Dear Song,
I’ve enjoyed reading this manuscript and working on language issues in it. You can see the
changes I made using “Track Changes.” I tried using search and replace for some terms, but (as
with many things with this very old Carleton laptop), it didn’t seem to function properly. Thus,
there are a few places where you will find me “correcting” words with exactly the same word. I
use double parentheses to set off my comments to you. I read through the manuscript twice, the
second time without the editing showing on the screen so I could check for the flow of language
with my first round of corrections.
I was particularly impressed by the thoroughness of your statistical investigation and the many
models you developed to examine a variety of interrelated hypotheses. I think you might be able
to highlight some issues of discrimination that interact with human capital (especially language
fluency) issues, so that readers cannot possibly misunderstand that you propose a voluntaristic
solution to Hispanics’ depressed wages (they just have to learn English and stop speaking
Spanish at home). Here I am thinking both substantively and in terms of the particular setting in
which you want to publish this article (and thus who the reviewers are likely to be).
You might also want to briefly discuss the limitations of using language spoken at home as an
indicator of English fluency, at least in a footnote, since there are both advantages (strongly
plausible assumptions) and disadvantages (language fluency is probably strongly related to level
of education, to amount of time in the United States, and to residence in ethnic enclaves vs. more
integrated settings [or to workplaces that are either ethnically/linguistically homogeneous or
heterogeneous, or to ego-centered networks that are either ethnically/linguistically homogeneous
or heterogeneous]; in these ways, language spoken at home might be a weak indicator on its
own; rather, it is likely to be part of a bundle of indicators of language fluency. It is, though, an
available and manageable indicator for your analyses.).
You discuss language fluency as an aspect of human capital. I wonder if it could also be
discussed in terms of cultural capital or social capital—which might help explain the particular
effect of language fluency on Hispanics alone. Bourdieu’s essay on forms of capital might be
helpful here (at least to cite it, and to be clear about these different forms).
There is no way to test for this, but I wonder about the effect of language fluency on African
Americans’ job chances and income outcomes. Here race/ethnicity combines with
class/education and the ability to “code switch” (switch from one form of language and social
interaction to another, according to context; this is something that Norbert Elias describes as an
essential part of the “civilizing process.”) While African Americans speak English, not all are
able to switch between “Ebonics” (Black English, or inner-city black dialect, with its own accent
and grammar) and standard (academic) English. Just something to think about, although I have
no idea if, how, or where it would fit into your paper!
Another issue you’ll want to keep a sharp eye out for as you read through the paper (and my
corrections and comments), is the use of racial terminology. I don’t have much available here,
but I did read the essay on race in a recent anthropology reader, Exotic No More: Anthropology
on the Front Lines, ed by Jeremy MacClancy, U of Chicago Press, 2002. The author, Faye
Harrison, uses both black (non-capitalized) and African American. She doesn’t discuss Hispanics
in the U.S. There are particularly interesting passages for you, on pp. 151 (on North Asian
Ambiguity) and 158 (regarding the Caribbean/Latin American color lexicon for phenotypical
gradations, because those of African descent are believed to be “without culture” and thus “race”
matters more, and the lack ot these for other ethnic populations such as East Indians and Chinese,
who are assumed to be culturally saturated). Such ideas might help interpret the differing effects
of various elements of human capital stocks for members of different racial categories. The
beginning of the essay might give you an idea of recent discourse regarding race and racial
terminology.
Regarding terminology, I suggest that you try to use the categories in the California Workplace
survey. (For example, you might want to replace “white” by “Caucasian” in many settings, as
long as it doesn’t get too awkward.) I also suggest that you try to use “income” rather than
“earnings” in most contexts. It is the term you use in your title, and is grammatically less
ambiguous! You may want to use search and replace to do this. Sorry it doesn’t work on my
machine.
Brief grammar lesson: you only make three mistakes consistently, so these would be things to
watch out for in future writing: 1) plurals; 2) use of articles (“a”, “the”); 3) use of semicolons (;),
which should only be used between two phrases that could each stand on their own as complete
sentences.
Good luck with getting this published. I think it’s a strong paper, and hope very much that it will
be published in this issue. It continues but expands your research focus on occupations and job
training/human capital, and connects that aspect of your work to other sociological subfields. It
could only be a wonderful additional feather to your cap! Don’t hesitate if you have any further
questions.
Best wishes,
Pamela
2
ENGLISH NON-FLUENCY AND INCOME PENALTY FOR
HISPANIC WORKERS*
SONG YANG
Assistant professor
Department of Sociology and Criminal Justice
211 Old Main
University of Arkansas
Fayetteville AR 72701
Email: yangw@cavern.uark.edu
Phone: 479-575-3205
*Direct all correspondence to Song Yang by writing to Department of Sociology and Criminal
Justice, 211 Old Main, University of Arkansas, Fayetteville, AR, 72701 or by email to
yangw@cavern.uark.edu.
3
ENGLISH NON-FLUENCY AND INCOME PENALTY FOR
HISPANIC WORKERS*
Abstract
Using the 2001-2002 California Workforce Survey, this contribution examines the income gap
between Hispanic and white workers. In light of human capital theory, I attribute the income gap
between Hispanic and white workers to differentials in their human capital. Data analyses
indicate that classical human capital indicators such as education, job training, and work
experiences are not sufficient to account for the observed income gap between Hispanics and
whites. Instead, English fluency is a highly valuable aspect of human capital for Hispanic
workers. English non-fluency, along with less education, job training, and work experience
explain why Hispanic workers earn less than white workers. However, variations in English
fluency do not affect the earnings of Asian workers. Those findings suggest that English nonfluency is a unique source of income penalty for Hispanic workers. I conclude with some
proposals for future studies.
4
ENGLISH NON-FLUENCY AND INCOME PENALTY FOR
HISPANIC WORKERS*
This research focuses on workplace inequality by investigating sources of the income
((could also be “sources of earning differences” but I think that is less strong)) gap between
Hispanic and white workers. For several decades, research on ascriptive workplace inequalities
has made significant contributions to our understanding of differentials in job training attainment
(Knoke and Ishio 1998; Caputo 2002), pay raises (Kaufman 1983; Browne et al. 2001), job
authority attainment (Smith 1997; Wilson 1997), and work dissolution (Elvira and Zatzick 2002).
However, the majority of those studies have focused on two groups: whites and African
Americans. Indeed, in much of the literature on workplace inequality, minority is synonymous
with African American. However, the turn to the new millenium has witnessed drastic changes in
the American demographic landscape. The 2000 U.S. Census Bureau reported that Hispanics
(12.5 percent) replaced African Americans (12.3 percent) to become the largest minority group
in the nation (http://www.census.gov/census2000/states/us.html). The newly released statistical
yearbook of the Bureau of Citizenship and Immigration Services (BCIS) reported that Mexico is
among the top five countries sending the most immigrants to the U.S. in recent years, along with
India, the Peoples Republic of China, the Philippines, and Vietnam
(http://www.bcis.gov/graphics/shared/aboutus/statistics/index.htm). The U.S. Census Bureau
projected that the Hispanic and Asian populations will double in the next 50 years, in contrast to
a slight increase of the African American population and a decline in the Caucasian population
(http://www.census.gov/population/www/projections/natproj.html). Indeed, a mosaic is emerging
5
in the American racial landscape, yet research addressing racial discrepancies between Hispanic
and white workers in crucial labor outcomes is scarce.
