18 Page Sociology Paper Topic: Gender Discrimination Against Women In The Workplace
Mrs.Shaw
source12.pdf
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 1/27
Full Text | Scholarly Journals
Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 Tomaskovic-Devey, Donald; Stainback, Kevin; Taylor, Tiffany; Zimmer, Catherine; et al.
American Sociological Review; Washington Vol. 71, Iss. 4, (Aug 2006): 565-
588.
Download PDF
Cite Email Print All Options
k 9
Abstract
Title VII of the 1964 Civil Rights Act made employment discrimination and segregation on the basis of
race, ethnicity, or sex illegal in the United States. Previous research based on analyses of aggregate
national trends in occupational segregation suggests that sex and race/ethnic employment segregation
has declined in the United States since the 1960s. We add to the existing knowledge base by
documenting for the first time male-female, black-white, and Hispanic-white segregation trends using
private sector workplace data. The general pattern is that segregation declined for all three categorical
comparisons between 1966 and 1980, but after 1980 only sex segregation continued to decline markedly.
We estimate regression-based decompositions in the time trends for workplace desegregation to
determine whether the observed changes represent change in segregation behavior at the level of
workplaces or merely changes in the sectoral and regional distribution of workplaces with stable
More
Full text Turn on search term navigation
Headnote
Title VII of the 1964 Civil Rights Act made employment discrimination and segregation on the basis of
race, ethnicity, or sex illegal in the United States. Previous research based on analyses of aggregate
national trends in occupational segregation suggests that sex and race/ethnic employment segregation
has declined in the United States since the 1960s. We add to the existing knowledge base by
documenting for the first time male-female, black-white, and Hispanic-white segregation trends using
private sector workplace data. The general pattern is that segregation declined for all three categorical
comparisons between 1966 and 1980, but after 1980 only sex segregation continued to decline markedly.
Sociological Abstracts
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 2/27
We estimate regression-based decompositions in the time trends for workplace desegregation to
determine whether the observed changes represent change in segregation behavior at the level of
workplaces or merely changes in the sectoral and regional distribution of workplaces with stable
industrial or local labor market practices. These decompositions suggest that, in addition to desegregation
caused by changes in the composition of the population of Equal Employment Opportunity Commission
monitored private sector firms, there has been real workplace -level desegregation since 1964.
The Civil Rights Act of 1964 prohibited employers from discriminating against and segregating employees
on the basis of sex, race, or ethnicity. Our descriptive knowledge as to the effectiveness of this defining
legislation is limited. All our systematic knowledge about national change in the employment opportunities
of women and race/ethnic minorities comes from surveys of individuals that describe employment in
terms of occupations detached from their workplace context, the presumed site of discrimination and
the target of antidiscrimination legislation.
The analysis of segregation is important not only because segregation was specifically targeted by the
Civil Rights Act and associated legislation, but also because it is associated with linked employment
inequalities such as wages and promotions (Kmec 2003a; Petersen and Morgan 1995; Tomaskovic-Devey
1993) as well as with the cementing of status expectations and distinctions (Ridgeway 1997; Tilly 1998).
The occupation-based literature suggests that since 1964, both women and minorities have had increased
access to a wider set of occupations, and that there have been declines in occupational segregation.
Occupation-based estimates, however, cannot distinguish change in the distribution of employment
across firms from change in the distribution of employment within firms. Observed declines in national
occupational segregation may simply reflect a shifting of employment to industries, regions, or firms with
lower initial levels of segregation. If this is the case, there may have been no change in the behavior of
employers, only a new distribution of employers with stable segregation practices.
In addition, occupation-based analyses give the potentially misleading appearance that change in sex and
race/ethnic segregation is a national phenomenon. Because employment segregation is essentially a
workplace process, it is likely that segregation levels and trends may vary substantially across
communities and types of workplaces .
In this article, we pursue two linked research questions. First, what are the trends in segregation when
measured at the establishment level? Second, how much of these trends is produced by industrial and
regional shifts in the composition of the economy, and how much is a function of changes in integration
within workplaces ?
Since 1966, the Equal Employment Opportunity Commission (EEOC) has collected yearly data on the sex
by race/ethnic composition of private sector employment as part of federal efforts to monitor compliance
with the Civil Rights Act and similar legislation. These data have rarely been available to the research
community.1 In this article, we summarize trends in private sector workplace segregation and
desegregation using EEO-1 reports available from 1966 to 2003.2 These survey-based reports contain
establishment information on sex and race/ethnic distributions across major occupational groups. These
data are unique in their ability to answer basic demographic questions about the pace and social location
of change in workplace segregation.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 3/27
We begin with a review of what is already known demographically about desegregation of the workforce
in the United States since 1964. We then use EEOC workplace data to describe national trends in male-
female, black-white, and Hispanic-white employment segregation since 1966.
Next we disaggregate these trends by industrial sector and region. This disaggregation accomplishes two
goals. First, it makes clear for the first time that the pace of equal opportunity advancement in the United
States is highly uneven, and that some social locations show limited social progress since 1966. Second, it
allows us to see demographically whether the progress toward increased occupational similarity by sex,
race, and ethnicity that previous research documents occurred in actual workplaces .
Finally, we include a regression-based disaggregation of time trends in male-female, black-white, and
Hispanic-white segregation. These models allow us to produce estimates of segregation trends within the
workplace , controlling for change in the industrial structure and geographic shifts in economic activity
since 1966.
There is a sustained literature on national trends in occupational sex segregation, less literature on trends
in black-white occupational segregation, and almost no literature on Hispanic-white occupational
segregation. At this point, we know very little about the spatial and industrial distribution of establishment
segregation, and essentially nothing over time. In fact, although speculation abounds, we know very little at
all about workplace segregation (see the review and this conclusion in Reskin, McBrier, and Kmec 1999).
NATIONAL TRENDS IN OCCUPATIONAL SEGREGATION
A comparison of multiple time series on sex segregation shows very little change between 1960 and 1970,
a dramatic decline between 1970 and 1980, and little or no change between 1980 and 1990 (Baunach 2002;
Carlson 1992; Jacobs 1989; Jacobsen 1994). Cotter, Hermsen, and Vanneman (2004) suggest a more nearly
linear decline in occupational sex segregation from 1970 to 1990. Studies of national occupational sex
segregation estimate about a 1 percent per year decline in occupational sex segregation between 1970 and
the mid-1980s (Beller 1985; Jacobs 1989), with slower declines in the 1990s (Cotter et al. 2004; Jacobs
2001).3
There is much less research available on trends in race/ethnic segregation. King's (1992) time series on
black-white occupational segregation is the longest. She estimates trends between 1940 and 1988
separately for men and women, showing that segregation increased for both between 1940 and 1960, then
declined thereafter (see also Fossett, Galle, and Kelly 1986). For women, King estimates black-white
occupational segregation declines from a very high base by 2.9 percent per year between 1960 and 1980.
Among men, black-white occupational segregation declined from a lower base 1.5 percent per year between
1960 and 1980, flattening to a 0.6 percent decline per year between 1980 and 1988 (see also Carlson 1992).
Hispanic-white occupational segregation shows rapid declines of 1 percent per year for men and 2 percent
per year for women across the 1970s, but is essentially flat across the 1980s and early 1990s (Catanzarite
2003).
The national trend data suggest that for all three forms of occupational segregation, the majority of the
observed trend toward integration happened in the 1970s, during the peak period of regulatory
enforcement (Bergmann 1996). For all three comparisons, integration stalled or nearly stalled across the
1980s. Only for sex segregation do we have a time series that extends far into the 1990s, and it suggests
some acceleration in desegregation during the 1990s, after the flat period of the 1980s.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 4/27
SECTORAL AND INDUSTRY SEGREGATION PATTERNS
To our knowledge, no comprehensive studies have estimated trends in occupational sex or race/ethnic
segregation by industry. Stearns and Coleman (1988) computed occupational segregation indices for 22
manufacturing industries in 1973 and 1982. Most industries display little or no change in sex segregation
across the period, although a few show dramatic declines and a few small increases in segregation. This
article suggests that there is considerable variation in sex segregation at the detailed industry level, and
that trends in segregation may be industry specific.
A few studies have examined cross-sectional sectoral and industrial differences in occupational (Lorence
1992) and establishment (Cartwright and Edwards 2002; Petersen and Morgan 1995) sex segregation.
These studies suggest that social services tend to be the most integrated, whereas personal services and
wholesale trade are moderately segregated. Conflicting findings, however, are far more frequent. In the
extractive sector, Lorence finds moderate segregation, whereas Cartwright and Edwards find high
segregation. Cartwright and Edwards find durable manufacturing highly segregated and nondurable
manufacturing only moderately segregated. Petersen and Morgan find both forms of manufacturing to be
highly segregated. These three studies estimate producer services such as banking and insurance to be
highly, moderately, or hardly sex segregated.
To our knowledge, Carrington and Troske (1998) provide the only race segregation industry estimates, yet
their findings conflict depending on the dataset from which they draw. Durable manufacturing is highly
segregated in a dataset of small firms, but moderately segregated in a sample of larger manufacturing
firms. Nondurable manufacturing is exactly the opposite, showing more segregation in the large firm
sample. These conflicting findings suggest variation in segregation within sectors, and possibly within
industries.4
Although little is known precisely about the levels or trends in industry-linked sex or race segregation, it is
clear that the service sector industries have higher proportions of women among their employees (Goldin
1990). Charles and Grusky (2004) see the rise of the service industries as producing sex segregation
through the cultural match between service provision and traditional female tasks (see also Hakim 2000).
