocal: women tend to be ghettoized into low-paying jobs and part-time employment (Armstrong and Armstrong, 1984; Ornstein, 19831, and women’s earnings remain lower than men’s despite a narrowing of the gendes earning gaps in more recent decades (Fox and Fox, 1986; Fillmore, 1990). Similarily, it has been shown that racial origin provides a useful basis for explaining unequal income levels of immigrants in the labour force (Satzewich and Li, 19871, and that ethnic and racial origin remains signifi- cant in differentiating income levels of working Canadians, although the ef- fect of origin on income is weak when variations in other factors have been taken into account (Li, 1988a).
Despite the rationale and the evidence, the literature has not dealt with how race and gender operate interactively in fractionalizing class, especially beyond the working class. The focus of this paper is to extend the analysis of race and gender as bases of class fractions to consider their interactive effects on earnings in various classes. The purpose is not to develop the empirical bases for measuring class fractions, but to apply the notion of class fractions as developed in the literature to assess what empirical con- sequences follow from class fractionalization by race and gender. The re- view of the literature suggests two logical inquiries. First, if class fractions are rationally manifested in the working class, as many previous empirical works have shown, does it mean that their effects are minimal in other classes? Second, if either race or gender constitutes a basis of class fraction, does it mean that they in fact operate jointly as grounds of class fractional- ization? Using income inequality as a consequence of class and class frac- tions, the following empirical questions are developed to address the above theoretical questions: V Is there an empirical basis for considering the in- teractive effects of race and gender in all classes? 2/ What are the con- sequences of race and gender as bases of class fractions on earnings within each specific class?
SOURCE OF DATA AND MEASUREMENT OF VARIABLES
The data in this analysis are taken from the 1986 Census of Canada, Micro- data File on Individuals. The data file has 600,434 cases, based on a 2 per cent probability sample of the Canadian population; the target population of the sample excludes i n s t i t u p a l residents and residents of incompletely enumerated Indian reserves. Only those individuals 15 years of age and over and those who worked in 1985 are included in this analysis;‘ the re- sulting sample has 277,246 cases.
493 EFFECTS OF RACE AND GENDER ON EARNINGS
The dependent variable is earnings from wages and salaries, and self-em- ployment. This variable differs from the total income of individuals in that the latter also includes investment income, pension, government transfer payments and other income. The ‘total income’ is partly determined by the amount of assets and investment, and partly by transfer payments, although empirically ‘earnings from employment and self-employment’ and ‘total in- come’ are virtually the same variable in that the correlation between them is 0.95, and wages and salaries made up 73.2 per cent of the total income, and self-employment income another 5.5 per cent. Since the present analy- sis is concerned mainly with how class fractions and class relations impact upon earnings which arise from labour force activities, the more narrow measurement of earnings from employment and self-employment is used. Statistics Canada ap lies upper and lower limits to income variables to en- sure confidentiality. The adjustments produce little effect on the overall earning functions, although they tend to underestimate the income of the highest earners, and consequently underestimate the disparity between groups in the two extremes (Li, 1988a: 68). The present analysis uses the actual income to facilitate direct interpretation, rather than using a loga- rithm transformation of income as adopted in some studies in which com- parisons of several countries make it necessary to standardize the income functions (Rosenfeld and Kalleberg, 1990).
Among the independent variables is ‘social class’ which is constructed along Wright’s formulation based on owership and control over money capi- tal, physical capital, and labour (Wright; 1977; 1978; 1979).7 Although cen- sus data do not include direct measurements of social class, several studies have been able to infer class categories from census data on self-employment and occupation (Cuneo, 1985; Carroll, 1987; Li, 1988a). Using Wright’s (1978: 1370) criteria of ownership of the means of production and control over others’ labour, those in the 1986 Census who were self-employed with paid help are classified as ‘employers,’ and hose self-employed without paid help are categorized as ‘petty bourgeoisie’! This is because both employers and petty bourgeoisie own and control money capital and physical capital of production, but unlike the petty bourgeoisie, employers also control the labour of others. Those ‘employed’ are subdivided into three groups, accord- ing to how much inferred autonomy and control they have over capital and labour. Employees in managerial and administrative positions occupy a con- tradictory class location because in acting on employers’ behalf, they have some control over money capital, and over physical capital and labour. Em- ployed ‘professionals’ form another contradictory class position in that, un- like other workers, they have certain autonomy over their work, and therefore some control over their l a b ~ u r . ~ Finally, employees not in manage- rial and professional occupations are classified as ‘workers,’ on the basis of their lack of control over capital and labour.
‘Gender’ and ‘race’ are dichotomized as ‘male and female’ and ‘white and non-white’. Individuals are ‘non-white’ if they are members of the ‘visible minority’ or an aboriginal ethnic group; otherwise, they are considered ‘white’. For the 1986 Census, Statistics Canada classifies a respondent as a
I!?
484 PETER S. LI
member of the visible minority if that person belongs to one of the 11 groups: Black, Indo-Pakistani, Chinese, Korean, Japanese, South East Asian, Fil- ipino, Other Pacific Islanders, West Asian and Arab, Latin Ameri and multiple visible minority group (Statistics Canada, 1990: 71-72)!‘Gsing this classification, there were 22,011 non-white persons in he 2 per cent sample who participated in the 1985 Canadian labour force.
Other independent varibles are: sector of employment or self-employ- ment (16 categories), nature of work (full-time or part-time), nativity (na- tive-born or foreign-born), age (actual year), number of weeks worked (1-49),12 and years of schooling. The ‘years of schooling’ is constructed from several variables, including ‘highest grade,’ ‘yeara of university’ and ‘years of other non-university education’. The years of schooling for individuals with post-secondary education is the sum of the number of years of univer- sity or non-university education, whichever is higher, and 12 years of elementary and secondary grades; those with only ‘Grades 5-8’ education are treated as having an average of 6.5 years of schooling; and those with ‘Less than Grade 5’ are treated aa having an average of three years of school- ing.
