The Use of Census Data in the

1
The Use of Census Data in the Analysis of
Ethnic Inequalities The Canadian Experience,
1996-2006 Nikolaos I. Liodakis Department of
Sociology Wilfrid Laurier University Waterloo,
ON, Canada nliodakis_at_wlu.ca
2

• Social Inequality in Canada
• The Theoretical and Research Contexts
• Ethnic perspective dominant in the analysis of
  social inequality in Canada since Porter (The
  Vertical Mosaic, 1965)

• Immigration and ethnic affiliation important
  factors in the process of social class formation
  in Canada

• The Canadian ethnic mosaic is vertical (Charter
  vs. entrance groups, etc.)
• Ethnicity, associated with class, implied blocked
  mobility

• Despite acknowledging their internal
  stratification (1965 72-73), Porter treated
  ethnic groups as homogeneous entities
• The gender and nativity dimensions of social
  inequality among groups have been adequately
  examined, but there exists, still, a relative
  silence on social class(es)

3

• Research on
• Social Inequality in Canada
• Ethnic perspective a dominant research
  tradition enjoys discursive hegemony it is the
  dominant narrative of social inequality in Canada

• Ethnic perspective treats ethnic and visibility
  groups as
• homogeneous entities
• This conceals important class, gender and
  nativity (and other) divisions and variations
  within groups

• Visibility Thesis colour has replaced ethnicity
  in the structure of social inequality in Canada
• Race The fundamental basis of social
  inequality in Canada (e.g., Gabaluzi 2007,
  Canada's Economic Apartheid)

4

• Inter-sectionalist Theorizing
• Mono-causal, mono-dimensional approaches
  inadequate for analyzing social relations
  complex character of social inequality

• Integrationist / inter-sectionalist approaches
  class, gender
• and race/ethnicity fundamental axes of social
  organization important in the matrix of
  domination (but not dominant perspective,
  Stasiulis,1999)
• Must understand the simulteneity, interactivity
  of class,
• gender, race/ethnicity
• Although analytically distinct inter-locking,
  mutually-
• determining, reinforcing social relations
• Any two sets of social relations are always
  present in, permeate, inform and (re-)define the
  third

5
Bringing Classes Back In

• Conspicuously absent from quantitative studies of
  social
• inequality in Canada, especially from analyses of
  ethnic
• earnings differentials
• Canadian social formation not only patriarchal,
  racist also
• capitalist
• What is the picture of social inequality that
  emerges if we begin from a different point of
  departure?
• Re-introduce classes, retain gender and ethnicity
  
• A holistic, more comprehensive, and arguably more
  accurate
• picture of social inequality in Canada?

6

• Findings
• Ethnic groups are not homogeneous entities

• There exist class, gender and nativity earnings
  structures

• The class and gender earnings differentials
  within ethnic groups are greater than the
  differentials among ethnic groups

• Class and gender are persistent bases of social
  inequality in Canada

• Human capital and labour market/production
  variables affect more earnings differentials
  than ascriptive variables, with the exception
  of gender

7
Ontology, Epistemology and Methodogical
Limitations
Data-gathering is theory-driven Censuses rely
upon dominant, well-entrenched paradigms tend to
privilege the ethnic perspective undermine other
perspectives (e.g., class) Running out of
single-origin groups (the categories are
emptying themselves out) ? Racialization of
analysis of social inequalities State-driven
scholarship? Language (construction of
unreal social reality) some ethnic/visibility
groups nominal-statistical categories without
social referents Stats Can constructs
Caribbeans, Balkans, Southern Europeans, South
Asians, Latin Americans, etc. Problematic
epistemic relationships? (Harrés taxonomic
collectives)
8
Ontology, Epistemology and Methodogical
Limitations

• Several problems
• By relying on the visible minority indicator
  alone, we conceal the important and sizable
  economic inequalities that exist within the
  category (e.g., earnings of Chinese-origin
  respondents vs. those of Filipino-origin)
• We cover the complexity of social inequality by
  reducing it to a two-tier hierarchy
  (visible/non-visible categories). We homogenize
  both the Self and the Other
• By overemphasizing the ethnic/visibility
  dimension of inequality, we obfuscate its class
  and gender dimensions