Decades of studies on workplace ascriptive inequalities have accumulated a large body of
knowledge on the causal factors of those inequalities. Early economic studies focused on both
sides of labor demand and supply. On the demand side, the observed wage gap between whites
and blacks was due to employer’s “discriminatory taste” (Becker 1957) or employer’s “statistical
discrimination” (Thurow 1975). On the supply side, classic human capital theory states that the
low level or low quality of education of education received by African Americans explains why
blacks make less money than whites (Becker 1993). Later sociological studies report that job and
workplace segregation and the devaluation of female and minority jobs are to be blamed for the
resulting wage gaps between men and women, whites and non-whites (England 1992;
Tomaskovic-Devey and Skaggs 1999; Tomaskovic-Devey and Skaggs 2002).
However, because most studies on racial inequalities have focused on white-black
comparisons, results and models from those studies are not readily applicable to explain
differentials between whites and Hispanics. For example, English fluency has been found as one
of the most significant factors that explains the income gap (McManus, Gould, and Welch 1983)
and occupational differences between Hispanic and white workers (Stolzenberg 1990). However
conventional studies comparing white and black incomes have paid scant attention to the issue
of language proficiency, as a vast majority of whites and blacks are native-English speakers. This
study uses data from the 2001-2002 California Workforce Survey to re-investigate the roots of
the income gap between Hispanic and white workers. I attempt to understand this income gap
with insights from human capital theory. Sections below review pertinent theories and propose
several testable hypotheses.
6
Human Capital Theory and Income Differentials between Whites and nonWhites
Theodore Schultz’s (1961) seminal inauguration speech on human capital and American
economic growth marked a revolutionary turn in the social sciences from national economic
growth to individual earning differentials. According to Schultz, human capital, like other forms
of capital, results from long term deliberate investments in areas such as education, job training
and health. As a consequence of investments in human capital, the American economy
experienced significant growth, especially after WWII. Other human capital advocates use the
human capital model to explain variations in personal income and the behavior of firms. In
several separate treatises, Mincer (1962; 1991; 1994) estimates earning returns to job training.
Although the exact figures vary depending on data, methodology, and time frame, job training
recipients are guaranteed returns that commonly are expressed as higher income in the posttraining session. Noble Prize laureate Gary Becker (1993) analyzed how firms react differently
depending on the consequences of job training. General training raised recipients’ marginal
products in the firms that provided the training and other competitive firms. Hence workers who
received general training see an increase in their salaries in and outside of the firm that supplied
such training. Consequently, generally trained workers, as the major beneficiaries of such
training, incur the cost of this training. In contrast, specific training augments trainees’
productivity and income only in the firms that supplied such training. Specifically trained
workers will not receive pay raises in other work settings, as the skills they learned are only
applicable to the firms that trained them. Thus employers that provided specific training will
keep trainees’ wage rates at the same level as those in the pre-training session and exploit the
extra profit between enhanced marginal product and the relatively lower wage. Consequently, the
employers pay for specific training.
7
Empirical evidence that human capital investments produce significant income returns is
considerable and compelling (Becker 1993). Scholars have fruitfully applied human capital
theory to explain gender and racial differences in salaries and training attainment (Duncan and
Hoffman 1979; Olsen and Sexton 1996; Barron and Black 1993; Altonji and Spletzer 1991).
Two aspects of human capital are used to explain earning differentials between men and women,
and between whites and minorities. The first aspect concerns the quality of human capital;
women and minorities receive lower returns from their human capital investment than do white
male workers because the quality of their human capital is relatively lower (Becker 1993: 195204). The second aspect concerns the quantity of human capital; women and minorities receive
lower pay than their white male coworkers because women and minorities have a lower level of
human capital stocks than do white male workers. ((Both of these appear to be testable
hypotheses that need to be investigated empirically. Becker seems to state them as theoretical
axioms.))
Empirical studies often support the quantity argument. For example, research has
identified training differentials as the source of earning gaps by race and gender. That women
and minorities complete less training explains why their income is lower than that of white men
(Duncan and Hoffman 1979). Women also tend to fill jobs that offer a shorter duration of on-thejob training. In one study, differences in training duration explained 45 percent of the income
difference between men and women in the post-training session (Baron, Black and Loewenstein
1993). However, empirical evidence diverges on the quality argument that attributes the
relatively low wages of women and minorities to the low quality of their human capital. Some
scholars report that the quality of schooling African Americans receive is lower than that
received by whites (Card and Krueger 1998; Farkas 1996). But others found that industrial
8
productivity is higher in industries with a high proportion of African American employment
(Galle et al. 1985), indicating a disjuncture between quality of schooling (an aspect of human
capital) and productivity. In fact, the very observation that minorities receive a lower rate of
return from their human capital investment, despite the lack of conclusive evidence indicating
their low quality of education (or productivity consequences of educational quality), has become
striking evidence of employment discrimination against minorities (Finkelstein and Levin 1990).
English Proficiency: a Crucial Component of Human Capital for Immigrant
Workers
Earlier human capital scholars stressed the pivotal roles of education, job training, and
work experience in affecting income. Because a vast majority of white and African American
workers are native English speakers, classic studies on white-black employment inequalities do
not include English language proficiency as one of the explanatory factors (Siegel 1965; Duncan
1969). But, English language proficiency has taken on increasing importance due to
demographic changes over the past three decades. Starting in the 1970s U.S. immigration grew
rapidly, following a fifty year slowdown due to restrictive immigration policies. The national
origins of the new immigrants shifted from European countries to Latin American and Asian
countries (Borjas and Tienda 1987). Since the 1970s, American workplaces have hired
increasing numbers of Hispanic and Asian workers, most of whom are not native English
speakers (Veltman 1990). Studies on those new immigrants identified a unique source of labor
market penalty: English language deficiency (Chiswick 1978; 1979). One study documented that
low earnings for minority groups are a consequence of their low English fluency, along with
other human capital factors such as low educational attainment and job training (McManus,
Gould, and Welch 1983). In a recent study, scholars have reported novel findings that sometimes
9
higher rate of rate to ((I don’t understand this sentence at all-sorry!)) language proficiency and
education can also indicate workplace discrimination (Stolzenberg and Tienda 1997). For
example, Asians and Hispanics were found to have higher rates of return to their language
proficiency and schooling, although a vast majority of them are not fluent in English and are less
educated (Stolzenberg and Tienda 1997: 44). Consequently, a very small number of Asian and
Hispanic workers earn a high income comparable to white workers, whereas the majority of
them have bottom wages due to their low language proficiency and low educational level.