Multiple Classification Analysis (Andrews et al., 1976) is used to analyse the effects of class fractions and class categories on earnings. The statisti- cal procedure treats the dependent variable as a linear combination of a set of variables, both categorical and interval, the effects of which can be assessed by the regression coefficients. For categorical independent varia- bles, Multiple Classification Analysis produces a regression coefficient for each category as a deviation from the grand mean. The gross deviations measure the effects when variations in other independent variables have not been adjusted; the net deviations are effects when inter-group variations in other independent variables have been taken into account. For interval variables, Multiple Classification Analysis produces the unstandardized re- gression coefficients.
d
DATA AND DISCUSSION
The effecta of race, gender and class on earnings are given in Table I. Column 2 shows the gross effect of each independent variable and of the interactive variable of race and gender. White Canadians’ average earnings are about $200 above the mean income of $18,155, whereas non-white Canadians earn about $2,200 below the national average. Thus, an income gap of about $2,400 can be attributed to race. However, in terms of its ability to account for income variation, albeit significant, race a variable only explains a small portion of the total variance (eta = 0.04). Gender as a variable pro- duces an average income gap of about $10,000 between men and women, with men earning over $4,000 above the mean, and omen $6,600 below. Gender as a variable alone accounts for 9 per cent (eta of the variations in earnings. When race and gender are considered jointly as an interactive vari- able, the data reveal that white males have the greatest income advantage, being $4,610 above the mean, while non-white males are second, with earn- ings about $1,000 above the mean. Women, whether white or non-white,
9 3
Y
495 EFFECTS OF RACE AND GENDER ON EARNINGS
TABLE 1 GROSS AND NET EFFECTS OF RACE AND GENDER ON ANNUAL EARNINGS FOR ALL CLASSES
Gms.9 Effect Net Effect on Annual on Annual
Number Earnings ($) Earnings ($) Eta Beta
Race White Non-white
Gender Male Female
Race and Gender White male Non-white male White female Non-white female
Social Claas Employers Managers Professionals Petty Bourgeoisie Workers
Sector Agriculture Other primary industries
Manufacturing Construction Transportation & storage
Communication & utilities
Wholesale trade Retail trade Finance, insurance & real estate Business services Federal Government services
Other Government services
Educational services Health and Social services
Accommodation and food services Other services
Nature of Work Full-time Part-time
243299 198.81 22011 -2197.55
0.04
149841 4321.32 115469 -5607.67
137884 4610.16 11957 990.57 105416 -5671.29 10054 -5989.12
11730 10026.53 23515 13549.79 39563 4205.43 13221 -3311.69 177291 -3151.81
10912 -8304.03
7627 5317.20 45004 2864.21 15749 -298.30
' 11761 4579.52
8540 0573.60 11903 2149.13 33599
14508 12196
8682
11522 17815
21397
16837 17258
208612 56698
-5772.05
2891.29 3339.33
5874.88
3076.63 4559.16
595.83
-10087.11 -7308.47
3127.84 -11508.44
0.30
3234.15 483.48
-3806.77 -4176.84
0.30 0.21
5458.11 7190.21 1497.78 -3721.69 -1371.41
0.33 0.17
-6893.98
6765.33 1737.13 -152.11
2723.59
4890.14 -178.82
-2996.92
1247.96 723.35
1938.37
1065.83 1216.01
894.55
-3020.20 -3562.91
0.31 0.16
1244.29
0.36 0.14 -4578.19
496 PETERS. LI
____
TABLE I contd.
Gross Efect Net Effect on Annual on Annual
Number Earnings (8 Earnings ($) Eta Beta
~~
Nativity
Foreign-born 49199 1777.20 174.12 Native-born 216111 -404.59 -39.64
0.06 0.00
Years of echooling Age Number of weeks worked
0.24 1040.16 0.24 200.81 0.49 356.77
Grand Mean 18154.93 18154.93 Multiple R 0.67
SOURCE: Compiled from the 1986 Censua of Canada, Public Use Microdata File on individuals, based on a 2% probability sample of the population. The figurea in the table do not include per- sons under 15 years of age, persons who did not work, and inmates
Multiple R2 0.44
have an earning level more than $6,000 below the mean. Non-white women have by far the worst income level, although their income gap relative to white women is relatively small. These figures clearly indicate that the ef- fect of race on earnings is dependent upon gender: race produces a greater effect on earnings among men than women. Conversely, the gender earning gap can be further differentiated by taking into account the effects of race, especially among men. Although race and gender j intly do not increase the proportion of explained variance in earnings (etJ=O.O9), they show how combinations of racial and gender groupings produce unequal effects on earnings.
Social class explains about 11 per cent (0.332) of the variation in earnings, with different class positions showing an advantage or a disadvantage. Em- ployers and managers enjoyed the highest income level, being over $10,000 above the national average. In contrast, the petty bourgeoisie and workers had the greatest disadvantage in earning, showing more than $3,000 below the average. In this respect, the petty bourgeoisie shares what Poulantzas (197813: 319) calls ‘form8 of distributive affinity’ with the working class.
Various sectors show differences in earnings, with ‘communications and utilities’ and ‘federal government services’ having the greatest advantage, and ‘accommodation and food services’ the greatest disadvantage. Those in full-time work earned about $14,600 more than those engaged in part-time work, and those who worked more weeks earned more (P = 0.49). However, those native-born earned about $2,100 less than those foreign-born, al- though like race, the ability of nativity to account for earning variations is relatively small. Finally, the effects of both ‘age’ and ‘schooling’ on earnings are positive and moderately strong (r = 0.24).
