English-Language Proficiency and Personal Employment Status

 

 

James Krein

The University of Akron

Department of Economics

Senior Project

May 11, 2007

 

 

 

Abstract

 

 

Using a sample of males ages 25-64, from the U.S. Department of Education’s National Assessment of Adult Literacy of 2003, this study explores how English-language proficiency affects labor market outcomes.  Using the Multinomial Logit method, the data is regressed to find the probability of two employment outcomes, self-employed and unemployed with respect to individuals who are employed by others, labeled in this paper as “employees”.   The empirical results show that increased English-language proficiency increases the likelihood of an individual having an employment status of self-employed with respect to “employee” status.  Alternatively, as English-language proficiency decreases, the probability that one will have an employment status of unemployed increases with respect to “employee” status as well.

           

 

 

The author would like to thank all individuals who provided constructive comments concerning the content of this paper.  Specific recognition must be given to Dr. Francesco Renna, for his continual guidance throughout the development and writing of my research, to Dr. Elizabeth Erickson, for taking the time to provide additional critiques, and Dr. Steven Myers, for econometrics assistance and for providing important policy explanations.

 

 

 

I.  Introduction:

Languages other than English have been a part of American culture since the nation’s inception.  The argument whether to institute a national language dates back to the earliest period of American history.  In 1780, founding father, John Adams, proposed for, “an official academy” to “purify, develop, and dictate the usage of,” the English language.[1]   Nearly a century and a half later in 1920, then President, Theodore Roosevelt exclaimed, “We have room for but one language in this country, and that is the English language, for we intend to see that the crucible turns our people out as Americans, of American nationality, and not as dwellers in a polyglot boarding house.”[2] 

Over the past twenty years, as the number of both legal and illegal immigrants from all across the globe has risen, the debate concerning the necessity to implement English as the official language of the United States has resurfaced as a topic of much deliberation.  In 1996, House Representative Bill Emerson (MO) designed a bill under the name of the “English Language Amendment” that sought to make English the official national language of the United States.  The bill passed through the House by a 259-169 vote, but was never addressed in the Senate.[3]   If such a bill would have been voted into federal law, the economic consequences for non-English speakers could be costly, and are currently unknown.  This paper investigates one such consequence, how English-language proficiency affects an individual’s current personal employment status using data from the National Assessment of Adult Literacy of 2003.    

From the conclusions that come from the regression results that shows how much English-language proficiency shifts the probability of a particular employment status, policy makers would be able to assess how previously discussed legislation would impact the U.S. labor force.  The general conclusion drawn from past research is that as one’s English-Language proficiency increases, so does the likelihood that the individual will be self-employed.  Many of these same studies, and others on the returns to self-employment, show that the self-employed earn more on average than employees who work for someone else do in the U.S. labor force.  When we combine both of these results, we find that there is incentive for poor English speakers to improve upon their skills and for non-English speakers to learn the language.  Conclusions drawn from the research carried out in this paper can be compared with those from the previous research to examine how robust previous theory is.

 

An essential skill, needed to talk with public officials, make a transaction at the bank, obtain personal services, and to above all hold a job, is the ability to communicate with others, especially co-workers as well as employers.  Census data from 2000, that focused on language-use shows that 18% of the total U.S. population, or 47 million people, reported that they spoke a language other than English while in the household.[4]  Of these 47 million, 21.4 million reported that they spoke English less than well.[5]  As population demographics have changed to include a growing number of non-native monolinguals and bilinguals, communication skills are becoming progressively more and more important.  Considering that these changes have decreased the English-language proficiency average of the United States, the English-language proficiency of potential workers has become a weighty determinant for employers in the United States. 

Another way in which the research done throughout this paper contributes to the broader study of employment status predictors is that its results can be used as a forecast of how a national English-Only Law (EOL) could restrict a poor speaker’s ability to acquire and maintain the employment status they desire.  It provides the continuing argument over English-Only Laws an economic context.   Not yet a federal policy, although readily discussed due to both legal and illegal immigration, EOL’s have been included as amendments in 24 state constitutions from Nebraska back in 1920, to Arizona just past fall, in 2006.[6]  Written into these laws are articles that eliminate these states from having to providing multiple language provisions for public services, such as state tax forms, benefit information, medical records, loan applications, election ballots, and a number of online state services and references.  Also, it is not mandatory for schools in these states to have bilingual teachers for non-English speaking students, which limits the capacity of these students to learn.  If the impact of EOL’s extends into the classroom, non-proficient students would be placed at a disadvantage and may be unable to acquire skills that employers demand.