They argue that two fundamental tendencies exist in the creation of sex segregation. The first is a vertical
dimension in which men tend to dominate the best jobs. The second is the horizontal sorting of women
into nonmanual and service-oriented positions. Their argument is at odds with the literature reviewed
earlier, which shows lower sex segregation in service-providing industries, but it is consistent with the
higher representation of women in these industries. On a more theoretical level, because service industries
tend to pay lower wages, previous researchers have expected that queuing (Reskin and Roos 1990) and
social closure (Tomaskovic-Devey 1993) pressures may be lower in service industries, leading to both more
female representation and lower segregation within these industries.
Finally, institutional organizational theory points to industries as developing normative expectations
concerning both the division of labor and the sex composition of typical jobs. In this framework, industry
as an institutional field is the normative source of segregation expectations among employers (Beggs
1995; Milkman 1987; Tomaskovic-Devey and Skaggs 1999).
REGIONAL AND LABOR MARKET SEGREGATION PATTERNS
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 5/27
Occupational sex and race segregation also vary across regions (Abrahamson and Sigelman 1987; Fossett
et al. 1986; King 1992; McCall 2001). The South historically has been more racially segregated than the
United States as a whole. Slavery, Jim Crow, and a late transformation from an agricultural to an industrial
economy all may have caused the South to lag behind other regions in decreasing racial segregation.
Stainback, Robinson, and Tomaskovic-Devey (2005) argue, instead, that the relatively strong segregation
decline in the U.S. South after the Civil Rights Act of 1964 reflects the earlier implementation of state-
level EEO laws restricting racial discrimination in employment before 1964 in the non-South. Further
disaggregation suggests that there also may be considerable variation in occupational sex and race
segregation within regions (Abrahamson and Sigelman 1987; Cohen and Huffman 2003; Cotter et al. 1997;
McCall 2001).
The literatures on trends in regional occupational segregation report the greatest decline in race
segregation in non-South regions between 1960 and 1970 and in the South between 1970 and 1980
(Abrahamson and Sigelman 1987; Fossett et al. 1986; King 1992). Looking at changes in occupational sex
segregation across cities, Lorence (1992) finds that growth in the service sector is associated with
declines in occupational sex segregation. He also finds that variation in sex segregation across cities is
strongly tied to variation in the industrial structure. Together, these findings suggest that some part of the
observed declines in employment segregation may reflect change in industrial structure. They also
suggest that regional differences in segregation may reflect differences in industrial structure rather than,
or in addition to, place-specific cultural-historical differences in sex and race/ethnic relations. It is likely
that regions may vary in their normative support for equal opportunity as well (Beggs 1995; Skaggs 2001).
Catanzarite (2000) points out that the uneven distribution of minority groups across geography means that
the likelihood of occupations developing into ethnic-typed employment locations is place specific. She
focuses on Los Angeles, partly because it has the Hispanic concentration necessary for the creation of
"browncollar" jobs. Catanzarite finds that the flow of Hispanic immigrants into Los Angeles during the
1980s led to increases in Hispanic-white occupational segregation.5 This finding is in contrast to the flat
national trend she documents elsewhere (Catanzarite 2003). Catanzarite's research suggests that ethnic
segregation should be more variable across regions than sex segregation, and that we might also expect
the desegregation trajectory to vary across regions because of local factors such as ethnic immigration.
STRENGTHS AND LIMITATIONS OF EEO-1 DATA
Baron and Bielby (1980) instructed us 25 years ago to bring the firm back into inequality research. Their
point was powerful. Most stratification research ignored the organizational context in which inequality was
produced and allocated. Their instruction, however, was underwhelming in its influence, precisely because
so much social and economic data were and still are collected from individuals with little or no regard for
organizational context. Quantitative data collection at the organizational level remains relatively rare, and
almost none is longitudinal (Reskin et al. 1999; see Kmec 2003b and Robinson et al. 2005 for summaries of
the strengths and weakness of extant and potential data collection strategies).
The EEO-1 reports we utilize document workplace -level segregation in the private sector since 1966.
Coverage is currently limited to establishments in private sector firms with 50 or more employees if the
firms are federal contractors and 100 or more employees if the firms are not federal contractors. Before
1983, separate reports were required for contractor firms with 25 or more employees and noncontractor
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 6/27
firms with 50 or more employees. By the year 2003, more than 4.5 million establishment observations had
accumulated, meaning plenty of data exist for disaggregating segregation levels and trends to the
community and industry level.
For each reporting establishment, the EEO-1 data contain sex-specific employment counts for five
racial/ethnic groups (black, Hispanic, Asian/Pacific Islander, Native American/Alaskan Native, and white)
across the following nine occupational categories: officials and managers, professionals, technicians, sales
workers, office and clerical workers, craft workers, operatives, laborers, and service workers.6 This allows
establishment-level estimates of occupational segregation by sex/race/ethnicity as well as more focused
models on access to specific occupational categories. Firms are instructed that employees do not include
temporary or casual workers, but do include leased employees as well as both part-time and full-time
employees.
The EEOC does not audit the accuracy of these reports, but does promise confidentiality. Robinson et al.
(2005) discuss the strengths and limitations of the EEO-1 reports for segregation analyses such as those
undertaken in this article. They conclude that the quality of these data is at least as high as that of U.S.
Census or Current Population Survey-based sources or other organizational surveys.
Bielby and Baron (1984, 1986) demonstrate that job-level sex segregation within establishments is
consistently higher than estimates based on the distribution of men and women across Census
occupational categories. One of the important methodological results of both the occupation-based
studies and the research of Bielby and Baron is that as the level of job or occupational detail increases, so
does observed segregation. This has led to a consensus in the field that more detail in describing the
employment context is always preferable to less. The fact that EEO-1 data describe actual workplaces is a
large advantage compared to occupation based estimates.
The EEOC data allow us to look at occupations, but not job titles, within specific workplaces . The use of
occupational measures in segregation analyses has been criticized because of the loss of detail relative
to job titles (Bielby and Baron 1986; Tomaskovic-Devey 1993). Robinson et al. (2005) show that EEO-1 data
underestimate the degree of segregation relative to job-within-establishment estimates, but produce
similar rank ordering of industries and workplaces . It is clearly worth remembering that these data shed no
light on within-occupation, within-establishment segregation. We still are underestimating actual
segregation.7
It seems to us that measurement error in estimated segregation using these aggregate occupational
categories is likely to be dependent on the degree to which the nine occupational categories mimic the
actual division of labor in a firm. This match will be better for some industries than others. Robinson et al.
(2005) show that the underestimates of job-level segregation in these data do not misrepresent
covariance with firm size or sex/race composition, but are associated with industry. For these reasons, they
conclude that these data are better for analyzing trends in segregation than for comparing cross-sectional
levels, especially across industries.
In the regression models that follow, we adjust for divisions of labor-derived error in the use of
occupational categories by controlling for observed occupational heterogeneity within establishments.
Establishments with lower occupational heterogeneity also will have lower segregation because of likely
greater dissimilarity between the EEO-1 occupational categories and the actual divisions of labor in the
respondent workplace .
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 7/27
Figure 1. Sex, Race, and Ethnic Homogeneous EEO-1
Establishments, 1966-2003
In addition, EEO-1 data are limited by the exclusion of small firms from reporting requirements. Bielby and
Baron (1984) show that sex segregation tends to be higher in smaller workplaces . Tomaskovic-Devey and
Skaggs (1999) show that the association of higher sex segregation with small establishment size is
primarily a function of the higher random probability of segregation as size declines. In the regression
models that follow, we control for establishment size. This also controls for the influence on segregation
trends of size-linked changes in 1983 reporting. Thus, we can with fair confidence describe trends in
workplace desegregation that occur net of size, industrial, and spatial reconfigurations of the sample.
TRENDS IN WORKPLACE SEX AND RACE-ETHNIC SEGREGATION SINCE 1966
The index of dissimilarity and most other summary measures of
segregation have been criticized for being margin dependent (Charles
and Grusky 2004; Grusky and Charles 1998). That is, the level of
segregation measured by D is partially dependent on the occupational
distribution in the observed workplaces . In studies of national
economies, often with an aim to control for national or temporal
differences in occupational structure when comparing national levels
of segregation, margin dependence is a potentially serious problem
(Charles and Grusky 2004). We believe the margin dependence of D
for our samples of workplaces presents a different threat to
inferences. Workplaces with broader and more equal occupational
coverage tend to have higher measured segregation. These
workplaces tend to be those that contain significant employment in
all or most major occupational categories. Thus, workplaces whose
divisions of labor map poorly onto the standard occupational
categories will have lower measured segregation. We see this as a problem of measurement error in
divisions of labor akin to having occupational-level rather than joblevel data. In our formal modeling, we
directly control for occupational heterogeneity at the workplace level to account for this source of
measurement error.