TABL
E 11
C
RO
SS E
AR
NIN
G DISPARITY O
F R
AC
WC
END
ER FRACTIONS BY
SO
CIA
L C
LASS
Gro
w E
anU
ngs as 5%
of
Gra
ss 'W
Me
Mal
e' in
com
e
Gro
ss
of R
acel
Gen
der
Gro
ss E
ffm
t
Mea
n E
arni
ngs
Mea
n E
umin
gs
Non
-whi
te
Whi
te
Non
-whi
te
on E
arni
ngs
soci
al C
hs
(a
ll G
roup
s)
Whi
te Male
Mal
e F
emal
e F
emal
e (E
ta)
~ ~
Em
ploy
ers
28,181
31,061
93.30
53.97
60.33
0.19
Mal
lage
I3
31,705
37,401
80.94
57.25
56.56
0.36
Prof
essi
onal
s 22,360
28,937
95.77
61.19
58.80
0.33
Pett
y bo
urge
oisie
14,843
16,525
l11.59
56.8
4 64.98
0.16
Wor
kers
15,003
19,203
79.88
53.14
51.97
0.34
All
clas
ses
18.155
23,271
79.80
52.54
48.57
0.30
SOU
RC
E: C
ompi
led
from
1986 Census of
Can
ada,
Pub
lic U
se M
icro
data
File
on
Indi
vidu
als, based
on a
2% p
roba
bilit
y sa
mpl
e. T
he figures
in th
e ta
ble
do n
ot in
clud
e persona u
nder
15 y
eara
of age, p
erso
ns w
ho d
id n
ot w
ork,
and in
mat
es
498 PETER S. LI
Some of the earning gaps among the class categories, and among the ra- cial and gender fractions can no doubt be attributed to variations in other variables, such as differences in full-time and part-time work status, school- ing, age and other factors. Column 4 of Table I shows the net effects of race and gender as an interactive variable. The earning disparities among the gender and racial fractions are smaller, when inter-group variations in other variations have been taken into account. For example, the income gap be- tween white males and white females is reduced to about $7,000, suggest- ing that some of the original disparity is attributable to gender differences in employment patterns, and individual factors such as age and schooling differences. The literature has shown that the substantial income gap be- tween men and women is related to the intermittent work career of women and their tendency to be employed part-time (Goyder, 1981; Ornstein, 1983). The relative income positions of the racial and gender fractions persist, however, despite controlling for other variables. The effect of race and gender fractions as a variable is given by the beta coefficient (0.211, which indicates a strong effect on earnings independent of the influence of other factors in the equation.
When variations in other variables are taken into account, the relative rankings of various classes in earnings also hold, despite a narrowing of the income gaps. The beta coefficient (0.17) indicates that the net impact of so- cial class on earnings is smaller than that of racial and gender fractions (0.21). These findings show that social class as well as race and gender are important in accounting for earning differences; and that there are substan- tial empirical grounds for considering race and gender interactively as pro- ducing social fractions in arnings. The model in column 4 statistically accounts for 44 per cent (R of the variations in income.
Although Table I establishes the importance of class categories and ra- cial and gender groupings on earnings, it is insufficient to show whether class fractions operate uniformly or differently across different class cate- gories. Table 11 provides the data for considering the effects of class fractions separately on each class. The income levels of non-white males, white females, and non-white females are calculated as a percentage of the gross earnings of white males. Thus, the smaller the percentage below 100, the greater is the income gap.
The differential effects of ‘race and gender’ on earnings for each class can be asssessed from the magnitude of eta. Table 11 shows that racial and gender fractions have stronger effects on earnings among the managerial class (0.36), the professional class (0.33), and the working class (0.34). By con- trast, their effects on those who have direct control over money capital and physical capital - the employers (0.19) and petty bourgeoisie (0.16) - tend to be weaker. This finding lends credence to the claim that gender and race fractionalize the working class, if the latter is defined broadly to include all wage-earners. Why class fractions based on gender and race tend to be weaker among employers and the petty bourgeoisie in producing a differ- ence in earnings is probably related to the fact that members of these classes are more immune from racism and sexism in that their self-employment in-
B
TABL
E 11
1 N
ET
EFF
ECT
OF
RA
CE
GE
ND
ER
FR
AC
TIO
NS ON A
NN
UA
L E
AR
NIN
GS BY S
OC
IAL
CL
ASS
Empr
OYe
rs
Petty
Bou
rgeo
isie
M
anag
ers
Prof
ess Wd
Wor
kers
Net
N
et
Net
N
et
Net
N
E
ffec
t N
E
ffec
t N
E
ffec
t N
E
ffec
t N
E
ffec
t B
eta
Bet
a B
eta
Bet
a B
eta
Race
and
Gen
der
White
mal
e 8762
Non
-whi
te m
ale
675
White
fem
ale
2086
Non
-whi
te fe
mal
e 207
Sect
or
Agr
icul
ture
1371
Oth
er p
rim
ary
indu
stri
es
334
Man
ufa
ctu
rin
g
823
Con
stru
ctio
n 1546
Tra
nspo
rtat
ion
&
&ra
ge
297
Com
mun
icat
ion
&
Util
ities
62
Who
lesa
le tr
ade
625
Ret
ail t
rade
2418
Fina
nce,
insu
ranc
e C
re
ales
tate
320
&si
neE
sser
vice
e 1201
Federal
Gov
ernm
ent
serv
ices
0
Oth
er G
over
nmen
t se
rvic
es
0 E
duca
tiona
l ser
vice
s 59
Health a
nd S
ocia
l se
rvic
es
839
2467
-2742
-8593
-8883
-10180
1414
935
-2702
154
-2936
2473
-6520
6532
ll678
-4605
32611
9383
537
3092
209
0.15
3772
450
535
1366
699
115
392
1719
296
1206
0 0 205
394
1697
1289
-5028
-5096
-4251
1434
611
1062
5101
769
808
-2182
4810
5585
-1261
22292
14620
95 1
7397
547
0.15
450
507
3552
737
750
950
1583
2576
2899
1596
1759
1870
1210
990
3738
-2055
-6583
-7306
-8265
11440
3355
293
2142
5961
1333
-4474
1783
135
-301
-2621
964
-2060
15037
1652
20888
1976
0.23
169
770
2971
331
333
1193
298
762
541
2677
1199
2444
11515
l234-
3335
22
69
-2300
-2962
-3677
7332
914
453
4281
5190
-913
-1952
1320
-540
2223
-42
119
-888
90082 2832
8142
246
71952 -3232
7115 -3453
0.17
0.23
5150 -5652
5566
5273
37123
1605
11769
278
9682 2983
6220 5141
9005 -243
26l24 -2407
10453
1312
5516 -1097
5724 2427
7208 2203
4826 -1332
6830
-1491
TABLE
III m
ad
EmO
love
rs
Pet&
Bou
rpeo
isie
M
anaa
ers
Pmfe
SsW
nals
W
orke
rs
Net
N
et
N
Eff
ect
N
Eff
xt
N
Bet
a B
eta
Net
E
ffec
t N
et
Net
N
E
ffec
t N
ef
fect
B
eta
Beta
Beta
Sect
or
Acc
omm
odat
ion
and
food
ser
vice
s 89
7 O
ther
ser
vice
s 93
8
Naia
u-e
of W
ork
Full-
time
1061
6 Pa
rt-t
ime
1114
N&
ity
Fore
ign-
born
29
80
Nat
ive bo m
87
50
Years
ofse
hool
iw
- A
ge
Num
ber
of weeks
wor
ked
Gra
nd Mean
Mul
tiple
R
Mul
tide
R2
-731
4 -5
752
0.38
631
-601
1 0.