To the extent that language proficiency facilitates communication with others in the
workplace, speaking the majority language can be considered as an integral component of
workers’ human capital. Accurate language communication is essential to customer satisfaction,
coordination with coworkers, and learning what to do and how to do a job (Stolzenberg and
Tienda 1997). In contrast, language non-fluency handicaps communication, limiting the range of
people with whom workers can provide services or coordinate work. Language non-fluency has
been found to reduce not only job opportunities in general, but also chances of obtaining highpaying jobs for which workers are otherwise qualified (Devine and Kiefer 1991). Studies of
economic ethnic enclaves found that workers’ lack of English skills often led them to obtain jobs
in economic sectors that use languages other than English (Portes and Manning 1986; Robinson
1988). But those non-English sectors often have harsh work environments, hire small numbers of
workers, and offer low wages (Sanders and Nee 1987). Therefore gaining language proficiency
in the majority language is an indispensable component of immigrant workers’ human capital
because fluency in the majority language can increase their potential earnings and outputs (Mora,
1998).
10
Workplace Discrimination and Segregation: other Sources of Earning
Penalty for Minority Workers
Besides the human capital model, workplace discrimination is another major source of
income gaps between predominant whites and minority groups. Early economic theory on
discrimination argued that employers exercise a “taste for discrimination” when they act either to
pay something directly or to incur reduced income in order to be associated with some persons
instead of others (Becker 1957). Discriminatory employers may refuse to hire African Americans
because employers erroneously underestimate economic efficiency of black workers. Although
employers might be ignorant of the true productivity of African American workers, such
ignorance could be eliminated through the proliferation of information that corrects the
erroneous underestimation of black’s productivity. However, prejudiced employers may still hire
whites rather than blacks despite the knowledge that both groups have the same level of
productivity (Becker 1957). To the extent that discriminatory employers refuse to hire people
whose marginal value products are higher than their marginal costs, these employers also incur
significant labor costs for their discriminatory practices. Therefore, market clearance
mechanisms in a completely competitive market will eventually drive the discriminatory
employers out of business because non-discriminatory employers are able to hire cheaper labor
(Becker 1957; Thurow 1975). ((But of course the market is not completely competitive, and
there are gradations of profit that do not necessarily lead to non-survival of firms. Thus,
discriminatory practices can persist.))
Sociological work has contributed greatly to later developments in the study of racial and
gender inequalities at workplaces (England 1992; Tomaskovic-Devey 1993; Tomaskovic-Devey
and Skaggs 1999; Tomaskovic-Devey and Skaggs 2002). Most sociological research focused on
how workplace racial and gender inequalities come about. Comparable worth models were
11
developed to ascertain how organizations create discriminatory job structures that sort women
and minority into minority-dominated and women-dominated jobs (England 1992). As a result,
those female- and minority-coded jobs normally require less training and schooling, are easily
replaced, are less likely to lead to promotion and have lower pay than those jobs whose
incumbents are dominantly white males. Some scholars also identified white male employees,
rather than employers, as the original discriminators (Tomaskovic-Devey and Skaggs 1999).
White male employees exclude women and minority workers from high-paying promising jobs
because they can garner the benefits from creating female-dominated and minority-dominated
job structures within an organization. ((How are these workers able to influence employers, who
I presume have the last say?)) A most recent study reported that the source of the gender gap in
earnings is that women are disproportionally placed, largely by white male employees allied with
employers, in jobs that require less skills, involve lower task complexity, and entail low job
authority (Tomaskovic-Devey and Skaggs 1999). Scholars also have argued for an eclectic
approach that simultaneously accounts for individual, job, and organizational variables in
assessing training differentials between men and women (Knoke and Ishio 1998), and
employment benefits (Kalleberg et al 2000). For example, women are concentrated in those good
job positions, occupations and workplaces which provided more job training to their incumbents.
Therefore, statistically controlling for these mediating factors expands, rather than reduces, the
training gap between men and women, making women further behind men in receiving job
training (Knoke and Ishio 1998). In other words, had men achieved the same work profile as
women, men would have received even more job training than women.
Although employment discrimination and job segregation are not the focus of this study,
the above discussion indicates that job-level and workplace-level variations((variables??)) may
12
mediate racial income differences. For example, if women and minorities are disproportionally
concentrated in less desirable positions or workplaces, which in turn provide low earnings to
their incumbents or workers, controlling for job and workplace variations would erase the
significant earning gap between minority workers and whites. Therefore, this study also
regresses earnings on race, along with mediating independent variables at the job and workplace
levels. The main purpose of including job and workplace independent variables is to identify the
original source that explains the income gap between Hispanic and white workers. By comparing
and contrasting job and workplace models with human capital models, we attempt to spot the
roots of income gap between Hispanics and white workers.
What Affects Earnings for Hispanics: An Empirical Assessment
The observed racial gap in income may result from different sources. Human capital
discrepancy and workplace discrimination all contribute to income differentials across different
racial and ethnic groups. However, attempts to extricate the causative mechanism for these
pervasive earning differentials are scarce and inconclusive (Stolzenberg and Tienda 1997; Kim
2002). A major purpose of this research is to readdress the issue of the causative mechanism of
the income gap between Hispanic and white workers. To examine what causes Hispanic workers
to earn less than whites we focus on changes in the Hispanic-white earning gap while
controlling for different sets of mediating independent variables and comparing how the changes
in the Hispanic-white income gap differ from those between whites and other minorities.
The orthodox human capital model has been used to explain a large portion of earning
differentials between whites and minority groups. As previously noted, the lower income for
minority workers relative to white workers has been attributed to their lower levels of human
capital stock (Becker 1993). Particular to Hispanic workers, English fluency was identified as
13
one of the main human capital factors that explain why Hispanics have lower occupational status
and income than do white workers (Mora 1998; Davila and Mora 2000; Carliner 1981; McManus
et al. 1983; Grenier 1984; Stolzenberg 1990; Stolzenberg and Tienda 1997). But the need to
overcome the language barrier for career development is not unique to Hispanic workers; other
groups, particularly Asian immigrants, face a similar obstacle (Schmid 2003). Earlier we
discussed that English language proficiency is one of the essential human capital factors for
immigrant workers. Therefore, in light of human capital theory, English fluency, much like other
conventional human capital factors such as education, training, and work experience, should
significantly increase earnings for Hispanic workers, and thus mitigate income gaps between
Hispanic and white workers. Likewise, language proficiency should also increases earnings for
Asian workers. This discussion leads to the following four testable hypotheses:
H1: The average income for Hispanic workers is significantly lower than that for white
workers.
H2: The income-gap between whites and Hispanics can be explained by the difference in
their stock of human capital factors including English proficiency, education, training,
and work experiences.
H3: Hispanic workers with great English fluency receive a higher income than do
Hispanic workers without English fluency.
H4: Asian workers with greater English fluency receive a higher income than do other
Asian workers without [with less??]English fluency. ((In H3 and H4, are you comparing
along a continuum, or do you have a way to define two clearly distinct groups at the far
ends of a language proficiency continuum? It would make a difference for the wording of
your hypotheses, as well as for the analysis and interpretation of results.))