9
Problems with ethnic questions in Censuses

• 1. No question on respondents own subjective
  ethnic/cultural identity we ask about their
  ancestry instead. This has created the multiple
  origins categories that are increasing in size
  (a problem in counting and reporting
  results).
• 2. A racist definition of visible minorities
  (Synnott and Howes 1996 a spurious category)
• not white in colour
• not Caucasian in race
• uncritical use of geography
• 3. Defining ethnicity through ancestry ? we know
  nothing about the economic inequalities among
  self-identified ethnics we know everything
  about offspring of ancestors

10
Ontology and Implications
We do not interrogate the categories. We
reproduce the language of the oppressor
(Goldberg). Social scientists paint a picture
of inequality that does not exist, or exists only
as a statistical reality. This image informs and
legitimizes the images reproduced by mass media,
but also the stereotypes in the minds of
non-social scientists, and thus, their social
praxis (Miles and Torres) It (mis)directs public
policy makers and social action. We need a
new, anti-racist meta-language a new
emancipatory vocabulary that transcends the
racial categories. Any suggestions?
11
Proposed Solution(s)

• A. Ask respondents about their own subjective
  ethnic/cultural identity retain the questions of
  ancestry for data comparability reasons
• The 2002 Stats Canada Ethnic Diversity Survey did
  ask about
• respondents self-identification of ethnicity,
  but the question
• sequence is questionable (it followed a number of
  questions
• about their ancestry skewed results?)
• Although it was clearly stated in the Survey that
  the information
• gathered might be used in conjunction with the
  2001 question
• on ethnic ancestry to enhance the information
  provided for the
• Survey, inexplicably, the self-identification
  question was not
• used in the 2006 Census.

12
Problematic Issues

• If we ask about respondents own subjective
  ethnic/cultural identity, we will get smaller
  numbers in some ethnic groups (those who have
  been longer in Canada).
• The category Canadian will increase in size even
  more (So? Why not? See Howard-Hassmann)
• Negative implications for the image of Canada
  as a multicultural society?
• Implications for the notion of the Canadian
  nation(s)?
• Political pressures there may be additional
  critiques against multiculturalism, from the
  right, and yet another call for dismantling the
  policy all together (for the wrong reasons)

13
Proposed Solution(s)

• B. No solution to the visibility issue in the
  foreseeable future.
• Any suggestions?
• How do we, or should we collect racial data,
  useful for
• combating inequality, discrimination and racism,
  without asking
• racist questions?
• Post-structuralists may propose the use of
  strategic
• essentialism, à là Derrida (see Thobani 2007).
• We do and will continue to reproduce racist
  language in the
• near future.

14
Circumventing the Issue

• Caveat the following pseudo-solution may not
  apply in other
• contexts, although it will certainly have
  implications
• Collect ethnicity data in Censuses by asking
  respondents to self-report their subjective
  ethnic identity retain the ancestry and
  visibility questions (but what we do with
  multiples?)
• Researchers then could disaggregate the category
  visible minority to its constituent parts
  (disentangle the self-identified ethnic
  categories contained in the visibility category)
• This will provide rich data on the economic
  inequalities within
• the visibility category. Arguably, a more
  comprehensive and
• accurate picture of economic inequality. A
  positive implication
• for social policy, especially if we include class
  and gender
• issues.

15

Evidence of Ethnic Heterogeneity

• Ethnic groups are not homogeneous, monolithic
  entities.
• They differ in their class, gender and nativity
  compositions.

16
The Sex Composition of Groups
17

• Evidence of Ethnic Heterogeneity
•  
• Ethnic groups are not homogeneous, monolithic
  entities.
• They differ in their class, gender and nativity
  compositions.