07
-2223
6 7
0 0.
01
1224
12
6 388
2818
2 0.
54
0.29
314
-194
10
10
1758
-1
837
1076
0.
26
1007
9 1338
2215
5 31
42
-429
2 13
60
0.01
1054
5 -6
7 19
250
2676
26
6 42
65
0.01
292
-30
227
14843
0.39
0.
15
-619
2 4548
558
-908
9
-62
279
225
1782
0.
18
3040
3 91
50
0.11
3202
23
7525
0.
01
2006
41
1 58
9
3170
5 0.
63
0.40
-221
7 14
391
-296
0 -3
078
1170
4 -3
172
0.11
0.
20
1901
13
5359
1236
-631
5 4
93
2 -
3989
0.
21
0.17
64
1455
38
-23
-271
31
753
107
0.01
0.
00
1686
333
454
587
160
335
2236
0 15
003
0.73
0.
69
0.53
0.48
%X
JRC
E:
Com
pile
d from t
he 1
986
Cen
sus of Canada, P
ublic
Use M
icro
data
File
on
indi
vidu
als,
bas
ed o
n a
2% pr
obab
ility
sam
ple of
the
popu
la-
tion.
The
figu
rea
in th
e ta
ble
do n
ot in
clud
e persons
unde
r l5
year
s of age, persons
who
did
not
wor
k in
198
5, a
nd in
mat
es
601 EFFECTS OF RACE AND GENDER ON EARNINGS
come does not depend on wage competition with other workers, nor on so- cial values attached to their work by employers.
Despite a weaker effect of race and gender among employers and among the petty bourgeoisie, there are substantial income gaps among gender and racial groups in these two classes. White females have the lowest income levels; their income is 54 per cent and 57 per cent of that of white males for the employer class and the petty bourgeoisie respectively. The earnings of white females are only marginally higher than those of non-white women among the managerial class, the professional class, and the working class, but substantially below that of white males. It would appear that in these three classes, white and non-white women experience similar low wages for their work, and that non-white women seem to have experienced a major jeopardy in earnings for being women, and only a marginal jeopardy for being non-white. The low earnings of women, white and non-white, in these three classes can be partly attributed to job ghettoization and discrimina- tion in pay, as the literature has well documented (Armstrong and Arm- strong, 1984; Fox and Fox, 1986; Fillmore, 1990). But for employers and the petty bourgeoisie, white and non-white women experienced similar earning disparity with men, as among wage-earners. This would suggest that owning and controlling capital and having control over labour do not reduce gender disparities in earnings. Such disparities cannot be attributed to unequal values being attached to male and female work by employers, since women who are self-employed or employers of others receive their earnings from self-employment. It is possible that part of such gender-based income dis- parity can be due to unequal business opportunity and differential pa- tronage of clientele for women employers and self-employera. But a more systematic explanation of persistent gender differences in earnings in all classes would have to include factors outside the sphere of production, such aa how domestic burden and patriarchal domination may adversely affect the earnings of women.
Non-white males, although having a higher income level than white and non-white women, earn less than white males in all classes, except among the petty bourgeoisie. This may suggest that small business where the petty bourgeoisie tends to concentrate provides a niche for certain ethnic enter- prises to thrive (Light, 1972; Ward and Jenkins, 1984; Cummings, 1980; Bonacich and Modell, 1980). However, in the managerial class and the work- ing class, non-white males earn about 80 per cent of what white men earn. These findings suggest that race puts non-white men in an income disadvan- tage relative to white men, but being male enables both groups to have an income advantage over women whether white or non-white. These differ- ences in earnings among the racial and gender fractions are not attributable to class variations, since a separate equation is used for each class. Other factors, such as differential patterns of employment and individual varia- tions in age and schooling may have accounted for some of the gross differ- ences.
Table 111 shows the net effect of race and gender on earnings for each class. The beta coefficients confirm that the overall effect of these fractions tends
602 PETER S. LI
to be smaller for employers and the petty bourgeoisie (0.15), but moderately stronger for the professional class (0.17), and much stronger for the manage- rial class and the working class (0.23). Overall, there is a reasonable ‘fit’ of the mode&mong the managerial class, the professional class, and the work- ing class.