So for these individuals, does their lack of English-language proficiency diminish the chance that they will be hired?  If a restriction does exist, are poor English-language speakers more likely to turn toward self-employment for income, or are the barriers to self-employment so much to overcome, that they fall into unemployment?

Literature Review: 

 

Fairlie (2005) investigates both the determinants of self-employment and its returns for male and female youths.  Data from the 1979 National Longitudinal Survey of Youth (NLSY79), a set of surveys designed to gather information over three decades on the labor market activities of several groups of men and women, concluded that white youths were more likely to be self-employed than blacks or Hispanics, and Hispanics more than blacks.  Furthermore, this order holds true for both genders, though at a lower percentage for females compared to males.  The second part of the study discussing the returns to self-employment found that for males, average earnings were $14,042 higher among the self-employed than wage or salary workers, and $4,979 higher for self-employed females.

Another previous study compares the formal self-employment rates of Mexican immigrants in the United States with that of Mexicans in Mexico.  Fairlie and Woodruff (2006) apply census data from both countries to draw several conclusions.  Most importantly for the research done in this paper, the authors suggest that for both male and female Mexican immigrants, English fluency increases the likelihood that an individual will be self-employed; as does residence in an “ethnic enclave.”[7]   The authors also found a large gap between Mexican self-employment domestically and Mexican immigrant self-employment in the United States (22.1% domestically and 6.2% in the U.S.).  This gap, according to the authors, can be attributed to, “differences in the structures of the economy [of the United States and Mexico] and would be even larger if not for the characteristics of the U.S. population – mainly being older.”[8]  They also found that Mexican’s who were self-employed in the United States earned less than those self-employed domestically when compared to the average income of each country.

Dávila and Mora (2004) study how English-language fluency affects the earnings of self-employed foreign-born males from 1980 to 1990.  Using the Integrated Public-Use Microdata Series (IPUMS), a U.S. census database provided from a prior study by Ruggles and Sobel (1997), the authors found that for observants with the same English-language fluency, earnings were higher in 1990 than in 1980 after adjusting for inflation.  The papers concluding remarks point to the, “intolerance of linguistic pluralism and higher consumer, supplier, and employee discrimination levels against foreign-born entrepreneurs in the early 1980’s,” and then the increased acceptance of foreign-born immigrants in the 1990’s to describe the earnings difference.[9]

In an expansion of their previous research, Dávila and Mora (2005) further conclude that local linguistic isolation (such as “ethnic enclaves”) promotes self-employment among English fluent immigrants.  This may occur because the immigrant is able to service both customers with the same ethnic and linguistic background as themselves and English monolinguals.  Once again, this second study also looked at discrimination, although this time to those who had employee status and had limited English proficiency.  A second conclusion drawn by the authors, the opposite of their first, is that for individuals with limited-English proficiency, linguistic isolation negatively relates to entrepreneurship, and for immigrants who are increasingly English proficient, their probability of employment, both as an employee and as an entrepreneur increases. 

Much of the remaining economic literature concerning the relationship between employment status and English-language abilities compares the wage returns for each labor force group.

 

            The rest of the paper is as follows, Section II will present the methodology of the paper, which will include a description of the dataset, the presentation of the Multinomial Logit model, and an explanation of the model.  Section III will provide the statistical analysis that contains the empirical analysis of the paper, regression results, and tables of the AM software output.  Section IV is where all conclusions that can be drawn from the results will be presented.  This final section will also include a discussion of possible policy implications, the limitations of the study, and where future studies may advance the topic.

 

 

II.  Methodology/Theoretical Background:


Data Description:

 

This paper utilizes data from the U.S. Department of Education’s, National Assessment of Adult Literacy of 2003, (NAAL), provided by their National Center for Educational Statistics division.  The NAAL data is a cross-sectional, nationally representative sample containing information on English proficiency, literacy, and second language proficiency.  Contained in the Assessment is information gathered from just over 26,000 participants, randomly selected from both the general U.S. population, as well as a national sample of prison inmates, all between the ages of 16 and 65.  It is the fourth and most recent assessment of adult literacy in the United States funded by the Federal government.[10]  

For the purpose of this study, both females and the prison sample have been eliminated from the dataset.  Employment status cannot be derived for the prison population, and females have the alternative choice of bearing child and the data cannot be manipulated to measure for this.  Other notable variable manipulations made to the dataset is that the 16-24 age group has been dropped from study, as most of the other research outside of youth study’s set an age minimum of 25.  “Excellent” is used as the base for the Health dummy variables, “still in high school” has been eliminated for the Personal Educational Attainment variable because the study eliminates the 16-24 age group, and “employed, but not at work” has been removed from the Employment Status variable because we have no measure for this case imbedded in the dependent variable. The overall number of observations for this study drops to 12,219.