• Sex Filipino- and Caribbean-descent respondents
  have a significantly higher percentage of females

18
The Nativity Composition of Groups
19

• Evidence of Ethnic Heterogeneity
•  
• Ethnic groups are not homogeneous, monolithic
  entities.
• They differ in their class, gender and nativity
  compositions.
• Sex Filipino- and Caribbean-descent respondents
  have a significantly higher percentage of females
• Nativity Filipino-, South Asian-, Caribbean-,
  Chinese-, Portuguese- and Greek-descent
  respondents have a significantly higher
  percentage of foreign-born

20
The Class Structure of the Sample
350000
300000

250000
Number of People
200000
150000
100000
50000
0

N
17,067
6
Employers (E)
24,013
8
Petty Bourgeoisie (PB)
11
Managers and Supervisors (MS)
32,742
19
57,924
Semi-autonomous Workers (SAW)
56
169,449
Proletariat (P)
Classes
21
The Class Composition of Groups ()
ETHNIC GROUP P SAW MS PB E
British (31,986) 55.7 18.1 12.4 08.4 05.4
Caribbean (3,460) 69.9 18.0 06.2 03.6 02.3
Chinese (8,763) 53.7 21.6 08.7 07.8 08.2
Filipino (2,811) 74.9 16.4 05.0 02.3 01.4
French (30,498) 55.4 22.4 10.7 06.7 04.8
Greek (1,784) 56.1 13.1 09.1 08.5 13.2
Italian (9,028) 58.4 15.6 12.4 05.8 07.8
Jewish (2,118) 33.0 28.7 11.9 11.8 14.6
Portuguese (2,936) 74.7 07.7 08.5 04.5 04.6
S. Asian (6,038) 63.8 16.4 08.5 05.1 06.2
SAMPLE (301,195) 56.0 19.0 11.0 08.0 06.0
22
The Class Structure of Men and Women in the
Sample
70
60
50
40
Percentage
30
20
10
0
P
SAW
MS
PB
E
Men
52.7
16.9
13.6
9.2
7.6
60.3
21.9
7.7
6.6
3.4
Women
Classes
23
The Class Structure of the Sample by Gender and
Nativity
70
60
50
40
Percentage
30
20
10
0
P
SAW
MS
PB
E
51
18.3
12.2
9.5
9
FB Men
53.2
16.5
14
9.1
7.2
NB Men
63.7
18.7
6.7
6.2
4.6
FB Women
59.4
22.8
8
6.7
3.1
NB Women
Classes
24
The Class Structure of Chinese-descent
Respondents Foreign-born and Native-born Men and
Women
70
60
50
Percentage
40
30
20
10
0
P
SAW
MS
PB
E
51
18.3
12.2
9.5
9
All FB Men
46.4
22.7
10.8
9.8
10.3
Ch. FB Men
53.2
16.5
14
9.1
7.2
All NB Men
37.7
36.3
13.6
6
6.4
Ch. NB Men
63.7
18.7
6.7
6.2
4.6
All FB Women
64.1
17.4
5.7
6.1
6.8
Ch. FB Women
59.4
22.8
8
6.7
3.1
All NB Women
47.4
33.3
11.6
6.1
1.6
Ch. NB Women
Classes
25
The Class Structure of Greek-descent Respondents
Foreign-born and Native-born Men and Women
70
60
50
Percentage
40
30
20
10
0
P
SAW
MS
PB
E
All FB Men
51
18.3
12.2
9.5
9
Gr. FB Men
49.4
7.4
8.2
12.9
22.3
All NB Men
53.2
16.5
14
9.1
7.2
Gr. NB Men
47.4
20.1
12.4
8
12.1
63.7
18.7
6.7
6.2
4.6
All FB Women
Gr. FB Women
68.6
9.7
7.3
6.5
8
59.4
22.8
8
6.7
3.1
All NB Women
61.3
21.4
10.1
3.1
4.1
Gr. NB Women
Classes
26

• Heterogeneity Within Ethnic groups
• Class only Jewish-descent respondents have an
  atypical class structure 

• If ethnic groups are conceived as units, no clear
  ethnic class pattern emerges

• Gender is introduced females are
  over-represented in the P and SAW
  under-represented in all other classes

• Nativity is introduced NB females are
  under-represented in the P and E. NB men are
  over-represented in the P and in MS

• When all three dimensions are examined within
  each ethnic group, no clear pattern emerges.