14
Data and Measures
The dataset used in my analyses is the 2001-2002 California Workforce Survey. The
survey was designed to assess working conditions in California and to measure the extent to
which various groups of workers differ in regard to wages, hours, benefits, and work control in
their working environment. The survey research center at the University of California, Berkeley
conducted telephone interviews to California residential households during 2001-2002. A
technique called list-assisted random-digit sampling was used to take advantage of large
computer databases of telephone directory information. Three steps were applied to eliminate
business and non-working phone numbers. Telephone interviews of the eligible residential
households produced a sample with 1404 respondents (For details on the survey design, see
2001-2002 California Workforce Survey Codebook). Among the total respondents, 1045 were
working full-time or part-time during the survey period. Because this study investigates work
wage differentials among multiple racial groups, these 1045 workers comprise the final sample
for my statistical analyses.
Income is the dependent variable, measured with questions “how much do you earn per
hour/month/year at this job?” Because respondents provided information on the number of hours
they work per week, I first computed weekly wage for those who reported their hourly wage by
multiplying their hourly rate with number of hours they work per week. I then computed their
annual salary by multiplying their weekly rate by 52. For those who reported their monthly
wage, I computed their annual salary by multiplying their monthly wage with 12. Thus the
dependent variable is the respondent’s annual income. I transformed the personal income into the
nature log form to stabilize sample variance and reduce heteroscedasticity (Allison 1999: 128).
15
Race is the crucial independent variable, whose relationship with income is the main
focus of this paper. The survey asked respondents “Which of the following best describes your
race or ethnic group?” The original coding has six racial groups: Caucasian (620), AfricanAmericans (68), Hispanics (246), Asian (69), Native Americans (17) and Middle Eastern (9).
((Why, in your article, do you not use the wording of this coding, e.g. replace “white” and
“black” with Caucasian and African-American. You should also be aware—and show your
awareness somewhere toward the beginning of the article—that Hispanic cannot be a valid racial
group, even if one continues to believe in “races” (which anthropologists don’t; phenotype is
only one biological criteria with which to distinguish among categories of people; of course, race
is still significant as a social category). Hispanics have many phenotypes and racial
identifications, ranging from “white” to “black.” What unites them is that they are Spanish
speaking. Some also say Latino or Latina (the latter for women). Others distinguish among
different groups of Hispanics according to their national origin (e.g. Chicano for those coming
from Mexico, Dominican for those from the Dominican Republic). I’d suggest looking at the
most recent writings by Alejandro Portes (who writes a lot about immigrants) to find the current
“correct” usage. You just don’t want any reviewers to be offended (even if the changes in
terminology might seem arbitrary). You should also follow any guidelines of usage indicated by
the call for papers for this special issue.)) Because the numbers for Native Americans and Middle
Easterners are too small to warrant a significant statistical inference, we created a new
classification that contains Caucasians (620), African-Americans (68), Hispanics (246), Asians
(69) and others (26) including Native Americans and Middle Easterners.
((Surprisingly few African Americans in this sample!))
16
Human capital is an important independent variable that mediates the relation between
race and income. It includes education, work experiences, employer-paid job training, workerpaid job training, and language ability. To capture the nonlinearity in the monetary returns to
education, I reclassified education into five dummy variables: less than high school, high school,
some college, BA, and postgraduate level. Work experience is measured with the number of
years respondents have worked for the current employer. Employer-paid and worker-paid
training ((be consistent with wording above (or change that wording to fit with your coding))) is
measured respectively with “did you participate in any employer-paid training in the last 3 years?
(yes=1, no = 0)” and “did you ever participate in a training not paid by your employer in the last
3 years (yes=1, no = 0)?” The survey did not directly ask about language fluency. Instead, it
asked respondents to report the language used in their workplace and language used in their
homes. About 93 percent of respondents report that English is the main language in their
workplaces. The rest of the respondents report that Spanish (6 percent) and other languages (1
percent) are the main language used in their workplaces. In addition, the survey asked
respondents to report the language used in their homes. About 85 percent report speaking
English, 11 percent speak Spanish, and the remaining 4 percent speak a variety of different
languages including Chinese, Vietnamese, Filipino, Russian, and other languages. Because this
paper focuses on language proficiency in English, I focused on those whose workplace language
is English and created three different groups based on their home languages. The three groups
are those who speak English at home, those who speak Spanish at home, and those who speak
other languages at home. We reasonably infer that those speaking English at home and work
have a better command of English at work than do those speaking Spanish and other languages at
home but speaking English at work.
17
The control variables include individual characteristics such as age and sex, job level
characteristics such as unionization, job supervision, full time work, and occupational
classification into seven major groups, and workplace characteristics such as independent
workplace, size of workplace, and industrial classification into eight main groups. Appendix
((number of appendix? Otherwise start the sentence “The appendix…”)) displays the details of
variable construction for the control variables.
Findings
Table 1 compares average income levels among the five racial groups. It shows that an
earning penalty associated with being a minority is pervasive in contemporary California
workplaces. While the average earning for the entire sample of California workers is $42,145 per
year, the annual incomes for Caucasians, African Americans, Hispanics, Asians and other
minority groups are $48,329, $42,984, $28,394, $43,845 and $42,905 respectively. Compared to
other groups, whites have the highest income, whereas Hispanics make the lowest wages ((salary
is a particular kind of raise, with a set annual income as opposed to being paid per hour; thus,
salaried workers usually have higher social status)), which accounts for merely 58.75 percent of
that for whites. The F ratio indicates that between-group variation in annual income is highly
significant, which mainly reflects the significant income disparity between whites and Hispanics.
This result supports H1 that average earnings for Hispanic workers are significantly lower than
that for whites. Asians made 90.72% of the income of white workers; the difference is not
significant. Similarly, the income gap between blacks and whites is not significant without
controlling for other independent mediating variables.
Sources of Income Penalty for Hispanic Workers
18
To account for the sources of income differentials among different racial groups, I regress
the annual income on races and steadily include other mediating variables of individual
characteristics, human capital factors, job classifications, and workplace features. Table 2 shows
five different models; each controls for different groups of mediating factors. The income
disparity between whites and Hispanic workers persists in Models 1, 3, and 4. In particular,
Hispanic workers make 84.54 (Exp. (-.168)=84.54%) percent of the income of white workers1 in
Model 1 that controls for age, gender, and classical human capital variables such as education,
training, and work experiences. Model 3 shows that Hispanics make 75.88% (Exp (-.276) =
75.88%) of white income, net of effects from job characteristics such as union status, full time
work, job supervisory power and occupational segregation. Model 4 controls for workplace
characteristics such as industry, size, independence, and economic sector. Yet, Hispanic workers
still make only 72.98% (exp (-.315) = 72.98%) of white income. These results suggest that
variations in job and workplace characteristics are not adequate to explain why Hispanic workers
make less money than white workers.