In conjunction with the literacy Assessment, a personal background questionnaire was taken by each participant.  These questionnaires asked for information concerning the participant’s general background, their language background, educational background and attainment, political participation, social participation, labor force participation, their engagement in literacy activities, and other demographic information. 

The exact purpose for the creation of this assessment by the Department of Education is listed in six objectives, jointly however, the main objective is to define and measure language literacy in the United States.[11]  In 1985, during the creation of the first U.S. literacy study, the following formal definition of literacy was adopted:

“Using printed and written information to function in society, to achieve one’s goals, and to develop one’s knowledge and potential.” [12]

This will be the definition for literacy used throughout the entirety of this paper.  Along with formation of this definition for literacy, seven tests were added within the Assessment to separate for different types of literacy.   The first of the three types of literacy is Prose literacy.  Prose literacy involves, “the knowledge and skills needed to understand and use information from texts that include editorials, news stories, poems, and fiction; for example finding a piece of information in an article, interpreting instructions, or being able to infer a theme or contrast a view expressed in written materials.”  The second form is Document literacy, and it concerns “the knowledge and skills required to locate and use information contained in materials that include job applications, payroll forms, maps, tables, and graphs.”  A few examples include entering the correct information on an application of work form, reading from a table or graph and being able to pull out the desired information, or to be able to create directions from information read off a map.  Lastly, Quantitative literacy, involves “the knowledge and skills required to apply arithmetic operations, either alone or sequentially, to use numbers embedded in print materials, and to describe how they would be set up in order to solve a problem.”  Brief examples would be balancing a personal checkbook, calculating a tip, completing a quantitative work form, or to determine the amount of interest earned from a bank loan or from an advertisement.[13]

            Within the Assessment, there is a rubric created for the use of the U.S. Department of Education to score the information after processing that reflects the performance level of each applicant for each of the three literacy type tests.  For simplification, this paper will just quantify these test scores by number of questions answered correct.


Presentation of the Model:

This paper will regress the data collected from the NAAL 2003 on the following model where an individual’s predicted employment status is the binary dependent variable:

Employment Statusi = β0 + β1X1i + β2X2i + β3X3i + εi

 

 

where, β0, β1, β2, β3, are all parameters to be estimated, i is a sub-script representing individual i, and εi is the error term.  X1i is a group of Personal Characteristics for each individual, including race, age, marital status, and immigrant status (country of birth, age of arrival to the U.S., and years living in the U.S.).  The inclusion of these variables allows for separate characteristics to me analyzed as theory from previous works shows that they too, have an effect on employment status.  X2i lists a set of Educational Attainment Characteristics for the surveyed and both of their parents.  These three variables are included because educational attainment goes a long way in determining what type of job one can obtain, which influences their economic status.  Parent’s educational attainment is included because often a child’s highest level of education is correlated to that of their parents.  The final vector, X3i, represents collected English Language Proficiency Characteristics for each observation in the study.  These include the results from three of the seven NAAL 2003 tests (Prose literacy [2 tests], Document literacy [3 tests], and Quantitative literacy [2 tests]), data on when the participant learned to speak English if at all, and the frequency in which they engage in English literacy building activities in a given week.  These English-language measures, in which actual ability is quantified by tests contained in the NAAL 2003, is the reason why I am using this dataset over others that only contain less detailed variables for proficiency.

Within vector X3i, only three of the seven tests given in the NAAL 2003 have been included in the regression in order to reduce the collinearity.  One goal of this research was to gauge each type of literacy tested, so inclusion of at least one test of each type was necessary to achieve this goal.  The three tests selected, one form of each type of test (Prose, Document, and Quantitative) were chosen after finding the correlation amongst all seven tests.  The one test with the least correlation to all of the others from each type was selected for the model.  Correlation ranged from -4.8% to 50.8% for Document Test 3, -7.3% to 76.2% for Prose Test 1, and from 6.55% to 44.3% for Quantitative Test 2.  Also, all three weekly literacy habit variables were included as their correlation was between 31.4% and 35.2%.  All measures of weekly literacy activity engagement were kept in the regression as well.  A test of correlation amongst these variables and the NAAL 2003 test scores provided only a maximum correlation of 25.8%.  Still, however, the all variables were kept in the model knowing that some collinearity does exist.