Figure 1 reports 1966-2005 time trends in the percentage
of EEO-1 establishments that are sex homogeneous, that
contain white or black employees but not both (racial
homogeneity), and that contain white or Hispanic workers,
but not both (ethnic homogeneity).8
In 1966, 11 percent of EEO-1-reporting establishments are
sex homogeneous. By 1980, this figure drops to only 1
percent of workplaces , a number that does not change
much since 1980. A much higher proportion of
establishments have no black or Hispanic employees
across the whole period. In 1966, 45.6 percent of all
reporting establishments are racially homogeneous. The
comparable number for Hispanics in 1966 is 71 percent. Thus, we begin the post-Civil Rights Act period
with very high levels of between- workplace segregation. We cannot comment at this point on what
proportion of this exclusion reflects racially or ethnically homogeneous community demography and what
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 8/27
Figure 2. EEO-1 Establishment-Level Dissimilarity
Indices (D) for Hispanic-White, Black-White, and Sex
Segregation, 1966-2003
part represents between-establishment racial segregation within ethnically heterogeneous communities.
The lesson from Figure 1 is that, at least among private sector EEO-1-reporting establishments, sex and
ethnically homogeneous workplaces have almost disappeared since the Civil Rights Act of 1964.
Although not reported in Figure 1, from the beginning of the period, almost no establishments are without
any men or whites. Specifically, in 1966, 1 percent of establishments have no men and 0.5 percent have no
whites. By 2003, only 0.5 percent of establishments have no men, whereas the proportion with no whites
does not change appreciably. By 2003, only 18.7 percent of establishments are without any Hispanic
employees and 13.1 percent of workplaces have no black employees. In 2003, only 5.4 percent of all
reporting establishments have no black or Hispanic employees.
These trends have methodological implications for the
interpretations of the index of dissimilarity in the
following analyses. Because establishment D's exclude all
sex or race/ethnic homogeneous workplaces by definition,
this measure of segregation underestimates total
segregation by excluding between- workplace segregation
from its calculation. Because these types of workplaces
rapidly decline over the period, the underestimation of
real segregation is higher early in the period than later.
This methodological exclusion is more consequential for
the calculation of Hispanic-white segregation, followed by
black-white segregation, and least consequential for
tracking trends in sex employment segregation. In the
regression decomposition that follows, we model this
definitional exclusion as a type of sample selection bias.9
Figure 2 documents trends in the mean level of establishment segregation using the index of
dissimilarity computed for EEO-1-reporting private sector workplaces . Sex segregation declines nearly
linearly from a very high level of 70 in 1966 to 35 in 2003. On the basis of comparisons with estimates of
D using joblevel data within establishments, Robinson et al. (2005) conclude that the use of EEO-1-
estimated D's exceeding 65 is likely to represent nearly total workplace segregation. The EEO-1 estimates
for 1966 also underestimate sex desegregation trends because they exclude 11 percent of establishments
that were sex homogeneous. Therefore, actual desegregation trends are even more dramatic than presented
in Figure 2.
Our estimates show that sex segregation declines by 1.4 percent per year between 1966 and 2003, 1.8
percent between 1966 and 1980, and 1.5 percent between 1980 and 2003. These workplace estimates
suggest a substantially larger and more linear decline in sex segregation than suggested by previous
occupationbased research (except for Cotter et al. 2004). Previous research suggests a 1 percent drop per
year before 1980 and little change thereafter. The EEO-1 reports suggest much more progress toward sex
equality in the workplace than did previous national occupation-based estimates.
The pattern for black-white segregation is more consistent with previous research on occupational
desegregation. We observe a 2.0 percent decline per year in black-white workplace segregation between
1966 and 1980, and a very slow decline (0.6 percent per year) thereafter. Similarly, the available research on
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 9/27
black-white occupational segregation suggests a 2.4 percent decline per year before 1980 and a flatter but
still declining level of segregation after 1980.
The time series for Hispanic-white segregation in Figure 2 suggests an aggregate pattern different from
both male-female and black-white employment segregation. Estimated Hispanic-white segregation is lower
than for the other two in 1966, with an initial D of 47.4. Observed segregation declines 0.8 percent per year
between 1966 and 1980, 0.5 percent per year across the 1980s, and 0.9 percent per year across the 1990s.
Although there appears to be a remarkable convergence in segregation levels, it is important to remember
that there still were many workplaces with no blacks or Hispanics in 2003. Thus, we underestimate
Hispanic-white employment segregation relative to black-white segregation, and both relative to male-
female segregation. On the other hand, measurement error in estimated segregation seems to be more
extreme for sex segregation, leading to larger underestimates.10 Regardless of the exact segregation level,
progress toward sex equality is more dramatic and more sustained than race/ethnic desegregation since
the Civil Rights Act of 1964. After an initial burst of desegregation along race/ethnic lines, workplace
segregation stalled at essentially 1980 levels for black-white and Hispanic-white segregation.
There is substantial establishment-level variation in observed segregation for every year observed. These
national trends may reflect desegregation in workplaces as a result of civil rights legislation and
enforcement as well as changes in the spatial or sectoral distribution of employment. We examine these
aspects of establishment location in the remainder of this article. We see the following analyses as
providing the first systematic description of workplace heterogeneity in the desegregation process.
TRENDS WITHIN MAJOR INDUSTRY SECTORS AND REGIONS
We examine segregation levels and trends for 11 industrial sectors: agriculture, construction, mining,
nondurable manufacturing, durable manufacturing, transportation-communication-utilities, wholesale trade,
retail trade, producer services, social services, and personal services. Each sector shows a decline in sex
segregation since 1966, but these lines are not strictly parallel." The decline in the construction sector
does not begin until after 1975 and stalls around 1990. Durable manufacturing and producer services show
steady declines in sex segregation, whereas transportation, communication, utilities, and nondurable
manufacturing do not begin to decline until after 1971 and 1975, respectively. The three sectors with the
lowest levels of sex segregation (retail trade, social services, and personal services) also show the least
dramatic change. In general, sex segregation is lower in the emerging service sector (retail trade, producer
services, social services, and personal services) than in the traditional production sectors of the
economy.
There is much less sectoral variability in black-white employment segregation than in sex segregation.
However, the sectoral trajectories are discontinuous compared with the nearly linear declines in sex
segregation across sectors. There are strong declines in black-white segregation beginning in 1966, but in
all sectors, these declines stall in the 1980s. Mining, construction, durable manufacturing, and nondurable
manufacturing appear to be resegregating after 1995. Again, in general, race segregation is lower in the
service sector, although social service race segregation is higher than race segregation in the other service
sectors.
There is very little change in Hispanic-white segregation in the traditional sectors (agriculture,
construction, mining, nondurable manufacturing, durable manufacturing, transportationcommunication-
utilities, and wholesale trade) of the economy since 1966, with only construction showing substantial
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 10/27
declines in segregation. Agriculture has the highest level of Hispanic-white segregation in 2003, with a
substantial increase since 1966. Similar to the pattern for black-white segregation, construction, wholesale
trade, durable and nondurable manufacturing, and social services all ended the period with relatively high
levels of Hispanic-white segregation. In 1966, Hispanic-white segregation is lowest in producer services,
but actually increases through 1990. For most of the period, Hispanic-white segregation is lowest in retail
trade, as it is for sex and black-white segregation, for which it shows a decline across the entire post-1971
period.
An examination of 2003 segregation levels across sectors shows some consistent patterns. Sex and
race/ethnic segregation is lowest in retail trade. Wholesale trade and both durable and nondurable
manufacturing tend to be relatively highly segregated across all three comparisons. Race/ethnic
segregation is highest in agriculture, whereas sex segregation is highest in construction and mining. Retail
trade, personal services, and producer services tend to be less segregated in all three status distinctions
than most goods-producing sectors. Transportation, utilities, and communication tends to have low levels
of race/ethnic segregation, but high levels of sex segregation. Conversely, social services tends to have
low levels of sex segregation and higher levels of race/ethnic segregation. The economy-wide shift toward
low-segregation services likely produced some of the desegregation trends we observe in U.S. society
since the Civil Rights Act. We model this source of change in the regression analyses that follow.
We also examine trends within census regions.12 All regions show similar linear declines in sex
segregation between 1966 and 2003, and there are no striking regional distinctions in the degree of sex
segregation. In all regions, black-white segregation drops dramatically between 1966 and 1980, then hardly
at all. Black-white segregation drops most dramatically in the southern regions. In fact, by 2003 the East
South Central and South Atlantic regions have the lowest levels of black-white employment segregation.
Mid-Atlantic and Northeast states have the lowest levels of black-white segregation in 1966, but show the
least change over the period. Hispanic-white segregation drops modestly in most regions, at least through
1980. It is highest in the East South Central region across the period, but also shows strong declines
across the entire period in that region. The Mountain states and the Pacific region have the lowest levels
of Hispanic-white segregation across the period, but they decline more modestly and turn essentially flat
after 1985 and 1980, respectively.