Model 2 presents striking results as the white-Hispanic income gap disappears when
controlling for human capital indicators such as education, training, work tenure, and English
fluency. This result supports H2 that controlling for human capital indicators eradicates the
income gap between Hispanic workers and whites. In other words, Hispanic workers earn less
because they have lower levels of human capital stock than do whites ((does this suggest that
they do not face job discrimination (which would be quite surprising), or that they face
1
Because the dependent variable is nature log of annual income and white is the reference group, the interpretation
of regression coefficient should follow the steps below
log (Hispanic income) – log (white income) = -.168
log (Hispanic income/white income) = -.168
Exponential (log(Hispanic income/white income)) = Exponential (-.168)
Hispanic income/white income = 84.54%
19
discrimination with regard to access to developing their human capital stock?)). To quantify this
allegation ((?? Word use?? Show this paragraph to a native-speaker colleague and ask about the
use of “allegation” in this context.)), I compared Hispanic and black workers to white workers
along crucial human capital dimensions. Table 3 shows that Hispanic workers on the average
receive 2.58 fewer years of education than do white workers. Breaking down the education level
into five different categories, the percentages of Hispanic workers receiving less than high school
or high school education are significantly higher than white workers by 21% and 9%
respectively. In contrast, Hispanic workers receiving BA and graduate degrees are 13 percent
and 17 percent significantly lower than are white workers. Comparison of other human capital
dimensions between Hispanics and whites yields a similar pattern. Hispanic workers have shorter
work tenure than do whites. Fewer Hispanic workers participate in employer-paid and unpaid job
training. About 29 percent of Hispanics speak Spanish at home, in contrast to 0 percent of white
workers who use Spanish at home. The evidence is striking that the source of the Hispanic-white
income gap lies in the discrepancy in their human capital stock. Whites are more educated,
receive more job training, have longer job tenure, and have greater English fluency than are their
Hispanic coworkers; all contribute to the higher income of whites compared to Hispanics. ((I
wonder if longer job tenure should really be included in human capital stock; while it indicates
on the job training and competency on the one hand, it also could be a measure of discriminatory
hiring/firing practices.))
However, as analyzed in Model 1, the conventional human capital indicators of
education, training, and work experience are not sufficient to explain income gaps between
Hispanic and white workers. The results suggest that English fluency is a highly critical element
of human capital for Hispanic workers. By merely adding the English fluency indicator to the
20
other human capital factors as control variables, the income gap between Hispanics and whites
disappears. In particular, those who speak Spanish at home earned only 70.26 percent (Exp. (.353) = 70.26%) of the income of their coworkers using English both at work and at home. In
contrast, there is no significant income gap between those speaking other languages at home and
those using English at home. Table 2 also shows that except for black workers, who made
73.79% (Exp (-.304) = 73.79%) of white income, income gaps between white and other minority
groups disappear in Model 5 when all the mediating independent variables are put under control.
((Somehow it seems that the LACK of a significant income gap between those speaking other
languages at home and English native-speaker whites, AND the rather similar income gap
between whites and blacks, indicates that something is going on IN ADDITION TO the human
capital/language fluency issue. Is there a way to put your one phenomenally strong result for
Hispanics’ language fluency into the context of your other results and into the context of the
sociology of workplace discrimination? See also my note (before the paper) regarding Ebonics.
My guess, though, is that prejudices about workplace behavior and productivity combine with
language to produce the income inequalities revealed in your study.))
English non-fluency, Income Penalty and Contingent Returns to Education
To probe how English fluency differentiates ((instead of “differentiates” would “affects”
or “influences” be better?)) earnings within the each racial group, we regress income on the
English fluency indicator, along with other human capital measures and other independent
control variables separately for the four racial groups of Caucasian, African American, Hispanic,
and Asian workers. Due to space restrictions, table 4 lists statistical results only for English
fluency and other human capital indicators; complete results including all other coefficients
21
shown in [which?] table [format] are available upon request to the author. Table 4 shows that
English fluency exerts an effect on income only for Hispanic workers. In support of H3, the
results show that Hispanic workers speaking Spanish at home make only 81 percent (Exp (-.203)
= 81.63%) of the income for their coworkers speaking English at home. Because there are no
white or black workers who speak Spanish at home, the income comparison for white and black
workers is restricted to English-speaking and other language speaking groups; none of the
comparisons is significant. However, because the vast majority of African American and white
workers speak English at both work and home (only 1 percent of them use other languages at
home), the relation between language proficiency and income for these two groups awaits future
scrutiny, when a larger sample is available to capture the divergence of language usage and
English fluency for white and African American workers. The results do not support H4 since
English fluency at work has no effect on the income levels of Asian workers. Those Asian
workers who speak Spanish or other languages at home do not make less money than other Asian
workers who speak English both at home and in their workplaces. Thus statistical analyses of
income variation within each racial group provide compelling evidence that English non-fluency
is a unique source of earning penalty for Hispanic workers. To the extent that language
proficiency facilitates communications among coworkers and with managers, American
employers only scrutinize Hispanic workers to ensure that their English is sufficient for effective
communication, and penalize those with low English skills by cutting their wages. ((Or, they
penalize Hispanic workers by hiring them only into lower paying jobs, but do not penalize nonfluent Asians in the same way. Or, is this not a possibility due to results from controlling for
occupation (but would the occupational categories have been subtle enough to capture this?))
22
Table 4 also shows conspicuous findings that returns to human capital are contingent
upon race. The impact of education and job training on income is more pronounced for white
workers than for minority workers. Compared to white workers with less than a high school
education, white workers with BA or graduate degrees receive 60.32% (Exp. (0.472)=1.6032)
and 94.25% (Exp. (0.664)=1.9425) higher earnings respectively. Receiving employer-paid job
training also raises income for white workers by 13.31% (Exp (0.125) = 1.1331). Only graduate
degrees enhance Hispanic workers’ earning: compared to Hispanic workers with less than a high
school education, Hispanic workers with graduate level degrees earn 63.72 percent (Exp. (0.493)
= 1.6372) higher income However, as table 3 reported, only 5 percent of Hispanic workers have
gained a graduate level degree. A majority of Hispanic workers have less than BA level
education, which does not significantly increase their earnings. Human capital variables do not
exert a significant impact on income for Asian and African American workers.
To statistically analyze how returns to human capital indicators differ between Hispanic
and white workers, we conduct a two-sample t test to compare their unstandardized coefficients
(Knoke, Bohrnstedt, and Mee 2002: 178). The last column in table 4 shows the results that a
discernible gap appears between whites and African Americans ((but did you really intend to say
Hispanic??)) in returns to graduate level education. Compared to white workers with less than a
high school education, graduate-level educated white workers earn 94.25 percent (Exp. (0.664) =
1.9425) higher income, whereas that comparison among Hispanic workers yields only 63.72
percent (Exp. (0.493) = 1.6372) earning differences. In other words, to the extent that graduatelevel training can boost one’s income, white workers can gain more from this training than can
Hispanic workers.
23
Comparing Hispanics with African American and Asian Workers
My data analyses of the 2001-2002 California Workforce Survey identified roots of
income differentials between Hispanic and white workers. That Hispanic workers have a lower
level of human capital stock—such as education, job training, work experience, and English
proficiency—than their white coworkers explains why they earn less than whites. It seems that
by increasing their human capital stock, Hispanic workers can significantly improve their income
and catch up with white coworkers. However, returns to graduate-level training are significantly
lower for Hispanics than they are for white workers. This suggests that Hispanic workers may
encounter a limit imposed on their efforts to totally eradicate their income gap with whites by
increasing their human capital.