            Whereas all of the included variables above are observable from the assessment and survey results, there are several important but unobservable factors missing from the NAAL 2003 that can influence one’s employment status.  Factors such as time preferences, personal and family wealth, and weekly or annual wages earned are just a few.  Because of these unobservable factors, the limitation of omitted variable bias will exist.    


Explanation of the Model:                                                                       

 

            The model presented above will be run as a Multinomial Logit, where Employment Status, the dependent variable can take on three values, coded (0) for those who have employee status, (1) for those “self-employed”, and (2) labeled as “unemployed”.[14]  This variable is coded as such so that (1) self-employed can be directly compared against the (0) worker/employee outcome.  This Multinomial Logit model emulates the empirical techniques used in a vast array of other Logit studies estimating the probability of employment outcome based on a host of observable characteristics.  The procedure is based on the cumulative logistic probability function, where the estimates provide the expected probabilities that a given independent variable produces a particular outcome compared to another base outcome. 

            After running the Multinomial Logit procedure on the data, the Marginal Effects of the data will be calculated with the estimated probabilities of each variable in the model.  What these marginal effects present are the probabilities that for each variable, a certain employment outcome will occur with respect to the base employment outcome, which is in this model, someone who is an employee not working for themselves.  The values for these probabilities range from 0% to 100%, and can take on both positive and negative values.  

            All variables included in the model are labeled and defined on the “Descriptive Statistics” Table located in Appendix A.  Their hypothesized relationship to the dependent variable is located in the “Hypothesized Effects” chart, which can be found in Appendix B.  Dummy variables have been created for most of the independent variables in order to see the true effect of each throughout a different range of values. 

 

 

III.  Statistical Analysis:


Discussion of the Results:

 

(a) English-Language Proficiency and its Marginal Effects

 

Based on the results from the Multinomial Logit model ran, one can clearly see that an individual’s English-language proficiency provides limited insight as to the particular employment status outcome of that individual.  Of the English-language variables regressed, those that measured unchangeable life characteristics and weekly literacy habits proved to be more significant than the proficiency tests administered in the NAAL 2003.  As a reminder, all the results for both “self-employed” and “unemployed” outcomes are with respect to an individual who has an outcome of “worker”/employee (Employment Status = 0).  

The age that the participant learned to speak English and whether English was their first language, the two uncontrollable variables, had significant coefficients for only two self-employed dummy variables.  When an individual was self-employed and spoke both English and another language, and if the individual learned English after the age of eleven, the coefficient was significant to the 90% level and 99% levels respectively.  Speakers whose first language was only English or any of the dummy variables that included English as a language spoken increased the probability that their employment status outcome would be self-employed by between 1.28% and 3.15%.  The only way in which one’s probability of being unemployed increased was for those who spoke both English and Spanish, but only by .47% compared to those who work as employees.  All other language combinations or knowing English only as their first language saw a probability decrease toward having an outcome of unemployed.   Only speaking Spanish and another language decreased one’s likelihood of being self-employed by 4.71%.  This result mimics the results of the other studies mentioned earlier in this paper’s introduction where increased English-language proficiency increases the likelihood that someone is self-employed.

Interesting as well is the result produced between the age one learned to speak English and the unemployed status outcome.  Only not learning English after the age of 11 decreases the likelihood that one will be unemployed compared to someone who is an employee.  As one gets older as they learn English, the chance that they will be unemployed increasingly decreases.  Not knowing English or learning it at any age increases the probability that one will be self-employed, from between 2.58% and 11.25%.  This result is especially odd because it concludes that those who cannot speak English still have an increased chance of being self-employed, a contradiction to the results of other variables and past research.  Also the output finds that younger learners see a reduced probability of being self-employed than working as an employee.

Weekly literacy habits also presented significant coefficients for the categories covering frequency of newspapers/magazines and books read in English for the unemployed outcome, and significance was found amongst the letter frequency variable for the self-employed employment outcome.  Reading any of the type of literature at all produced positive marginal effects for the unemployed outcome, meaning that any literacy activity increases the chances one would be unemployed as compared to being someone else’s employee.   This may be due to the fact that the unemployed have more time to engage in these activities, but this is only speculation.  Reading newspapers/ magazines reduced the probability that when compared to an employee, a worker could be self-employed, especially for those who read them a few times a week, the likelihood that they would be self-employed decreased by 12.64%.  The smallest decrease in self-employment probability came for those who read the newspaper or magazines daily.  This is probably due to the fact that an individual that is self-employed needs to be kept up-to-date on the latest local and business happenings to compete.