MODELS AND MEASURES
We use regression analyses to estimate national time trends, controlling for sample selection, measurement
error, and industrial and geographic shifts in employment, leaving a residual time trend, which we interpret
as a within-organization rather than a between-organization change in segregation.13
There is substantial evidence in the descriptive material to suggest that the general sectoral shift in the
economy toward services and away from manufacturing and extractive sectors will be a significant source
of the observed desegregation in U.S. workplaces . Although we do not find dramatic regional differences
in employment trends in the descriptive analyses, it is possible that if we examine shifts in employment
at the local labor market level we will find that some of the observed desegregation in U.S. workplaces
since 1966 represents the movement of employment toward less segregated labor markets. Our basic
strategy is to regress observed D 's on time and then control for sectoral and regional shifts in employment,
with fixed effects for industry and local labor market. We report decompositions of time trends based on
three-digit industry and local labor market fixed effects.14
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 11/27
We see three potential sources of error in the observed desegregation trends: sample selection,
occupation-based underestimates of job segregation, and variation across firms in the appropriateness of
the nine EEO-1 occupational categories.
SAMPLE SELECTION PROBABILITY. First, D, by definition, excludes workplaces that are homogeneous for
any comparison category. For allmale (female) or all-white (black, Hispanic) workplaces , D is undefined.
Thus, we underestimate true segregation with D, especially for the early years in which a higher proportion
of workplaces are sex or ethnically homogeneous. Because this is a form of selection bias in our
estimates, we first compute the probability of a workplace being homogeneous as a function of
employment size, year, industry, region, and estimated random probability of workplace homogeneity.
These logistic regressions, interesting in their own right, are discussed in the next section. We use these
logistic regressions to produce an estimated probability of exclusion from the sample and use that as a
control variable in the decomposition regressions. When estimating the true intercept for desegregation
trends (i.e., estimated D in 1966), we do so when the probability of exclusion from the sample is zero.15
OCCUPATIONAL HETEROGENEITY INDEX. Measurement error in estimated segregation also is dependent
on the degree to which the nine occupational categories used in reporting to the EEOC mimic actual firm-
level divisions of labor. Firms use job titles, not occupational groups, to make internal social and task
distinctions (Strang and Baron 1990). The EEOC data capture workplace -level, between-occupation
segregation, but miss within- workplace , within-occupation job segregation. Thus, the EEOC data will
always underestimate true segregation, but that measurement error will vary across workplaces . We reason
that this measurement error is higher when workers are observed in fewer of the nine EEOC occupational
categories. Establishments with low occupational heterogeneity also will have low segregation because of
an increased disjunction between the EEOC occupational categories and actual divisions of labor. An
occupationalIy heterogeneous workplace may have substantial employment in all nine occupational
categories. As such, it has an increased chance of displaying high levels of segregation because there are
more positions across which people can be distributed. A firm with all its employment in only one
occupational category will have no observed segregation in the EEO-1 data. In the real world, however,
this firm might make numerous job distinctions within that one occupational category and thus have high
segregation in practice.
ESTABLISHMENT SIZE (LN). The EEO-1 reports, because they use
occupational groups within establishments rather than job titles to
collect status distribution information, tend to underestimate true job-
level segregation in the workplace , and this error rises
logarithmically with establishment size (Robinson et al. 2005). This
underestimation of segregation that arises from using occupational
groups rather than job titles will be greater the larger the workplace .
This is a potential source of error in our desegregation trend analyses
because in 1983, the size rules for EEO-1 data collection change, and
establishments with fewer than 100 employees (50 in the case of a
federal contractor) no longer report. There is also a secular decline in average establishment size after
1985. We use the natural log of total employment size as a control for measurement error, reporting
change and compositional change, and estimate the 1966 intercept for desegregation trends (i.e.,
estimated D in 1966) with ( In ) size at its 1966 mean.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 12/27
Reskin et al. (1999) conclude that establishment size is positively
correlated with the employment of women and racial minorities. They
suggest that larger establishments can make room for women and
racial minorities without directly challenging the jobs reserved for
privileged groups. A positive relationship also exists between
organizational size and the development of personnel offices,
formalization, and job differentiation (Konrad and Linnehan 1995).
Other research shows that formalization reduces sex segregation
(Tomaskovic-Devey 1993; Tomaskovic-Devey, Kalleberg, and Marsden
1996; Tomaskovic-Devey and Skaggs 1999) and the influence of sex
composition on the pay gap (Anderson and Tomaskovic-Devey 1995;
Elvira and Graham 2002; Pfeffer and Cohen 1984), whereas it increases
women's access to managerial and supervisory jobs (Hultin and Szulkin 1999; Reskin and McBrier 2000).
Net of formalization, Tomaskovic-Devey et al. (1996) find that there is an additional decrease in sex
segregation in very large workplaces . They suggest that this may represent the enhanced visibility of
large firms. Job title proliferation in large firms also may increase segregation (Baron and Bielby 1986;
Strang and Baron 1990), but it is precisely this type of segregation that the EEO-1 reports cannot detect.
Thus, although we use size primarily as a control variable, we also suspect that larger size may reduce
segregation for substantive reasons around formalization and visibility.
Our regression models for changes in segregation begin with a model that regresses D on year. The second
model controls for the three sources of measurement error outlined earlier: sample selection, occupational
heterogeneity, and establishment size. The final model adds fixed effects for three-digit industry and local
labor market.
Random sampling from a large database such as the EEO-1 reports is a practical means of handling its
magnitude and does not reduce the reliability of the results. For this reason, we took a 20 percent sample
for 1966 and 1971, then in 5-year increments for 1975 to 2000, and finally for 2003 ( the last year of
available data). This led to a sample of 276,749 establishments. Table 1 shows the first stage of our
analysis, in which we focus on sample selection issues. We perform dichotomous logistic regression with
three dependent variables, one for each type of establishment homogeneity (no females or males, no
blacks or whites, no Hispanics or whites).
ODDS of ESTABLISHMENT SEX AND RACE/ETHNIC HOMOGENEITY
Table 1 reports the logistic regression of sex and race/ethnic homogeneity upon time, with controls for
measurement error, industry, and region. We also include a measure for the random probability of
homogeneity associated with the size of the organization.17 We first summarize the effects of the
control variables for all models. Very small organizations are more likely to be sex and race/ethnic
homogeneous. A one-unit increase in the random integration ratio reduces the odds of sex homogeneity
by 3 percent and race/ethnic homogeneity by 4 percent. Occupational heterogeneity has essentially no
influence on race/ethnic homogeneity, but a one-unit increase in occupational heterogeneity is associated
with a 3 percent decrease in the odds of sex homogeneity. The EEOC occupational categories are more
closely tied to between-firm sex segregation than they are to race/ethnic segregation. Increases in
establishment size dramatically reduce the odds of sex and race/ethnic homogeneity.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 13/27
Table 1. Sample Selection (Logistic Regression)
Models of Establishment Homogeneity (Odds
Ratios Reported)
The first column of Table 1 shows that the odds of sex
homogeneity decrease drastically with time. The odds of
an establishment being sex homogeneous in 2003 are only
10 percent of what they were in 1966, the baseline year.
Sex homogeneity effects associated with industry fall into
two broad categories. Relative to durable manufacturing,
we see higher odds of sex homogeneity among old
economy industries (mining, construction, transportation,
communication, and utilities) and lower odds of
homogeneity among new economy industries (wholesale
trade, retail trade, producer services, social services,
personal services). Such effects are not surprising given
the traditionally male-dominated nature of the
occupations in the older industries. Regional effects on
sex homogeneity are positive, indicating that the odds of
establishment sex homogeneity are higher in all regions
than in the Pacific region. In New England and the
Mountain region, these differences are not statistically
significant. The mid-Atlantic and East and West South Central regions stand out as having much higher
odds of containing sex homogeneous establishments than other regions.
Black-white homogeneity also declines over time, but not as dramatically as sex homogeneity, dropping to
33 percent of the 1966 odds by the year 2003. Relative to durable manufacturing, only agriculture and
mining have higher black-white establishment homogeneity. The lowest incidence of racial homogeneity is
found in the personal service sector (.378), followed by the transportation and communication sectors, in
which the average odds of black-white homogeneity are less than half as great as in durable
manufacturing. There is no clear correspondence between old and new economy sectors in the incidence
of black-white establishment homogeneity. Rather, all sectors have lower homogeneity than agriculture and
mining, and the two manufacturing sectors have higher homogeneity than the remaining sectors. The
largest contrast is in the regional effects. The odds of black-white homogeneity are about two times
those of the Pacific region in New England (1.675), the West North Central region (2.164), and the
Mountain (2.272) region, whereas the same odds are substantially lower than the those of the Pacific
region in the South Atlantic (.230), East South Central (.318), and West South Central (.416) regions. These
results reflect, at least in part, the geographic distribution of the black population in the United States.
In the last column, Hispanic-white homogeneity is modeled. The time trend again is similar to that for the
other two types of homogeneity. By the year 2003, the odds of Hispanic-white homogeneity are 9 percent
of their 1966 level. This reflects the strong growth in the size of the Hispanic workforce as well as any
desegregation in hiring patterns. Industry effects are generally weak, with only mining displaying markedly
higher odds of Hispanic-white homogeneity than the other sectors. Personal service shows the lowest
odds of Hispanic-white homogeneity, as it did for both sex and black-white comparisons as well. The
regional effects, however, are substantial and quite a bit stronger than in the other two models.