In terms of the actual earning gaps, Table 111 further shows that when differences in other variables are taken into account, white men have the highest income level, and non-white women have the lowest in all classes, although their earnings are only marginally less than those of white women. Similar to non-white women, white women’s income is substantially below the mean earnings for all classes. The earning gap between women, white and non-white, and white men, is over $10,000 in the employers class and the managerial class, where the average earnings are higher. In contrast, such an income gap is about six to seven thousand dollars in the petty bour- geoisie and the working class, where the average income is the lowest. This finding is consietent with that reported in the literature which indicates that gender discrepancy in earnings is greater in high-wage occupations (Fill- more, 1990). Non-white men earn more than women, but less than white men, in all classes. Their advantage (gross effect) over white men’s income in the petty bourgeioisie disappears when variations in other variables are taken into account, thus suggesting that some of the original advantage may be attributed to differences between white males and non-white males in their patterns of work, such as working in different sectors and for unequal number of weeks, and other individual variations.
CLASS, GENDER AND RACE RECONSIDERED
The writings of Marx and Poulantzas have provided the theoretical grounds for analysing class fractions based on productive, ideological and political relations. The literature has considered race, and to a lesser extent gender, as bases of fractionalizing the working class. However, how race and gender operate interactively in fractionalizing all classes has not been systemati- cally examined.
This paper analyses the effects of race and gender on earnings in five classes constructed on the basis of ownership and control of capital, and con- trol of labour. The findings indicate that while both race and gender have an effect on earnings, there are compelling empirical grounds to consider them interactively to produce four categories that have unequal effects on earnings. As expected, men have an income advantage over women; however, race fractionalizes the earnings of white and non-white men to a much greater extent than for women. When inter-class differences in earn- ings are removed, the data show that race and gender remain important in explaining earnings, although their effects, in terms of explained variance, are stronger among the managerial class, the professsional class, and the working class, and weaker among the employers and the petty bourgeoisie. This pattern persists despite adjusting for differences in other variables in the equation. Therefore, the findings lend support to the claim that race and
503 EFFECTS OF RACE AND GENDER ON EARNINGS
gender fractionalizations operate more strongly among wage-earners than among those who own and control capital.
In terms of income gaps, the findings suggest that despite controlling for inter-class variations and inter-group differences in the pattern of work and individual factors, white men have the highest income, followed by non- white men; non-white women have the lowest income, although only margi- nally lower than white women. The gender gap in earnings is more pronounced, but race remains important in segmenting the earnings of men. Non-white males have an income advantage over white males among the petty bourgeoisie, but such an advantage disappears when differences in such variables as schooling, age, and the nature and duration of work are taken into account.
The earnings of non-white women are slightly lower than those of white women among all the classes being considered, and the difference among white and non-white women in the petty bourgeoisie class is virtually nil. Previous works in Canada and the U.S. have reported similar findings about non-white women earning not much less than white women, especially &er variations in o t p variables have been statistically adjusted (Goyder, 1981; Epstein, 1973). For example, using data from the 1973 Canadian Mobility Study, Goyder (1981: 334) shows that net income difference between immi- grant women born in the West Indies and Canadian-born English-speaking women was small. The literature has also provided further insights into this phenomenon. It has been shown that gender gaps in earnings tend to be higher among high wage occupations, and that the economic rewards are low for both men and women in low-paying jobs (Krahn and Lowe, 1988; Fillmore, 1990). In other words, income disparity tends to attenuate at the lower income levels, where there is what Goyder (1981: 332) calls the ‘floor effect’. The findings in this paper provide further credence to this observa- tion in that earning gaps are relatively large between white and non-white women in the managerial class, the group with the highest average earn- ings, and smallest among the petty bourgeoisie which has the lowest aver- age earnings. The same argument may also elucidate why race is more important in producing disparities among higher income groups (white and non-white males), but less important among lower income groups (white and non-white females). In other words, in the same way that gender dis- parity in earnings tends to be less pronounced in lower-paying occupations, racial differences in earnings among women (lower income levels) are less as compared to racial differences among the higher income groups for men.
The findings that women systematically earn less than men in all classes irrespective of race, and that race seems to make a difference in the earn- ings of men and only marginally in the earnings of women invite further theoretical considerations. On a general level, the data certainly lend sup- port to Stasidis’ assertation that all social relations in Canada have ele- ments of class, gender and race (Stasidis, 1990: 269). Specifically, the persistent earning gaps among the gender and racial groups call for further consideration of the debate concerning the primacy of class, gender and race in producing social inequality.
604 PETERS. LI
The fact that white and non-white women systematically earn less than their male counterparts in all classes suggests that gender inequality can- not be adequately subsumed under class relations. Nor is it sufficient to argue that gender and race produce class fractions for the benefit of capital. These theoretical positions would lead to the empirical expectations that earning gaps between gender and racial groups would disappear except among wage-earners, contrary to what this paper reports. David H.J. Mor- gan and Daphne E. Taylorson (1983: 9-10] have warned against that ver- sion of Marxist functionalism which holds the dominance of class relationships over other relationships, and reduces gender relationships to class contradictions under the capitalist social formation. It is equally in- adequate to attribute the source of gender inequality to the imperatives of capitalism, in the same way that Robert Blauner (1972) and Miles (1980) have found the reduction of racism to the struggle between capital and labour unsatisfactory. The data in this paper confirm that such reduction- ism is inadequate and misleading.