27

• Earnings Differentials
•  
• There exist class, gender and nativity earnings
  structures.

1. Males make more than females in all classes

3. NB males make more than the FB. NB females
make more than FB expect in the PB
28
Earnings of the Sample
Median
29
Sample Mean of Earnings by Class
CLASS N Mean Median SD
Proletariat 169,449 24,720.27 23,000.00 18,133.27 -17.7
Semi-autonomous Workers 57,924 38,696.11 38,000.00 23,642.01 28.8
Managers Supervisors 32,742 45,573.20 40,505.00 30,735.07 51.7
Petty Bourgeoisie 24,013 19,929.24 13,250.00 24,859.07 -33.6
Employers 17,067 37,810.65 26,353.00 39,570.58 12.6
Sample 301,195 30,034.67 26,292.00 24,521.12  
30
Sample Mean of Earnings by Sex
SEX N Mean Median SD of Mean
Males 160,961 36,138.56 32,319.00 27,718.33 20.3
Females 140,234 23,028.60 20,482.00 17,825.07 -23.3
Sample 301,195 30,034.67 26,292.00 24,521.12  
31
Sample Mean of Earnings by Nativity
PLACE OF BIRTH N Mean Median SD of Mean
Foreign-born 66,108 28,518.97 24,000.00 24,966.49 -5.0
Native-born 235,087 30,460.89 27,000.00 24,377.50 1.4
Sample 301,195 30,034.67 26,292.00 24,521.12  
32

• Earnings Differentials Among and
• Within Ethnic Groups
•  
• If ethnic groups are conceived as units, then,
  there is an ethnic rank order of earnings

33
Earnings by Ethnic Group
ETHNIC GROUP N Mean Median SD of Mean
Jewish 2,118 43,269.28 33,024.00 38,794.69 44.0
British 31,986 33,434.28 30,000.00 26,752.78 11.3
Italian 9,028 31,155.10 28,714.00 23,443.39 3.7
French 30,498 29,916.59 26,899.50 23,121.85 -0.4
Portuguese 2,936 26,521.62 25,000.00 18,069.63 -11.7
South-Asian 6,038 25,792.93 21,000.00 23,200.98 -14.1
Chinese 8,763 25,747.03 21,000.00 22,945.79 -14.3
Greek 1,784 24,723.16 20,000.00 20,825.40 -17.7
Caribbean 3,460 24,005.59 22,481.00 18,296.79 -20.0
Filipino 2,811 22,548.16 20,000.00 16,876.62 -24.9
Sample 301,195 30,034.67 26,292.00 24,521.12  
34
Examples of Earnings Differentials Within Ethnic
Groups
Jewish-descent Respondents Male, Employers,
Native-born 84,876 - Females, PB,
Foreign-born 15,946 68,930 South
Asian-descent Respondents Males, MS,
Native-born 59,833 - Females, PB,
Foreign-born 14,304
45,529 Filipino-descent Respondents Males,
Employers, Foreign-born 39,426 - Males, PB,
Native-born 09,415 30,011
35

• Earnings Differentials Among and
• Within Ethnic Groups
•  
• If ethnic groups are conceived as units, then,
  there is an ethnic rank order of earnings
• If we examine the class, gender and nativity
  dimensions of earnings within ethnic groups, the
  overall sample patterns hold
• Class and gender earnings differentials are
  greater than ethnic differentials

36
Bivariate Regressions Gross Effects on log
Earnings
Variable R Adj. R2 Explained Class
and Sex (interaction) 0.423 0.179 17.9 Weeks
Worked (1995) 0.383 0.147 14.7 Class 0.338 0.
114 11.4 Full-time 0.322 0.104 10.4 Industry
0.282 0.079 07.9 Sex 0.266 0.071
07.1 Years of Schooling 0.257 0.066
06.6 Age 0.145 0.021 02.1 Ethnicity 0.120
0.014 01.4 Visibility 0.070 0.005
0.05 Place of Birth 0.022 ? 0.00
? 0.00 Language(s) 0.007
? 0.00 ? 0.00
37
Multiple Regression Models on log Earnings
Variables Model Adj. R2 Schooling
1 0.066 Full-Time 2 0.167 Weeks
Worked 3 0.247 Industry
4 0.286 Age 5 0.311 Sex
6 0.342 Class 7 0.379 Ethnicity
8 0.384 Visibility 9 No
improvement Place of Birth 10 No
improvement Official Language(s) 11 No
improvement Class and Sex (interaction)
12 0.386
38

What variables account for inequalities in
earnings?