Establishments in regions other than the Mountain region have much higher odds (2.559-46.357) of
Hispanic-white homogeneity than the Pacific region. This makes sense for the regions that have few
Hispanic residents, such as New England. However, it is surprising that the odds of homogeneity are
highest in one area with many Hispanics residents: the East South Central region. Here homogeneity likely
is not being driven by a lack of Hispanic workers, but by some kind of exclusion.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 14/27
Table 2. Regression Models of Establishment Sex,
Race and Ethnic Segregation
In the next step of our analysis, we take the predicted probability of establishment homogeneity and
include it as a control variable in the models of establishment segregation. This will control for any
sample selection bias attributable to undefined D values.
REGRESSION DECOMPOSITION OF DESEGREGATION
Table 2 shows the regression decompositions of establishment segregation. The first model includes only
the time trend. The second model includes the controls for measurement error in D and the sample
selection probability. In the third model, we add three-digit industry and commuting zone fixed effects.
From 1966 to 2003, observed sex segregation (Model 1)
drops 35.0 points. Remarkably, 16 percent of the total
variation in sex segregation is accounted for simply by
time. There is a dramatic and widespread decline in sex
segregation after 1966. Sex segregation adjusted for
temporal shifts in industry, local labor market employment,
sample selection, and measurement error drops 22.9 points
(Model 3). By comparing the year coefficients in Models 1
and 3, we can see that about one third of the observed
decrease in sex segregation over time is attributable to
these shifts in the employment composition of the
private sector economy.
Establishments with high predicted probabilities of being
sex homogeneous have higher observed sex segregation
(Model 2), but lower estimated sex segregation net of
industry (Model 3), yet this latter effect is substantively
small (beta = -.06 in Model 3). Thus, both sex homogeneity
and sex segregation are strongly tied to industry. When
occupational heterogeneity rises one unit, sex segregation
increases about half a unit (.55). As organizational size
increases, observed sex segregation decreases as well.
Black-white segregation exhibits a similar pattern in some respects. Observed black-white segregation
drops 19.2 points between 1966 and 2003. After adjusting for temporal shifts in sources of sample section,
measurement error, industry, and local labor markets, we observe a decline of 12.9 points. Again, about one
third of the decrease in segregation over time is attributable to compositional shifts in the economy. The
difference here relative to the sex segregation estimates is that relatively little unique explanatory power is
added by industry and local labor market fixed effects. Black-white segregation rises slightly with an
increased sample selection probability. As with the models for sex segregation, black-white segregation
declines as establishment size rises and increases with increased occupational heterogeneity.
For Hispanic-white segregation, the picture changes still more. From 1966 to 2003, we observe an 11.4
point drop in Hispanic-white segregation. After adjustment for sample selection and measurement error
(Model 2), there is essentially no change in Hispanic-white segregation between 1966 and 2003. This
reflects the strong decrease in workplaces with no Hispanics after 1966. That is, increased Hispanic
representation in the labor force has been particularly influential in producing observed desegregation
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 15/27
Table 3. Decomposition of 1966-2003
Desegregation into Observed Change and Estimated
Within Workplace Change
trends (Figure 2). After control is imposed for local labor market changes in the private sector economy,
the decline in Hispanic-white segregation is 7.0 points. This 7.0-point decline happens entirely within local
labor markets. Hispanic-white segregation declines are profoundly local. 18
A consistent finding across these models is that larger establishments have lower segregation of all three
types. Another consistent finding is that the addition of the measurement error controls explains a large
percentage of the crosssectional and temporal variation in segregation. In models not shown here, we
find that this is attributable primarily to the introduction of the occupational heterogeneity control
variable. This variable is strongly associated with industry because establishments in the same industry
tend to have similar divisions of labor.19 Thus, shifts in industrial composition influence desegregation
trends for two reasons: the move to services is a move to both less segregated workplaces and
workplaces whose internal divisions of labor are less well described by the EEO-1 occupational
categories.20
Table 3 shows the results of these regression estimates used to create period-specific trends in observed
desegregation and in estimated within- workplace segregation. Observed segregation change is based on
Model 1 year coefficients from Table 2. The estimate of within- workplace change uses year coefficients
from Model 3. Estimated change in workplace organizational practices- the intended goal of Title VII of
the 1964 Civil Rights Act-is probably the better estimate of the influence that this legislation and the
associated Civil Rights movement pressure had on workplace desegregation. All three comparisons show
significant real (i.e., residual) change in employment segregation. The estimate of total observed change
between 1966 and 2003 is much higher for sex segregation (averaging a drop of 1.07 percent per year) than
for either black-white (0.77 percent per year) or Hispanic-white (0.55 percent per year) segregation.
Hispanic-white workplace employment segregation changes the least. Our estimates of within- workplace
change are smaller, reflecting the influence of industrial and other economy shifts in desegregation
trends.
These desegregation trajectories are, however, highly
variable by period. In the initial Civil Rights period (1966-
1971), black-white segregation declines 2 percent per year,
and almost all that decline happens within workplaces (87
percent) rather than as a function of compositional change
(13.0 percent). During this same period, there is a strong
1.4 percent per year decline in sex segregation, but it is
largely (67.2 percent) a function of compositional change
in EEOC-reporting establishments. This early period shows
the slowest rate of Hispanic-white desegregation (0.49
percent a year), although most of that (96.4 percent)
happens within workplaces .
Across the 1970s, all three forms of segregation show
strong desegregation tendencies, and the vast majority of these shifts happen within workplaces . Our
estimate of withinworkplace desegregation is 1.1 percent per year for both sex and black-white segregation
in the 1970s. Hispanic-white within- workplace segregation declines at half that rate (0.60 percent per
year), but this decade shows the strongest Hispanic-white desegregation of any period.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 16/27
There is a slowing down of sex desegregation across the 1980s and since 1990, but observed
desegregation remains above 1 percent a year, although about one third of that in both periods is produced
by composition shifts in private sector firms reporting to the EEOC. In the 1980s and after 1990, black-
white desegregation slows considerably, and what does occur is almost entirely produced by compositional
shifts in the EEOC-reporting private sector. Hispanic-white desegregation also stalls after 1980, although
there is some evidence of increased within- workplace desegregation after 1990.
Figure 3 displays our estimated time trends in EEO-1-based segregation for both observed and within-
workplace change. To produce these estimates, we adjust the 1966 intercepts to take into account sample
selection and measurement error. Specifically, we calculate the 1966 intercepts in this figure from the
Model 2 intercept, plus the predicted D when the Gibbs-Martin index is at its observed maximum (89) (and
presumably measurement error in the use of EEO-1 occupational categories is at its minimum), average
1966 establishment size, and when sample selection is zero. We treat the 1966 intercepts as our best
guess as to the actual average level of segregation in 1966. We suspect that they still are underestimates
of the actual segregation level.21 The observed line is calculated as the Model 1 year coefficients
subtracted from this 1966 intercept estimate. The adjusted line is calculated in the same way using the
Model 3 year coefficients.
These estimates are quite different from the observed trends in Figure 2. The 1966 intercepts are
uniformly higher, primarily reflecting the adjustment for underestimated segregation using the EEO-1
occupational categories and the sample selection exclusion associated with homogeneous organizations.
Substantively, for all three demographic comparisons, the slopes of the segregation indices are less
dramatic after adjustment for compositional shifts in the sample.
Although it has the most dramatic downward trajectory, observed male-female segregation is the highest
of the three across the entire period. If there had been no compositional changes in the economy
between 1966 and 2003, sex segregation still would have remained very high, with an estimated D of 65. If
we focus on the adjusted sex segregation line, it is clear that there was very little workplace change before
1971, a dramatic within- workplace decline in segregation through 1980, and a more modest, but still
sustained, drop after 1980.
Although high, adjusted black-white segregation is lower than sex segregation across the entire period and
shows almost its entire decline before 1980. Across the 1970s, the average rate of decline for the
adjusted estimate is 1.4 percent per year, dropping to a negligible 0.14 percent decline in the 1980s and
0.02 percent after 1990. The adjusted line is much more flat than the observed change, suggesting that
most white-black desegregation since 1980 was produced by shifts in the composition of the private
sector EEOC-reporting economy rather than actual workplace change.
Hispanic-white segregation is the lowest of the three across the whole period, although it also displays
the least decline. After adjustment for sectoral and labor market shifts in the economy, there is very little
workplace change in Hispanic-white segregation after 1980.
CONCLUSIONS
The academic and policy communities have had some well-founded reservations about our knowledge of
desegregation in U.S. workplaces . We have known for two decades that occupational data were potentially
misleading, but have had no alternative for estimating changes in status distinctions in employment. The
prior literature on occupational segregation suggests that all three forms of segregation (male-female,
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 17/27
Figure 3. Observed and Adjusted (On the Basis of
Model 3, Table 2) Trends in Male-Female, Black-
White, and Hispanic-White Workplace Segregation
black-white, and Hispanic-white) declined through about 1980 and were relatively flat thereafter. Our
estimates allow us to speak more confidently about actual workplace desegregation, at least in private
sector workplaces that report to the U.S. Equal Employment Opportunity Commission.
Sex segregation declines strongly after 1966, although before 1971, almost the entire decline is created by
compositional shifts in the economy. Both observed and estimated within- workplace sex segregation
declines dramatically and consistently after 1971, although there is some slowing down after 1980. This
decline is robust, happening in all regions and sectors. To the extent that there has been widespread
national progress in equal employment opportunity, it seems to be most clear for sex desegregation. Of
course, sex segregation starts at very high levels, so there was considerable room for improvement. In
addition, the shift to a service economy produces a substantial portion of the observed decline in sex
segregation.