Since gender gaps in remuneration persist in all classes, the disparities would suggest that there are fundamental differences in the working ex- periences of men and women. Such differences have been clearly docu- mented in the literature. For example, Eva Gamarnikow et al. (1983) show that the work experiences and consciousness of working-class women are different from those of working-class men and middle-class women in Eng- land. In Canada, S.J. Wilson (1986: 75-98) has followed an extensive litera- ture which shows how women had been proletarianized under industrial capitalism in Canada, and how in the process, women’s work and working experiences had been marginalized. Regarding non-white women, Daiva K. Stasiulis (1990) has argued that their experiences are different from white women in work, political power and ideology. Roxana Ng (1988) and Ng and Judith Ramirez (1981) have shown that the unique marginality of immigrant women in Canada is connected to their vulnerable economic, political and ideological positions. Elsewhere, Bell Hooks (1984, 1988) and Angela Y. Davis (1981) have described the complex and diverse experiences of non- white women; they have argued that racism and sexism should be considered jointly to understand how such experiences are distinct from those of white women and black men.
Why the working experiences of women are different from those of men has been the subject of intense scrutiny. There is little doubt that some of the gender differences in work experiences can be attributed to the way in which the labour market has been constituted and the manner in which re- lationships of production have been transformed. For example, the arrange- ments of the labour market according to gender-linked jobs, gender segregation and sectorial segmentation have set the structural parameters under which low pay, intermittent employment and undesirable working conditions become the characteristics of women’s work in the labour market (Armstrong and Armstrong, 1984; Connelly, 1978; Carroll, 1987; Ornstein, 1983). However, the fact that substantial earning gaps persist between men and women despite statistical control of major market factors, and that
606 EFFECTS OF RACE AND GENDER ON EARNINGS
gender gaps prevail across all classes suggests that labour market arrange- ments and work experiences are insufficient to understand the earning out- come for men’s work and women’s work. The present data also illustrate the inadequacy of the argument premised upon attributing gender segrega- tion only to the benefits of capital.
Many feminist scholars have been arguing that gender relationships are found not only in relationships of production, and that a system of partri- archy, which includes culture, ideology, the family, the educational system and the state, has been responsible for the production and perpetuation of gender relationships (Barrett, 1980; Brittan and Maynard, 1984; Wilson, 1986). Such a system has enabled men to gain and to maintai dominance over women for much of human history (Ramazanoglu, 1989). It is fair to say that the source of gender inequality is rooted only in part in the way capitalist production is structured, but also in an extensive way in the patri- archal structure and ideology of society which produce and sustain the dom- inance of men over women from precapitalist economy to capitalist production (Ramazanoglu, 1989; Gamarnikow et al., 1983).
Neither the present data nor the present state of theory is refined enough to enable linking empirically gender relationships outside the sphere of pro- duction to job outcomes such as earning disparities. But given the compel- ling theoretical arguments developed by many feminist scholars, it would be inconsistent to consider gender inequality in the sphere of production as not having some of its prior roots in the production and maintenace of gender relationships in society.
Finally, it should be noted that income inequality is only one of the many aspects of inequality. Obviously, those who could not find work or others who did not work would not have an employment or self-employment in- come, and therefore would not be considered in this study. It is quite likely that ‘race’ and ‘gender’ operate in a rather different way in differentiating the opportunity of employment and self-employment beyond what earning differentials can capture. At the very least, the present study suggests that there are considerable theoretical and empirical merits in considering the interactive effects of race and gender as fractionalizing the earnings within each social class, However, the findings also challenge the simplistic inter- pretation of subsuming race and gender relations under capitalist social for- mation, and they suggest the need to link the formation of race and gender relations outside the sphere of production to class relations as a important step to intergrate a better theory of gender, race and class.
NOTES
1
ili
There are alternative theoretical paradigms which can be used to study the relation- ships of race, gender and status. A substantial portion of the literature in stratification research in Canada and the U.S. has followed the ‘status-attainment model,’ which prem- ises primarily on the Weberian notion of social status and shares with some of the as- sumptions in human capital theories (see, for example, Horan, 1978; Kerckhoff, 1976). The most comprehensive report on status attainment in Canada using this framework is provided by Boyd et al. (1986).
606 PETER S. LI
Barrera (1979: 211-2) uses the term ‘class segments,’ rather than ‘class fractions’ to refer to intraclass divisions. He further distinguishes two types of divisions; the first, b a d on the structure of occupations, is called structural class segments, and the aec- ond, baaed mainly on race and sex, is called wriptive clam segments. Fox and Fox (1986) show that the gender income gap, as measured by the ratio of mean female to mean male earnings, diminished from 1941 to 1961, and widened from 1961 to 1971, but then decreased again from 1971 to 1981. For a discuseion of the sample design and eatimation, and problems dec t ing data relia- bility, see Statistice Canada (1990: 127-49). By definition, immigrants who immigrated to Canada in 1986 are not included, since Statistice Canada treats them as not having worked in Canada in 1986, and automati- cally assigns their 1986 income as zero (Statistics Canada, 1990: 108). Wages and ealariea are always positive, but self-employment income can take on a nega- tive value. The upper and lower limits are $100,000 and -$30,000, except for males out- side the Atlantic region, whose limits are $140,000 and -$60,000 (Statistics Canada, 1990: 140). The application of income ceilings changes the wages and salaries of 697 in- dividuals, or 0.2% of all recipients from this source of income, and changes the self-em- ployment income of 463 individuals, or 1.9% of all recipients from this source of income. In terms of actual aggregate income being af€&, the ceilings alter 0.9% of total wages and ealariea, and 4.4% of total self-employment income (Statistics Canada, 1990: 143). In subsequent writings, Wright (1986) reformulates the typology of class locations in capitaliet societies. His reformulation involves using the notion of exploitation baaed on ownership of assets in the means of production, organizational assets, and credential as- sets to produce 12 class locations. However, the new typology lacks the theoretical and operational elegance of the original formulation. For a debate of Wright’s reformulation, see Carchedi (1987) and Wright (1989). Class categories constructed from census data are subject to several measurement er- rors. For example, many members of the grand bourgeoisie are holders of controlling stocks and shares of corporations, but they may show up in the census as employees of corporations or they may be classified as not having participated in the labour force. Cuneo (1986: 471) indentifiea two other problems: I/ since 1961, Statistics Canada col- lapsed ‘own account workers’ and ‘employers’ as self-employed; and 2/ since 1971, owners of incorporated businesses were classified as employees of their corporations. The census categories used to classif) ‘professionals’ as a class include those engaged in scientific fields (natural sciences, engineering, mathematics and social sciences) and those employed in teaching, medicine, health, artistic, literary and recreational occupa- tions.