Comparing earnings between white and African American workers yields a different
pattern: none of the mediating independent controls from various sources can explain the income
gap between whites and African Americans. The failure to identify the roots for producing such
an income gap suggests that little can be done to bring up income for Black workers. ((Or, it
suggests that what needs to be done is the work toward reducing or eliminating discriminatory
wage practices; this suggests a systemic and political (in terms of law-making and compliance
enforcement) solution, rather than one that individual workers can implement (e.g. seeking
further education or other ways of increasing their human capital stocks).))To the extent that
more mediating factors are needed to explicate the white-black income differentials, much
deeper social economic factors exist to account for the disparity between white and black
workers than the differentials between whites and Hispanics. The persistent income gap between
African American and white workers, despite controls of all mediating factors, suggests that
African American workers may encounter greater discrimination from employers than do
24
Hispanics, and that there is little they can do from their own initiative to change the current
situation. (Yes, see my comment immediately above!)
Asian workers share many similarities with Hispanic workers in several important work
profiles. Both groups have a considerable number of new immigrant workers who do not speak
fluent English (Stolzenberg and Tienda 1997), and are highly heterogeneous in their ethnic
backgrounds. However, the income comparison between Asians and whites contrasts starkly to
that between Hispanics and whites. First, the income gap between Asian workers and white
workers is the smallest (see table 1) and insignificant (see table 2), whereas the gap between
Hispanic and white workers is the largest (table 1) and significant, except for the model that
controls for human capital indicators (table 2). Second, although English fluency appears to be so
vital to the earnings of Hispanic workers, the impact of language on income is completely absent
for Asian workers. This finding concurs with a previous report that occupational inferiority for
Hispanic workers is most pronounced when they have low English fluency and low schooling, a
negative stereotype commonly held by many American employers for Hispanic workers
(Stolzenberg 1990:151). As this profiling is absent for Asians, English proficiency is not a
significant factor for Asian workers’ income levels.
Discussion
The turn of the century witnessed significant changes in the American population
landscape. The Hispanic population has replaced the African American population to become the
largest minority group. Yet, systematic studies on issues related and unique to Hispanic people
are scarce. Analyzing data from a workforce survey from California, a state with the most
diverse population composition, this paper made important contributions to the study of the
Hispanic workforce. It identified English deficiency as a unique source of earning penalty for
25
Hispanic workers. Less educated Hispanic workers with English difficulties receive the largest
earning penalty. In contrast, Hispanic workers with graduate degrees and English fluency can get
ahead to receive a significant income increase. However, only 5 percent of Hispanic workers
received graduate level degrees, and returns to graduate-level schooling for Hispanic workers are
significantly lower than those for white workers. These results suggest that even with a
significant increase in their human capital, Hispanic workers may still have an income gap with
white workers.
To the extent that language proficiency facilitates coworker communication and enhances
productivity, language fluency is considered an integral component of workers’ human capital.
Hence, language proficiency, much like education, should increase workers’ income. This effect
should be particularly pronounced for a group with great variation in their language proficiency.
Hispanics and Asians are good cases in point as both groups are highly heterogeneous in their
ethnic backgrounds and both have a considerable number of non-native English speakers.
However, my research reveals that the effect of English fluency on earning is a contingent one.
Hispanic workers who speak Spanish at home are the only group that pays an income penalty for
English deficiency. This finding strongly concurs with previous studies that Hispanic workers
fitting the stereotype of less education and poor English skills incur the greatest loss in
American labor market (Stolzenberg 1990). This line of work also holds great promise. For
example, an employer-level survey can determine the existence and extent of employer profiling
of different racial groups. A critical question awaits scrutiny as to the consequences of such
employer profiling on employees fitting into that profiling. The evidence presented in this
research seem to suggest that when employer profiling is in place, employees with negative
profiles are likely to endure a great deal of financial loss. Pending future evidential support, this
26
finding entails some preliminary implications for Hispanic workers. First, individual Hispanic
workers can improve their earning by increasing their human capital stocks, including gaining
more education and greater English fluency ((which of course could not be measured, as you did
here, by what language they speak at home!)). Second, it may take a long time and collective
efforts from the entire Hispanic workforce, in conjunction with policy-makers, to eliminate the
employer profiling that assigns negative stereotypes such as low education level and poor
English skills to Hispanic workers. Until then, the chronic and idiosyncratic bias against
Hispanic workers will continue to handicap their earnings. ((I think you’ve repeated this last
phrase enough that it is clear in this context.))
27
References
Allison, Paul. 1999. Multiple Regression: a Primer Thousand Oaks, California: Pine Forge Press
Altonji, Joseph G. and James R. Speltzer. 1991. "Worker Characteristics, Job Characteristics,
and the Receipt of On-the-Job Training." Industrial and Labor Relations Review 45:5879.
Barron, John, Dan Black. 1993. “Gender Differences in Training, Capital, and Wages.” Journal
of Human Resources 28:343-64.
Becker, Gary. 1957. The Economics of Discrimination Chicago IL.: University of Chicago Press
Becker, Gary. 1993. Human Capital: a Theoretical and Empirical Analysis, With Special
Reference to Education Chicago: The University of Chicago Press
Browne, Irene, Cynthia Hewitt, Leann Tigges and Gary Green. 2001. “Why Does Job
Segregation Lead to Wage Inequality among African Americans? Person, Place, Sector,
or Skills?” Social Science Research 30/3: 473-495
Borjas, George and Marta Tienda. 1987. “The Economic Consequences of Immigration” Science
235:645-651
Caputo, Richard. 2002. “Discrimination and Human Capital: A Challenge to Economic Theory
& Social Justice” Journal of Sociology and Social Welfare 29/2:105-124
Card, David and Alan Krueger. 1998. “School Resources and Student Outcomes” Annals of the
American Academy of Political and Social Science 559: 39-54
Carliner, Geoffrey. 1981. “Wage Differences by Language Group and the Market for Language
Skills in Canada” The Journal of Human Resources. 16/3:384-400
Chiswick, Barry. 1978. “The Effect of Americanization on the Earnings of Foreign-born Men”
Journal of Political Economy 86:897-921
Chiswick, Barry. 1979. “The Economic Progress of Immigrants: Some Apparently Universal
Patterns” Pp 357-399 in the Contemporary Economic Problems, edited by Fellner,
William Washington: American Enterprise Institute for Public Policy Research
Davila, Alberto and Marie Mora. 2000. “Hispanic Ethnicity, English-Skill Investments, and
Earnings” Industrial Relations 40/1:83-88
Devine, Theresa and Nicholas Kiefer. 1991. Empirical Labor Economics: the Search Approach
New York: Oxford University Press
28
Duncan, Greg, and Saul Hoffman. 1979. “On-the-Job Training and Earnings Differences by Race
and Sex.” Review of Economics and Statistics 61:594-603.