Charles and Grusky (2004) analyzing cross-national and
cross-temporal occupation by sex contingency tables
conclude that there are horizontal and vertical dimensions
to sex segregation. The horizontal dimension is the
tendency of women to be overrepresented in nonmanual
service occupations, whereas the vertical dimension is the
tendency of men to get the better jobs within nonmanual
and manual occupational structures. Although we did not
investigate their models directly, our results suggest that
the growth of service industries actually has reduced sex
segregation because they are internally less segregated
than other industries.
Black-white desegregation begins earlier than sex or
Hispanic-white desegregation, but is uneven across regions
and industrial sectors. Most strikingly, black-white
workplace desegregation essentially stops after 1980. The
observation in the occupation-based literature that equal
opportunity progress stalls at about 1980 is generally
supported, although there is also some disturbing evidence
of resegregation after 1995 in old economy sectors
(mining, construction, durable and nondurable
manufacturing). We think there is still much to be learned about the temporal, industrial, and regional
context of black-white workplace segregation.
The shift to a service economy reduces black-white segregation only marginally. Because the most
integrated sector in 2003 is retail trade, this desegregation shift is not likely to produce strong earnings
growth for the African American population. There is a clear irony in the result that integration is strongest
in low-wage sectors of the economy. Of course, this is exactly the result predicted by queuing (Reskin
and Roos 1990) and social closure theories (Tomaskovic-Devey 1993). Pressures for exclusion should be
weaker where jobs are less desirable.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 18/27
The time trends in Hispanic-white segregation are much less dramatic, even in the peak enforcement
years of the 1970s. National Hispanic-white desegregation essentially stops during the 1980s, although it
appears to pick up again after 1990. Across sectors and regions, Hispanic desegregation is even more
uneven than black-white desegregation. Hispanic-white segregation shows very little change in the
traditional sectors (agriculture, construction, mining, nondurable manufacturing, durable manufacturing,
transportation-communication-utilities, and wholesale trade) of the economy since 1966, with only
construction showing substantial declines in segregation. Hispanic-white segregation actually increases in
agriculture across the post-Civil Rights period. Because Hispanic rights were never the intention of EEO
laws nor a strong component of the Civil Rights movement, weaker shifts are not surprising. At the same
time, the rapid and regionally specific growth in the Hispanic population probably has created new
opportunities for employment segregation in many places and industries.
Spatial variation in ethnic groups was much more consequential for predicting race and ethnic
homogeneity in workplaces than it was for predicting segregation levels within them. We believe that we
have only just scratched the surface of the spatial dynamics of segregation. Future research might pay
particular attention to the influence of both the size and growth of different demographic groups on
workplace segregation and inequality. McCall (2001) reports that sex and race segregation varies locally in
a complex web of class and status processes. Her work suggests that future analyses of local dynamics
need to treat these various labor forces as linked in the local economy through processes of competition,
complementarity, or both.
Regional differences in segregation are apparent, but not as dramatic as sectoral differences. Only for
Hispanic-white segregation do regional shifts in employment help explain much of the temporal change
in segregation. In fact, after adjustment for sample selection and measurement error, there is essentially
no national desegregation pattern for Hispanic-white employment. When labor market fixed effects are
added in the subsequent model, the modest decline we report in Table 3 and Figure 3 becomes apparent.
This suggests that Hispanic-white desegregation has occurred within local labor markets, but that Hispanic
employment is rising in workplaces with relatively high levels of segregation.
The weak trends in Hispanic-white segregation are not surprising. Although Hispanics were legally granted
the same rights as women and African Americans under the 1964 Civil Rights Act, they are socially
different in many ways. Being Latino is potentially associated with multiple cultural markers: immigrant,
illegal, non-English-speaking, and ethnic. In addition, the Latino population is growing and diffusing across
the United States, leading to many more opportunities for new contact with non-Latino populations and
new patterns of incorporation into the economy. Although we are convinced that it is appropriate to begin
our understanding of workplace segregation and desegregation by looking at trends since the Civil Rights
Act of 1964, there are many reasons to regard that political and legal moment as much less consequential
for Hispanic Americans than for women and African Americans.
Understanding Latino segregation shifts probably requires a unique set of explanations, perhaps along the
lines suggested by Catanzarite (2000, 2003), that focus on local immigration flows and the resulting
tensions between new and old status groups.
The contrast between sex and black-white segregation trends is instructive as well. Although African
Americans were the primary target of the 1964 Civil Rights Act, women may have been the primary
beneficiaries. The strong decline in both race homogeneous workplaces and black-white segregation,
starting in 1966, suggests real Civil Rights era progress toward more equal race relations in workplaces .
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 19/27
That progress stopped about 1980 suggests that African American progress was predicated upon political
action, which at least at the national level stalled with the election of Ronald Reagan in 1980. In another
article, we have shown that black-white desegregation is strongly tied to political era; stalling after Ronald
Reagan became president, federal enforcement efforts waned, and legislative initiates stopped (Stainback
et al. 2005). Political and legal pressure for gender equality, which did not start effectively until the 1970s,
continued through the 1990s, as did sex desegregation (Stainback, 2006). Continued convergence in
education and labor force experience between men and women may play some role as well. We see this
article as far from definitive in this regard, and think that there is room for much research on the political
and labor market context of workplace desegregation for all groups.
Our use of EEO-1 establishment reports allows us to say with greater confidence that real workplace
desegregation has occurred since 1966. The regression models suggest that contemporary segregation
still is quite high as well. This article also suggests that sex segregation has declined strongly in most
contexts, but that race-ethnic desegregation is temporally, spatially, and industrially uneven.
It is important to remember that all the estimates in this article are limited to the part of the private
sector economy that is required to report to the EEOC. These are firms with at least 100 employees (50
employees if they are federal contractors). They are essentially the population of private sector
organizations that confront a federal equal opportunity regulatory environment. Because prior research
shows higher sex and race homogeneity and segregation in smaller workplaces , this regulatory selection
means that our estimates are likely lower than the true scores for the entire private sector economy
because small firms are excluded. Desegregation in the entire private sector is likely to be lower than it is
among EEO-reporting firms both because excluded firms are not in the regulatory environment and
because they are likely to be smaller.
In 2003, employment segregation remains quite high for all groups. After adjustment for measurement error
and sample selection, sex segregation has an index of dissimilarity estimated to be 53.3 in 2003. Because
this still probably is an underestimate, since job title detail is lacking, we can be confident that well more
than half of all workers would have to switch jobs to create a sex-neutral employment distribution. Black-
white segregation in 2003 is only marginally lower, with an estimated D of 48.4, Again, given the
underestimate of segregation from the lack of job title detail, more than half of all African Americans
probably would need to switch jobs to create completely desegregated private sector employment.
Hispanic-white segregation is only marginally lower in 2003 (D = 44.5).
The availability of EEO-1 reports on private sector establishment level of employment since 1966 offers
the research community many avenues for expansion on this article. Descriptive work can be extended to
include other ethnic groups and sex within race comparisons; distribution of groups to specific desirable
occupational destinations such as manager, professional, or craft work; and more specific industries or
places. These data are ideally suited for researchers' use to explore more fully the contours of employment
segregation within industrial, spatial, organizational, and temporal contexts. We suspect that explanatory
models used to develop the implications of horizontal versus vertical segregation mechanisms,
institutional fields, community norms and politics, and firm practices and environments for equal
opportunity dynamics all are promising avenues for future theorizing and research with these data. They
clearly are promising for the field in general as well.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 20/27
In a real sense, this article has answered only a basic question: Has there been any workplace progress in
segregation since 1966? Although the answer is affirmative, the results are far from reassuring. The Civil
Rights Act was aimed at improving the status of African Americans in the United States, but it was far
from uniform in its influence. After 1980, it seems to have led to no new improvements. Women have
benefited more continuously. We clearly need to develop theories and models that examine the similarities
and differences in these two trajectories. Finally, Latino progress is weak and not strongly tied to the
temporal pattern of Civil Rights legislation. One of the disturbing implications of this article is that
aggregate race/ethnic equal opportunity employment progress in the United States stopped a quarter of a
century ago, after less than a quarter century of progress.
One message from this article for those still waiting for equal opportunity to diffuse across the United
States is that unless substantial shifts occur in political pressure and employment practices, future
progress is unlikely. The conclusion for sex desegregation is clearly more optimistic because the
combination of real workplace desegregation and sectoral shifts has continued to produce declines
though 2003. Of course, continued integration must be produced by something, and until we know why
gender segregation has continued to decline since the stalling of race/ethnic progress, we cannot assume
this trend will continue. In the absence of additional political pressure for equal opportunity in U.S.
workplaces , it is quite possible that desegregation for all groups will converge toward current levels and
then remain stable. More than 40 years after the Civil Rights Act, it is clear that the march toward
employment equality is far from over.
Footnote
1 Before the year 2000, these data were used only occasionally by academics (Adams 1972; Becker 1980:
Leonard 1984; Smith and Welch 1984). To achieve access to these data, the first author became an unpaid
employee of the U.S. Equal Employment Opportunity Commission. We anticipate that these data will
become a major resource for the academic community in the future.