• More
• Labour market/production Full-time, weeks
  worked, industry, classes of P, SAW, MS
• Human Capital Years of Education
• Ascriptive Gender, Age

39
What variables account for inequalities in
earnings?

• More
• Labour market/production Full-time, weeks
  worked, industry, classes of P, SAW, MS
• Human Capital Years of Education
• Ascriptive Gender, Age
• Less
• Labour market/production Classes of PB and E
• Human Capital Language(s)
• Ascriptive Ethnicity, Visibility, Place of Birth

40

• Findings
• Ethnic groups are not homogeneous entities

• There exist class, gender and nativity earnings
  structures

• The class and gender earnings differentials
  within are greater than the differentials among
  ethnic groups

• Class and gender are persistent bases of social
  inequality in Canada

• Human capital and labour market/production
  variables affect more earnings differentials
  than ascriptive variables, with the exception
  of gender

41
Relative Silence on Class(es)?

• In most qualitative studies focus on immigrant
  women of
• colour actualities of life, multiple
  jeopardies
• Small occupational groups proxies for working
  class
• (domestic labourers, nurses)
• Nativity dimension (foreign-born) Gender
  dimension
• (women)
• Conflate race and/or gender with class?
• Race, gender, and class essentialism (Jhappan,
  1996)
• Relative silence on classes
• In most quantitative studies focus on
  income/earnings
• gender and nativity dimensions within
  ethnic/visibility groups,
• but silence on classes

42
Female Employers by Ethnic Group
43
Semi-autonomous Workers by Ethnic Group
44
Earnings Differentials within Semi-autonomous
Workers
CLASS SEX PLACE OF BIRTH N Mean SD
S-A Workers Males Foreign-born 6,523 44,761.76 28,682.97
Native-born 20,656 45,290.25 26,131.39
Total 27,179 45,163.41 26,766.38
Females Foreign-born 5,689 32,899.10 20,428.09
Native-born 25,056 32,997.06 18,298.15
Total 30,745 32,978.93 18,710.25
Total Foreign-born 12,212 39,235.50 25,861.61
Native-born 45,712 38,552.01 23,011.00
Class Total 57,924 38,696.11 23,642.01
45
Earnings of Chinese-descent Respondents Foreign-bo
rn and Native-born
PLACE OF BIRTH N Mean Median SD
Foreign-born 7,950 24,862.42 20,000.00 22,544.36 -12.8
Mean FB 66,108 28,518.97 24,000.00 24,966.49  
Native-born 813 34,397.31 32,000.00 24,966.84 12.9
Mean NB 23,5087 30,460.89 27,000.00 24,377.50  
46
1960s
THESIS METHODS EMPHASIS
Vertical Mosaic Quantitative European ethnic
National data, groups gross
effects Blocked Mobility Occupation
No Gender or and Income Nativity
dimensions Occupation a
proxy for Class
47
1970s-1980s
THESIS METHODS EMPHASIS
Fading Quantitative European Vertical
Mosaic? National data, and some
non- gross/net effects European
ethnic groups Earnings Occupation and
Convergence Income/Earnings Gender and and
Nativity Mobility dimensions
Relative Silence on Class
48
1990s - Present
THESIS METHODS EMPHASIS
New Mosaic Quantitative Mostly Race,
colour has National data, Visible
minority replaced gross/net effects groups Eth
nicity Occupation and Earnings
Visibility Qualitative
Intersections of Thesis Small focus groups
Gender, Race/ (immigrant women of
Ethnicity, Nativity, No Mobility? colour)
Class (but not the dominant Life
experiences paradigm) Relative
Silence on Class(es)