Duncan, Otis. 1969. “Inheritance of Poverty or Inheritance of Race?” Pp 85-110 in On
Understanding Poverty: Perspectives from the Social Sciences, edited by Moynihan,
Daniel P. New York: Basic Books
Elvira, Marta and Christopher Zatzick. 2002. “Who's Displaced First? The Role of Race in
Layoff Decisions” Industrial Relations 41/2:329-361
England, Paula. 1992. Comparable Worth: Theories and Evidence New York: Aldine de Gruyter
Farkas, George. 1996. Human Capital or Cultural Capital?: Ethnicity and Poverty Groups in an
Urban School District New York: A. de Gruyter
Finkelstein, Michael and Bruce Levin. 1990. Statistics for Lawyers New York: Springer
Galle, Omer, Hinson Wiswell and Jeffrey Burr. 1985. “Racial Mix and Industrial Productivity”
American Sociological Review 50/1:20-33
Grenier, Gilles. 1984. “Shifts to English as Usual Language by Americans of Spanish Mother
Tongue” Social Science Quarterly 65/2:537-550
Kalleberg, Arne, Barbara Reskin and Ken Hudson. 2000. “Bad Jobs in America: Standard and
Nonstandard Employment Relations and Job Quality in the United States” American
Sociological Review 65/2:256-278
Kaufman, Robert. 1983. “A Structural Decomposition of Black-White Earnings Differentials”
American Journal of Sociology 89/3:585-611
Kim, Jongsung. 2002. “Learning and Earnings: How College Matters in Earnings Gap between
Hispanic Workers and Non-Hispanic White Workers” Journal of Hispanic Higher
Education 1/2:125-143
Knoke, David and Yoshito Ishio. 1998. “The Gender Gap in Company JobTraining” Work and
Occupations 25/2:141-167
Knoke, David, George Bohrnstedt and Alisa Potter Mee. 2002. Statistics for Social Data
Analysis, 4th Edition Itasca, Illinois: F.E. Peacock Publishers
McManus, Walter, William Gould and Finus Welch. 1983. “Earnings of Hispanic Men: the Role
of English Language Proficiency” Journal of Labor Economics 1:101-30.
Mincer, Jacob. 1994. “Human Capital: A Review” Pp 109-141 in Labor Economics and
Industrial Relations: Markets and Institutions edited by Kerr, Clark and Paul
Staudohar Cambridge, Mass.: Harvard University Press
29
Mincer, Jacob. 1991. “Job Training: Costs, Returns, and Wage Profiles.” Pp. 15-40 in Market
Failure in Training, edited by David Stern and Josef Ritzen. New York: Springer Verlag.
Mincer, Jacob. 1962. “On-the-Job Training: Costs, Returns, and Some Implications.” Journal of
Political Economy (Supplement) 70:50-79.
Mora, Marie. 1998. “Did the English Deficiency Earnings Penalty Change for Hispanic Men
between 1979 and 1989?” Social Science Quarterly 79/3:581-594
Olsen, Reed Neil, and Edwin A. Sexton. 1996. “Gender Differences in the Returns to and the
Acquisition of On-the-Job Training.” Industrial Relations 35:59-77.
Portes, Alejandro and Robert Manning. 1987. “The Immigrants Enclave: Theory and Examples”
Chapter 2 in Competitive Ethnic Relations edited by Olzak, Susan and Joanne Nagel
Orlando: Academic Press
Robinson, Chris. 1988. “The Distribution of Language Skills and Earnings in a dual-language
Economy” Research in Labor Economics 9:53-90.
Sanders, Jimmy and Victor Nee. 1987. “Limits of Ethnic Solidarity in the Enclave Economy”
American Sociological Review 52/6:745-767
Schmid, Carol. 2003. “Immigration and Asian and Hispanic Minorities in the New South: An
Exploration of History, Attitudes, and Demographic Trends” Sociological Spectrum
23/2:129-157
Schultz, Theodore. 1961. “Investment in Human Capital” American Economic Review 51:1-17
Siegel, Paul. 1965. “On the Cost of Being a Negro” Sociological Inquiry 35/1:41-57
Smith, Ryan. 1997. “Race, Income, and Authority at Work: A Cross-Temporal Analysis of Black
and White Men” Social Problems 44/1:19-37
Stolzenberg, Ross. 1990. “Ethnicity, Geography, and Occupational Achievement of Hispanic
Men in the United States” American Sociological Review 55/1:143-154
Stolzenberg, Ross and Marta Tienda. 1997. “English Proficiency, Education, and the Conditional
Economic Assimilation of Hispanic and Asian Origin Men” Social Science Research
26/1:25-51
Thurow, Lester. 1975. Generating Inequality: Mechanisms of Distribution in the U.S. Economy
New York: Basic Books
Tomaskovic-Devey, Donald and Sheryl Skaggs. 1999. “An Establishment-Level Test of the
Statistical Discrimination Hypothesis” Work and Occupations 26/4:422-445
30
Tomaskovic-Devey, Donald and Sheryl Skaggs. 2002. “Sex Segregation, Labor Process
Organization, and Gender Earnings Inequality” American Journal of Sociology
108/1:102-128
U.S. Bureau of the Census. 2000. “Census 2000 Data for the United States”
http://www.census.gov/census2000/states/us.html
U.S. Citizenship and Immigration Services. 2002. “Statistical Yearbook of the Immigration and
Naturalization Service prior to the Fiscal Year 2002 edition”
http://uscis.gov/graphics/shared/aboutus/statistics/ybpage.htm
U.S. Bureau of the Census. 2000. “National Population Projections”
http://www.census.gov/population/www/projections/natproj.html
Veltman, Calvin. 1990. “The Status of the Spanish Language in the United States at the
Beginning of the 21st Century” International Migration Review 24/1:108-123.
2001-2002 California Workforce Survey Codebook
31
Table 1: Earning Differentials among the Four Racial Groups
Total
(N=774)
African Americans
(N=48)
Hispanics
(N=206)
Asian
(N=50)
Other Minorities
(N=17)
White
(N=445)
Average Income
(St. Deviation)
42,145
(34,670)
42,984
(60,189)
28,394
(23,807)
43,845
(33,297)
40,141
(31,762)
48,329
(34,144)
Percentage of white income
t test
(compared to white)
87.20
---
88.94
-1.04
58.75
-7.00***
90.72
-.889
83.06
-1.23
---
---
*** P < .001; F ratio = 12.287 (p< .001); adjusted R square = 5.6%
32
Table 2 Unstandardized Coefficients of OLS Regression of Income
Predictors
Model 1
Model 2
Model 3
Model 4
Model 5
Constant
8.777***
(.163)
8.975***
(.195)
9.461***
(.113)
9.973***
(.174)
9.222***
(.240)
-.168**
(.065)
-.195
(.102)
-.251*
(.103)
-.045
(.166)
---
-.044
(.075)
-.171
(.116)
-.242*
(.104)
-.013
(.172)
---
-.276***
(.060)
-.117
(.095)
-.335***
(.098)
-.148
(.168)
---
-.315***
(.066)
-.187
(.112)
-.293**
(.111)
-.066
(.180)
---
-.035
(.068)
-.259
(.154)
-.304**
(.096)
-.187
(.167)
---
.394***
(.050)
.007**
(.002)
.409***
(.053)
.007**
(.002)
.250***
(.052)
.006**
(.002)
.340***
(.057)
.012***
(.002)
.169**
(.053)
.003
(.002)
.903***
(.122)
0.745***
(.118)
.391***
(.108)
.232*
(.112)
---
.709***
(.160)
.551***
(.157)
.203
(.148)
.058
(.153)
---
---
.617***
(.153)
.381*
(.149)
.152
(.137)
.039
(.138)
---
.258***
(.053)
.019
(.060)
.079***
(.018)
.246***
(.055)
.015
(.062)
.078***
(.018)
-.353**
(.130)
-.106
(.158)
---
---
.156**
(.054)
.027
(.057)
.035
(.018)
-.228*
(.109)
.012
(.136)
---
Individual Characteristics
Hispanics
Asians
Blacks
Others
White (reference group)
Male
Age
Human Capital Elements
Graduate
BA
Some college
High school
Less than HS (ref.)