2 These represent all reports from all years since 1966 that currently are available in machine-readable form
at the EEOC. Data are missing for the years 1967-1970, 1974, and 1976-1977.
3 Although the index of dissimilarity (D) is computationally identical across studies, levels of D are
strongly influenced by the number of categories available. Thus, we summarize trends in terms of
percentage change in D rather than changes in absolute values.
4 Sectors refer to aggregations of similar industries, such as durable manufacturing, whereas industry
refers to specific product groups such as airplane manufacture.
5 See also Jacobs and Blair-Loy (1996) for a discussion of occupational segregation by race in local labor
markets.
6 A simpler set of ethnic distinctions is used in Hawaii. In the analyses that follow, we exclude
establishments in that state. We also exclude establishments in Alaska because they are few and thus
cannot be grouped into meaningful labor market areas. We do not analyze all ethnic distinctions because of
space and the lack of prior literatures. Clearly, much more should and can be done with these data.
7 Robinson et al. (2005) find that EEO-1-based estimates of segregation produce absolute values on the
index of dissimilarity of about the same level as those produced by detailed (n = 360) occupationbased
analyses. Thus, the loss of detail produced by using only nine occupational categories is roughly
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 21/27
counterbalanced by the gain in detail attained from having workplace -level data.
8 The dramatic drop in homogeneous workplaces between 1980 and 1985 reflects a change in reporting
requirements instituted by the Reagan administration. After 1982, only firms with 100 or more employees
that were not federal contractors reported to the EEOC, whereas before 1982, smaller firms with 50 to 100
employees also reported. Smaller workplaces are more likely to be racially homogeneous. In the
multivariate analysis, the inclusion of a measure for establishment size effectively controls for this change
in reporting coverage.
9 We did consider simply setting D to be 100 in cases of sex or ethnic homogeneous establishments. This
approach makes some sense for sex segregation, under the assumption that all local labor markets across
all periods have at least some women who might be employed in reporting firms. It does not make sense
for either black-white or Hispanic-white segregation because in all periods there are some local labor
markets with no or almost no African Americans or Hispanics in the EEO-1 labor force.
10 In the regression-based decomposition models that follow, we adjust for measurement error in D
associated with the EEO-1 occupational categories. Estimates adjusted for selection and measurement
error presented in Figure 3 suggest that sex segregation remains the highest across the entire period and
Hispanic-white segregation the lowest.
11 See Supplemental Table 1 on the ASR Web site for estimated sector segregation levels and trends.
12 Supplemental Table 2 on the ASR Web site provides these estimates.
13 Strictly speaking, these are estimates of withinindustry, within-place change controlled for measurement
error and size shifts in the composition of reporting establishments, and thus also could represent the
growth of new organizations in local labor markets within industries. Even in such cases, we treat these
changes in local organizational populations as true organizational change. Future research with these data
might examine how much change results from new divisions of labor within existing organizations, as
compared with different divisions of labor in new firms.
Footnote
14 We use 1990 commuting zones as described in Tolbert and Sizer (1996). Commuting zones are
aggregations of contiguous counties defined in terms of county-to-county flows of residence to
workplace commuting. Thus, they describe labor markets.
15 Descriptive statistics for the full sample are presented in Supplemental Table 3 on the ASR Web site.
16 The average observed occupational heterogeneity drops from 59 in 1966 to 53 in 2003, reflecting the
increasing mismatch between an occupational scheme designed during an industrial era and an
increasingly service-dominated economy.
17 The smaller the workplace , the higher the level of expected segregation (Mayhew 1984). Firms with
fewer incumbents have a higher baseline probability of segregation, even in cases that involve no group
preferences in assignment. To account for random sources of segregation, we include a measure we call
the random integration ratio (RIR), which measures the baseline probability that a workplace is integrated:
RIR = 100*[(N-1)/(N+1)] (for further discussion, see Tomaskovic-Devey and Skaggs 1999).
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 22/27
18 Although Model 3 includes fixed effects for both industry and local labor market, supplementary
analyses demonstrate that this entire effect is a function of the labor market fixed effect.
19 When we regress a fixed effect for industry upon occupational heterogeneity, we get an R^sup 2^ of .39
across the entire period and an R^sup 2^ of .41 for a single year (2000). The industry fixed effect
contributes more to the total explained variance than any measurement error variable. The Gibbs-Martin
index of heterogeneity (H) was the most influential of the measurement error variables.
20 For example, in 2000, the mean occupational heterogeneity in durable manufacturing is 64.1, whereas
in retail it is only 38.7. The mean occupational heterogeneity in 1966 is 59, but drops to 53 by 2003 as the
economy becomes increasingly service oriented.
21 We base this conclusion on comparisons with Bielby and Baron (1986), who observe jobs within
workplace D levels of sex segregation to be about 94 for circa 1970, 6 points higher than our measurement
error-adjusted 1966 estimate. We have no job level estimates of black-white or Hispanic-white segregation
for that period, and thus can only assume that they are similarly underestimated in these data.
References
REFERENCES
Abrahamson, Mark and Lee Sigelman. 1987. "Occupational Sex Segregation in Metropolitan Areas."
American Sociological Review 52:588-97.
Adams, Arvil. 1972. "Black-White Occupational Differentials in Southern Metropolitan Employment." Journal
of Human Resources 7:500-17.
Anderson, Cynthia and Donald Tomaskovic-Devey. 1995. "Patriarchal Pressures: An Exploration of
Organizational Processes That Exacerbate and Erode Sex Earnings Inequality." Work and Occupations
22:328-56.
Baron, James and William T. Bielby. 1980. "Bringing the Firm Back In : Stratification, Segmentation, and the
Organization of Work." American Sociological Review 45:737-65.
_____. 1986. " The Proliferation of Job Titles in Organizations." Administrative Science Quarterly 31:561-86.
Baunach, Dawn Michelle. 2002. "Trends in Occupational Sex Segregation and Inequality, 1950 to 1990."
Social Science Research 31:77-98.
Becker, Henry Jay. 1980. "Racial Segregation among Places of Employment." Social Forces 58:761-76.
Beggs, John J. 1995. " The Institutional Environment: Implications for Race and Sex Inequality in the U.S.
Labor Markets." American Sociological Review 60:612-33.
Beller, Andea H. 1985. "Changes in the Sex Composition of U.S. Occupations." Journal of Human
Resources 20:235-50.
Bergmann, Barbara R. 1996. In Defense of Affirmative Action. New York: Basic Books.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 23/27
Bielby, William T. and James N. Baron. 1984. "A Woman's Place Is with Other Women: Sex Segregation within
Organizations." Pp. 27-55 in Sex Segregation in the Workplace : Trends, Explanations, Remedies, edited by
Barbara Reskin. Washington, DC: National Academy Press.
_____. 1986. "Men and Women at Work: Sex Segregation and Statistical Discrimination ." American Journal
of Sociology 91:759-99.
Cantanzarite, Lisa. 2000. "Brown-Collar Jobs: Occupational Segregation and Earnings of Recent-Immigrant
Hispanics." Sociological Perspectives 43:45-75.
_____. 2003. "Race-Gender Composition and Occupational Pay Degradation." Social Problems 50:14-37.
Carlson, Susan M. 1992. "Trends in Race/Sex Occupational Inequality: Conceptual and Measurement
Issues." Social Problems 19:268-85.
Carrington, William and Kenneth Troske. 1998. "Interfirm Segregation and the Black-White Wage Gap."
Journal of Labor Economics 16:231-60.
Cartwright, Bliss and Patrick Ronald Edwards. 2002. "Sex Segregation Measured by Job Groups and
Industry." Paper presented at the annual meeting of the American Sociological Association, August,
Chicago, IL.
Charles, Maria and David B. Grusky. 2004. Occupational Ghettos: The Worldwide Segregation of Women
from Men. Stanford, CA: Stanford University Press.
Cohen, Philip and Matt Huffman. 2003. "Occupational Segregation and the Devaluation of Women's Work
across U.S. Labor Markets." Social Forces 81:881-907.
Cotter, David A., JoAnn DeFiore, Joan M. Hermsen, Brenda Marsteller Kowalewski, and Reeve Vanneman.
1997. "All Women Benefit: The Macro-Level Effect of Occupational Integration on Gender Earnings
Inequality." American Sociological Review 62:714-34.
Cotter, David A., Joan M. Hermsen, and Reeve Vanneman. 2004. Gender Inequality at Work. New York:
Russell Sage Foundation.
Elvira, Marta M. and Mary E. Graham. 2002. "Not Just a Formality: Pay System Formalization and Sex-
Related Earnings Effects." Organization Science 13:601-17.
Fossett, Mark A., Omer R. Galle, and William R. Kelly. 1986. "Racial Occupational Inequality, 1940-1980:
National and Regional Trends." American Sociological Review 51:421-29.
Gibbs, Jack and William Martin, 1962. "Urbanization, Technology, and the Division of Labor-International
Patterns." American Sociological Review 27:667-77.
Goldin, Claudia. 1990. Understanding the Gender Gap. Oxford, England: Oxford University Press.
Grusky, David B. and Maria Charles. 1998. " The Past, Present, and Future of Sex Segregation Methodology."
Demography 35:497-504.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 24/27
Hakim, Catherine. 2000. Work-Lifestyle Choices in the 21st Century: Preference Theory. New York: Oxford
University Press.