10 Statistics Canada (1990: 71-72) classifies respondents who indicated multiple responses to the ethnic origin question as belonging to the appropriate single ‘visible minority’ if the multiple responses included any one of the officially defined ‘visible minority’ group; otherwise, they are considered as belonging to the multiple ‘visible minority’ group if the multiple responses fell into more than one of the visible minority groups.
11 Although the number of visible minorities included in the analysis appears to be large enough to permit finer analyis of the subgroups, the actual number for some subgroups is small, especially when ‘gender’ and ‘class’ are considered simultaneously. For ex- ample, there were only 207 non-white females in the employer class, and 209 non-white female and 637 non-white male in the petty bourgeoisie class (Table 111).
12 Since some full-time workers excluded their paid vacation or sick leave, contrary to in- etructions, those who reported working 49-52 weeks are collapsed, as suggested by Statistice Canada (1990: 46).
13 Several statistical features affect the ability of a variable to explain the variance of the dependent variable. For example, if the variance within the categories of the inde-
607 EFFECTS OF RACE AND GENDER ON EARNINGS
pendent variable is relatively large, it would reduce the size of the correlation coeffi- cient, but the mean difference can remain significant. The number of cases in the catego- ries of the independent variables also affects the correlation coefficient. For example, other things being equal, the disproportionate number of cases in the white and non- white categories would tend to minimize the variance for ‘race,’ in the same way that the almost equal number of cases in the male and female categories would tend to maxi- mize the variance for ‘gender’. It is quite likely that these differences produce in part the differential effects of gender and race on earnings.
14 The R2 is one measure of the goodness of fit of a model as it shows the total non-redun- dant overlap of all the independent variables with the dependent variable. However, statisticians have warned against the problem of attenuation in measurements of asso- ciation as a result of measurement errors, and have suggested that comparisons of corre- lation coefficients across populations do not yield very useful information. For this and other reasons, Blalock (1972: 46-62) argues that correlation coefficients should be seen only as measuring the amount of unexplained variation in testing the adequacy of a given model. In the present context, the correlation coefficients are merely used to show in which class category there is a better statistical ‘fit’ of the model.
16 There are different interpretations of this finding in the literature. Epstein (1973) ar- gues that the ‘multiple negative’ of being black and female operates in favour of some black female professionls in that poverty compels many black mothers to have to work, and thereby set work role models for their daughters. This interpretation contradicta the descriptions and arguments of black women provided by black feminist scholars such as Davis (1981) and Hooks (1984; 1988). In Canada, Goyder (1981: 334) finds that the ‘multiple negative’ of being female and non-white produces negative effects in earn- ings for immigrant women born in the West Indies.
16 There are substantial disagreements among feminist scholars regarding the definition of patriarchy and whether patriarchy is a universal or historical phenomenon. For a lucid discussion of these debates, see Ramazanoglu (1989) and also Barrett (1980).
REFERENCES
Andrews, Frank M., James N. Morgan, John A Sonquist, and Laura Hem 1976 Multiple Classification Analysis. Ann Arbor: Institute For Social Research,
Armstrong, Pat, and Hugh Armstrong 1984 The Double Ghetto. Toronto: McClelland and Stewart Barrera, Mario 1979 Race and Class in the Southwest: A Theory of Racial Inequality. Notre
Barrett, Michele 1980 Women’s Oppression Today. London: Verso Blalock, Hubert M. Jr. 1972 Causal Inferences in Nonexperimental Research. New York: W.W. Norton Blauner, Robert 1972 Racial Oppression in America. New York: Harper and Row Bolaria, B. Singh, and Peter S. Li 1988 Racial Oppression in Canada, Second Edition. Toronto: Garamond Press Bonacich, Edna 1972 ‘A theory of ethnic antagonism: the split labour market.’ American Sociologi-
University of Michigan
Dame: University of Notre Dame Press
cal Review 37: 647-59
608 PETER S. LI
1975 ‘Abolition, the extension of slavery, and the position of free blacks: a study of split labour markets in the United States, 1830-1836.’ American Journal of Sociology 81: 601-28
1976 ‘Advanced capitalism and blacwwhite race relations in the United States: a split labour market interpretation.’ American Sociological Review 41: 34-51
Bonacich, Edna, and John Model1 1980 The Economic Basis of Ethnic Solidarity. Berkeley: University of California
Boyd, Monica, John Goyder, Frank E. Jones, Hugh A. McRoberts, Peter C. Pineo,
1985 Ascription and Achievement: Studies in Mobility and Status Attainment in
Brittan, Arthur, and Mary Maynard 1984 Sexism, Racism and Oppression. Oxford: Basil Blackwell Carchedi, Guglielmo 1987 Class Analysis and Social Research. Oxford and New York: Basil Blackwell Carroll, William K. 1987 ‘Which women are more proletarianized? Gender, class and occupation in
Press
and John Porter
Canada. Ottawa: Carleton University Press
Canada.’ The Canadian Review of Sociology and Anthropology 24(4): 571- 85
Castles, Stephen, and Godula Kosack 1985 Immigrant Workers and Class Structure in Western Europe. New York:
Oxford University Press Connelly, Patricia 1978 Last Hired, First Fired: Women and the Canadian Work Force. Toronto:
Cox, Oliver C. 1948 Caste, Class and Race: A study in Social Dynamics. New York: Doubleday Creese, Gillian 1988 ‘The politics of dependence: women, work, and unemployment in the Van-
couver labour movement before World War II.’ Canadian Journal of Soci- ology 13: 121-42
problem”.’ B.C. Studies 80: 24-51
Women’s Press
198fi-89 ‘Exclusion or solidarity? Vancouver workers confront the ”Oriential
Cummings, Scott (ed.) 1980 Self-Help in Urban America: Patterns of Minority Business Enterprise. New
Cuneo, Carl J. 1985 ‘Have women become more proletarianized than men?’ The Canadian
Davis, Angela Y. 1981 Women, Race and Class. New York: Random House Epstein, C.F. 1973 ‘Positive effects of the multiple negative: explaining the success of black
Fillmore, Catherine J. 1990 ‘Gender differences in earnings: a re-analysis and prognosis for Canadian
York: Kennikat Press
Review of Sociology and Anthropology 22(3): 465-95
professional women.’ American Journal of Sociology 78: 912-35
women.’ Canadian Journal of Sociology 15: 275-99
609 EFFECTS OF RACE AND GENDER ON EARNINGS
Fox, Bonnie J., and John Fox 1983 ‘Effecta of women’s employment on wages.’ Canadian Journal of Sociology
1986 ‘Women in the labour market, 1931-81: exclusion and competition.’ The
Freedman, Francesca 1975 ‘The internal structure of the American proletariat: a Marxist analysis.’