Employer-paid job training
Self-paid job training
Work tenure
Home language: Spanish
Home language: Others
Home language: English (ref.)
---
---
---
33
34
Job characteristics
Full time work
Model 1
Model 2
---
---
Supervisory
Union membership
Occupation-managerial
Occupation-secretary
Occupation-machine operators
Occupation-craft
Occupation-farming
Occupation-service
Occupation-PT (Ref.)
Model 3
.891***
(.067)
.158***
(.023)
.147***
(.060)
-.073
(.084)
-.379***
(.065)
-.684***
(.098)
-.469***
(.096)
-.779***
(.175)
-.674***
(.082)
---
Workplace characteristics
Workplace size
Workplace independence
Non-profit public
Non-profit private
Profit (Ref.)
Industry-agriculture
---
---
---
--34.8%(13)
753
--32.2%(15)
696
--51.8%(15)
632
Industry-manufacturing
Industry-finance
Industry-personal service
Industry-professional service
Industry-retail
Industry-whole sale
Industry-transportation
Industry-public admin. (Ref)
Model R2 (df)
Number of cases
Table 2 continues
35
Model 4
---
Model 5
.764***
(.071)
.133***
(.023)
.082
(.071)
-.134
(.082)
-.256***
(.070)
-.358**
(.120)
-.272*
(.110)
-.340
(.264)
-.349***
(.095)
---
.099***
(.021)
-.028
(.063)
-.165
(.086)
-.246*
(.106)
---.442**
(.151)
-.418**
(.152)
-.081
(.143)
-1.096***
(.254)
-.222
(.113)
-.795***
(.147)
-.611**
(.226)
-.244
(.144)
--27.3%(18)
725
.056**
(.020)
-.109
(.057)
-.251**
(.085)
-.267**
(.094)
---.132
(.152)
-.248
(.144)
-.084
(.132)
-.713**
(.231)
-.216
(.111)
-.363**
(.140)
-.142
(.204)
.063
(.144)
--59%(36)
563
Numbers in parenthesis are standard errors
*P<.05; **P<.01; ***P<.001 (two-tail test)
36
Table 3 Mean Value Differences in Human Capital Factors between White
and Hispanics
Variables
Hispanic/N
(St.d)
12.13/243
(3.45)
.23/243
(.42)
.27/243
(.44)
.34/243
(.48)
.11/243
(.31)
.05/243
(.21)
6.69/217
(1.70)
.37/217
(.48)
.15/217
(.36)
.69/176
(.46)
.29/176
(.45)
.17/176
(.13)
Overall education
Less than high school
High school
Some college
BA
Graduate
Work tenure
Employer-paid job training
Unpaid job training
Speak English at both home and work
Speak Spanish at home and English at work
Speak other language at home and English at work
37
White/N
(St.d)
14.71/620
(1.89)
.02/620
(.13)
.18/620
(.38)
.35/620
(.48)
.24/620
(.43)
.22/620
(.41)
7.26/510
(1.55)
.64/510
(.48)
.25/510
(.44)
.99/615
(.11)
0/615
(0)
.01/615
(.11)
Hispanic-White
(t test)
-2.58***
(-14.02)
.21***
(11.17)
.09**
(2.99)
-.01
(-.26)
-.13***
(-4.29)
-.17***
(-6.15)
-.57***
(-4.41)
-.27***
(-6.94)
-.10**
(-2.95)
-.30***
(-14.78)
.29***
(16)
.16***
(16.31)
Table 4 Unstandardized Coefficients of OLS Regression of Income by Four
Races*
Coefficients
Blacks
Asians
Hispanics
White
Less than high school (Ref.)
High school
-1.067
(.795)
.241
(.196)
.405
(.905)
.602
(.883)
.101
(.865)
1.520
(.724)
.174
(.215)
--
--.196
(.765)
-.425
(.493)
-.019
(.429)
.548
(.925)
.453
(.343)
.187
(.390)
.106
(.195)
-.889
(1.002)
.004
(.382)
-40
-.126
(.191)
-.106
(.198)
.230
(.240)
.493***
(.013)
.120
(.114)
-.025
(.134)
.024
(.033)
-.203*
(.102)
-.381
(.326)
-123
-.037
(.193)
.292
(.191)
.472*
(.198)
.664***
(.065)
.125*
(.061)
-.038
(.064)
.038
(.021)
--
t test: Hispanic –
white
-.089
(.328)
-.398
(-1.447)
-.242
(-.778)
-.171**
(2.579)
-.005
(-.039)
.013
(.088)
-.014
(-.325)
--
-.300
(.362)
-349
-.081
(-.166)
---
Some college
BA
Graduate
Employer-paid training
Unpaid-training
Work tenure
Speak Spanish
Speak other languages
Speak English (Ref.)
Total cases
1.452
(1.590)
-39
*Space restriction prevents display of coefficients and standard errors for all other control variables
including individual, job and workplace characteristics. Those information is available upon request to the
first author.
Numbers in parenthesis for the last column are t test values. Numbers in parenthesis for other columns are
standard errors.
*P<.05; **P<.01; ***P<.001 (two-tail test)
38
Appendix: Item Constructions for Independent Control Variables
Variable names
Measuring Items
Coding Methods
Gender
Are you male or female?
Male=1
Female = 0
Age
How old were you on your last birthday?
Respondents’ actual age in years
Full-time work
Are you currently working for full time (35 +
hrs/wk) or part time?
Full time = 1
Part time = 0
Job Supervisory
As part of your job, do you (1) supervise the
work of other employees? (2) Influence the pay
or promotion of the people you supervise? or (3)
Hire and fire the people you supervise?
Agreement to each statement equals 1,
and then results are summed up, thus
producing a scale from 0 to 3.
Union membership
Do you currently belong to a labor union?
Respondents
occupations
What is your job title called?
Yes = 1
No = 0
A multiple dummy variable including
the following groups: managerial,
professional and technical, service,
secretary, machine operator, craft, and
farming. Professional/technical is the
reference group in regression
Workplace size
About how many people are employed where
you work?
1: fewer than 10
2: From 10 to 50
3: From 51 to 100
4: From 101 to 1,000
5: Over 1,000
Workplace
Independency
Is the place where you work part of a larger
company?
1: Independent
2: Yes, dependent
Workplace types
Do you work for a business, a government, or a
non-profit organization?
A multiple dummy variable where
business is reference group
Workplace industries
What kind of business or industry do you work
for at this job?
A multiple dummy variable including
the following groups:
agriculture, manufacturing, finance,
personal services, professional services,
retail, wholesale, transportation, and
public administration. Public
administration is the reference group
39