Hultin, Mia and Ryzard Szulkin. 1999. "Wages and Unequal Access to Organizational Power: An Empirical
Test of Gender Discrimination ." Administrative Science Quarterly 44:453-72.
Jacobs, Jerry A. 1989. "Long-Term Trends in Occupational Segregation by Sex." American Journal of
Sociology 95:160-73.
_____. 2001. "Evolving Patterns of Sex Segregation." Pp 535-50 in Sourcebook of Labor Markets: Evolving
Structures and Processes, edited by Arne Kalleberg and Ivar Berg. New York: Plenum Press.
Jacobs, Jerry and Mary Blair-Loy. 1996. "Gender, Race, Local Labor Markets, and Occupational Devaluation."
Sociological Focus 29:209-30.
Jacobsen, Joyce P. 1994. "Sex Segregation at Work: Trends and Predictions." The Social Science Journal
31:153-69.
King, Mary C. 1992. "Occupational Segregation by Race and Sex, 1940-88." Monthly Labor Review 115:30-36.
Kmec, Julie. 2003a. "Minority Concentration and Wages." Social Problems 50:38-50.
_____. 2003b. "Collecting and Using Employer-Worker Matched Data." Sociological Focus 36: 81-96.
Konrad, Alison M. and Frank Linnehan. 1995. "Formalized HRM Structures: Coordinating Equal Employment
Opportunity or Concealing Organizational Practices?" Academy of Management Journal 38:787-820.
Leonard, Jonathan. 1984. "Employment and Occupational Advance under Affirmative Action." The Review
of Economics and Statistics 66:377-85.
Lorence, Jon. 1992. "Service Sector Growth and Metropolitan Occupational Sex Segregation." Work and
Occupations 19:128-56.
Mayhew, Bruce. 1984. "Baseline Models of Sociological Phenomena." Journal of Mathematical Sociology
9:259-81.
McCall, Leslie. 2001. Complex Inequality: Gender. Class, and Race in the New Economy, New York:
Routledge.
Milkman, Ruth. 1987. Gender at Work: The Dynamics of Job Segregation by Sex During World War II.
Urbana, IL: University of Illinois Press.
Petersen, Trond and Laurie Morgan. 1995. "Separate and Unequal: Occupation-Establishment Sex
Segregation and the Sex Wage Gap." American Journal of Sociology 101:329-65.
Pfeffer, Jeffrey and Yinon Cohen. 1984. "Determinants of Internal Labor Markets in Organizations."
Administrative Science Quarterly 29:550-72.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 25/27
Reskin, Barbara F. and Debra Branch McBrier. 2000. "Why Not Ascription? Organizations' Employment of
Male and Female Managers." American Sociological Review 65:210-34.
Reskin, Barbara F, Debra McBrier, and Julie A. Kmec. 1999. " The Determinants and Consequences of
Workplace Sex and Race Composition." Annual Review of Sociology 25:335-61.
Reskin, Barbara F. and Patricia A. Roos. 1990. Job Queues, Gender Queues: Explaining Women's inroads into
Male Occupations. Philadelphia, PA: Temple University Press.
Ridgeway, Cecilia. 1997. "Interaction and the Conservation of Gender Inequality: Considering Employment."
American Sociological Review 62:218-35.
Robinson, Corre L., Tiffany Taylor, Donald Tomaskovic-Devey, Catherine Zimmer, and Matthew W. Irvin, Jr.
2005. "Studying Race or Ethnic and Sex Segregation at the Establishment Level: Methodological Issues and
Substantive Opportunities Using EO-1 Reports." Work and Occupations 32:5-38.
Skaggs S. 2001. Discriminating Litigation: Implications for Women and Minorities in Retail Supermarket
Management. Unpublished doctoral dissertation, North Carolina State University, Raleigh, N.C.
Smith, James P. and Finis Welch. 1984. "Affirmative Action and Labor Markets." Journal of Labor Economics
2:269-301.
Stainback, Kevin. 2006. " Workplace Sex Segregation in the post-Civil Rights Era, 1966-2002: Politics,
Uncertainty, and Organizational Change." Department of Sociology, University of Massachusetts, Amherst,
MA. Unpublished manuscript.
Stainback, Kevin, Corre Robinson, and Donald Tomaskovic-Devey. 2005. "Race and Workplace Integration: A
Politically Mediated Process?" American Behavioral Scientist 48:1200-29.
Stearns, Linda Brewster and Charlotte Coleman. 1988. "Occupational Segregation by Sex in High and Low
Technology Manufacturing Industries." Research in the Sociology of Work 4:289-310.
Strang, David and James N. Baron. 1990. "Categorical Imperatives: The Structure of Job Titles in California
State Agencies." American Sociological Review 55:479-95.
Tilly, Charles. 1998. Durable Inequality. Berkeley, CA: University of California Press.
Tolbert, Charles and Molly Sizer. 1996. "U.S. Commuting Zones and Labor Market Area: A 1990 Update."
Economic Research Service, U.S. Department of Agriculture, Number 9614. Washington D.C.: Government
Printing Office.
Tomaskovic-Devey, Donald. 1993. Sex & Racial Inequality at Work: The Sources and Consequences of Job
Segregation. Ithaca, NY: ILR Press.
Tomaskovic-Devey, Donald, Arne L. Kalleberg, and Peter V Marsden. 1996. "Organizational Patterns of Sex
Segregation." Pp. 276-301 in Organizations in America: Analyzing Their Structures and Human Resource
Practices, edited by Arne L. Kalleberg, David Knoke, Peter V Marsden, and Joe L. Spaeth. Thousand Oaks,
CA: Sage Publications.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 26/27
Tomaskovic-Devey, Donald and Sheryl Skaggs. 1999. "Degendered Jobs? Organizational Processes and Sex
Segregated Employment." Research in Social Stratification and Mobility 17:139-72.
AuthorAffiliation
Donald Tomaskovic-Devey
University of Massachusetts, Amherst
Kevin Stainback
University of Massachusetts, Amherst
Tiffany Taylor
North Carolina State University
Catherine Zimmer
University of North Carolina at Chapel Hill
Corre Robinson
North Carolina State University
Tricia McTague
North Carolina State University
AuthorAffiliation
Direct correspondence to Donald Tomaskovic-Devey, Department of Sociology, University of
Massachusetts, Amherst, MA 01003-9278 ([email protected]). An earlier version of this
article was presented at the 2004 meeting of the American Sociological Association. The authors thank
Robert Kaufman and David Brady, the ASR reviewers, and the ASR editor Jerry A. Jacobs for helpful
comments on earlier versions of this article. Research supported by The National Science Foundation
grants SES-0216424 and SES-0216843.
AuthorAffiliation
Donald Tomaskovic-Devey is a Professor of Sociology at the University of Massachusetts, Amherst. He is
interested in processes of organizational inequality and change, particularly with regard to equal
employment opportunity. He also is exploring interorganizational market relationships, contrasting market,
hierarchy, and embeddedness approaches.
Catherine Zimmer is a Senior Research/Statistical Consultant at the Odum Institute for Research in Social
Science and Adjunct Associate Professor of Sociology, both at the University of North Carolina at Chapel
Hill. She is interested in organizational sociology, workplace inequality, and the application of quantitative
analysis techniques. Currently, she is studying the distribution of different types of organizations in
Australia and the relationships of these organizations to their customers and suppliers. In addition, she
consults on a variety of research projects by providing statistical analysis expertise. Her recent work has
appeared in Social Forces, Work & Occupations and Social Science and Medicine.
4/20/2021 Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966-2003 - Sociological Abstracts - ProQuest
https://search.proquest.com/socabs/docview/218830374/9FAE6418C79D4A43PQ/14?accountid=14275 27/27
Kevin Stainback is a Visiting Assistant Professor at the University of Massachusetts, Amherst. He recently
completed his dissertation examining the influence of the political environment on changes from 1966 to
2002 in white male, white female, black male, and black female access to private sector management jobs.
With support from the Russell Sage Foundation, he currently is working on a book manuscript describing
changes in U.S. workplace equal employment opportunity since the Civil Rights Act of 1964.
Corre Robinson is a Ph.D. student at North Carolina State University and Senior Research Associate with
McNeil Research and Evaluation Associates. His research interests are racial disparity in employment and
housing. His dissertation examines the influence of city-level political dynamics on private sector
employment change using the EEOC data. In 2005 he was the lead author of a methodological paper on
the use of EEOC data published in Work and Occupations.
Tiffany Taylor is a Ph.D. student and instructor in Sociology and Women s Studies at North Carolina State
University. Her research includes studies of workplace inequality, spatial variability in inequality,
organization volunteerism, and the relationship between family and work. Her dissertation research
examines how organizations providing Temporary Assistance to Needy Families related services construct
and maintain legitimacy and effectiveness despite considerable obstacles and constraints.
Tricia McTague is a Ph.D. student at North Carolina State University. Her research interests include
workplace and organizational inequality, specifically the use of teamwork in the retail sector, plant
closings, and job loss, as well as the formation ofracialized and gendered identities among middle school
children. She currently is writing a dissertation on union organizing drives in a well-known retail firm.
Copyright American Sociological Association Aug 2006
Tennessee Electronic Library
Copyright © 2021 ProQuest LLC.