Socialist Revolution (0ct.-Dec.): 41-83 Gamarnikow, Eva, David H.J. Morgan, June Purvis, and Daphne E. Taylorson
(eds.) 1983 Gender, Class and Work. London: Heinemann Goyder, John C. 1981 ‘Income differences between the sexes: findings from a national Canadian
Hooks, Bell 1984 Feminist Theory: From Margin to Center. Boston: South End Press 1988 Talking Back: Thinking Feminist, Thinking Black. Toronto: Between the
Lines Horan, Patrick M. 1978 ‘Is status attainment research atheoretical?’ American Sociological Review
Kerckhoff, Alan C. 1976 ‘The status attainment process: socialization or allocation?’ Social Forces
Krahn, Harvey J. and Graham S. Lowe 1988 Work, Industry and Canadian Society. Scarborough, ON: Nelson Li, Peter S. 1988a Ethnic Inequality in a Class Society. Toronto: Wall and Thompson 1988b The Chinese in Canada. Toronto: Oxford University Press Light, Ivan H. 1972 Ethnic Enterprise in America: Business and Welfare Among Chinese,
Japanese, and Blacks. Berkeley: University of California Press Lowe, Graham S. 1980 ‘Women, work and the office: the feminization of clerical occupations in
Makabe, Tomoko 1981 ‘The theory of the split labour market: a comparison of the Japanese ex-
Marx, Karl 1963 The Eighteenth Brumaire of Louis Bonaparte. New York: International
Miles, Robert 1980 ‘Class, race and ethnicity: A critique of Cox’s Theory.’ Ethnic and Racial
1982 Racism and Migrant Labour. London: Routledge and Kegan Paul Morgan, David H.J., and Daphne E. Taylorson 1983 ‘Introduction: Class and work: Bringing women back in.’ Pp. 1-10 in Eva
Gamarnikow et al. (eds.), Gender, Class and Work. London: Heinemann
8: 319-29
Canadian Review of Sociology and Anthropology 23(1): 1-21
survey.’ The Canadian Review of Sociology and Anthropology 18(3): 32142
43: 53441
55: 368-81
Canada, 1901-1931.’ Canadian Journal of Scoiology 5: 361-81
perience in Brazil and Canada.’ Social Forces 59: 786-809
Publishers
Studies 7: 217-37
510 PETER S. LI
Muszynski, Alicja 1988 ‘Race and gender: structural determinants in the formation of British
Columbia’s salmon cannery labour force.’ Canadian Journal of Sociology 13: 103-20
Ng, Roxana 1988 The Politics of Community Services: Immigrant Women, Class and The
State. Toronto: Garmond Ng, Roxana, and Judith Ramirez 1981 Immigrant Housewives in Canada. Toronto: Immigrant Women’s Centre Ornstein, Michael 1983 Accounting for Gender Differentials in Job Income in Canada: Results from
Phizacklea, Anne, and Robert Miles 1980 Labour and Racism. London: Routledge and Kegan Paul Poulantzas, Nicos 1978a Political Power and Social Classes. London: V e m 1978b Classes in Contemporary Capitalism. London: Verso Ramazanoglu, Caroline 1989 Feminism and The Contradictions of Oppression. London and New York
Routledge Rosenfeld, Rachel A,, and Arne L. Kalleberg 1990 ‘A cross-national comparison of the gender gap in income.’ American
Journal of Sociology 96: 69-106 Satzewich, Victor, and Peter S. Li 1987 ‘Immigrant labour in Canada: the cost and benefit of ethnic origin in the job
Stasiulis, Daiva K. 1990 ‘Theorizing connections: Gender, race, ethnicity, and class.’ Pp. 269-305 in
a 1981 Survey. Ottawa: Women’s Bureau, Labour Canada
market.’ Canadian Journal of Sociology 12: 229-41
Peter S. Li (ed.), Race and Ethnic Relations in Canada. Toronto: Oxford University Press
Statistics Canada 1990 Public Use Microdata File On Individuals: Documentation and User’s Guide Ward, Robin, and Richard Jenkins (eds.) 1984 Ethnic Communities in Business: Strategies for Economic Survival. Cam-
Wilson, S.J. 1986 Women, The Family and The Economy, Second Edition. Toronto: McGraw-
Hill Ryerson Wright, Erik Olin 1977 ‘Marxist class categories and income inequality.’ American Sociological
1978 ‘Race, class, and income inequality.’ American Journal of Sociology 83: 1368-
1979 Class Structure and Income Determination. New York: Academic Press 1985 Classes. London: Verso 1989 The Debate on Classee. London: Verso
bridge: Cambridge University Press
Review 42: 32-55
97