Only daily and occasional book readers see their chances of being self-employed improve, all other weekly habits for reading books produces decreases likelihood of that individual being self-employed.  Probabilities do not ever decrease for the marginal effects of both self-employed and unemployed for letter/note readers.  Everyday readers have the greatest prospect of being self-employed than being an employee at just under 24%.  Similarly, those who read letters or notes on a daily basis also have the greatest chance of being unemployed when compared to employees, by 12.31%.

Results for the actual English-language assessments in the NAAL 2003 were for the most part insignificant.  Document test 3 was by far the least significant, and when compared to employee status, both self-employed and unemployed saw increased outcome probabilities.  This means that for every correct answer, the probability that one would be self-employed increases by 1.98%, and for unemployed, increases by only .001%; hardly a change at all. 

Prose test 1 had a significant coefficient for the unemployed outcome to the 90% level.  With respect to being an employee, every correct answer increased one’s probability of being self-employed by 3.01% and decreased ones chance of being unemployed by 1.05%.  For Quantitative test 2, the parameter estimate coefficient for self-employed was significant to 99% level.  Different for this test, however, is that the likelihood that an individual will have an employment status outcome of self-employed slightly decreases, while the possibility that one would be unemployed increases by even less.  Viewed against someone working as an employee, the probability of the self-employment outcome decreases by .47% with every correct response, and increases the chance of an unemployed outcome by .19%. 

 

 

 

(b) The results and Marginal Effects of other variables in the model

 

Other non-English language variables in the model that had significant coefficients in the regression include Age (those over 40), residence in the West[15], Personal Educational Attainment at the High School graduate and High School equivalent levels, being single, Personal Health Status above fair, and some of the dummy variables contained in the two Parent’s Educational Attainment variables.  Surprisingly, only the dummy variable for Black under race had a significant coefficient for the unemployed outcome. 

Results show that as one gets older, the likelihood increases that they will be both self-employed and unemployed increases.  For those in the age range of 50-65 years old, the chance one will be self-employed increases by 73.71%, and the probability that a person will be unemployed increases by 22.70%.  Personal educational attainment lessens self-employment chances throughout all attainment levels when compared to the base employee outcome.  Having a vocational or trade certificate decreases self-employment outcome by only 2.85%, while some one who only has a high school diploma sees their self-employment probability decrease by 18.94%.  On the flip side, having a degree at any level (2 year, 4 year, 4+) increased one’s chances of being unemployed.  This increase is greatest for those with graduate degrees, 17.28%, and could exist because these workers are waiting for a job with their desired salary and skill level.  The greatest decrease in likelihood that one is unemployed came for those who had less than a high school diploma; speculation is that this occurs because those without increased skills are forced to take low paying service sector jobs, and there are plenty in the market that need to be filled.

Living in the Northeast decreases one’s likelihood of being self-employed or unemployed, but for those living out West and in the South, the probability that one is unemployed increases by.  Self-employment likelihood increases by nearly the same amount for those that live in the South and West with respect to those who are employees.  Finally, being black increases the odds that one is unemployed by 13.46%.  Only whites see the probability that they will be unemployed decline.  All three races measures are less likely to have an employment outcome of self-employed; the decrease is minimal for whites, while for both blacks and Hispanics, the decrease in the probability that they will be self-employed is over 9%.

Complete results for all variables can be found in Appendix C later in the document.

 

 

Results Summary Table:

 

 

English language Proficiency Characteristics

Variable

Self-employed

M. E. (Self)

Unemployed

M. E. (Un)

 

 

 

 

 

Document Test # 3

 

 

 

.198

(.248)

1.98%

.009

(.208)

.001%

Prose Test # 1

 

 

 

.141

(.010)

3.01%

-.098*

(.063)

-1.05

Quantitative Test # 2

 

 

 

-.190***

(..071)

-.47%

.007

(.069)

.19%

Frequency one reads a the newspaper or a magazine in English

Everyday

 

 

Few times a week

 

 

Once a week

 

 

Less than once a week

 

 

 

 

 

 

-.310

(.248)

 

-.354

(.287)

 

-.183

(.264)

 

-.334*

(.266)

 

 

 

-.14%

 

 

-12.64%

 

 

-3.65%

 

 

-5.38%

 

 

 

.566**

(.284)

 

.475*

(.286)

 

.375

(.313)

 

.090

(.286)

 

 

 

28.33%

 

 

18.13%

 

 

8.00%

 

 

1.55%

Frequency one reads a book in English

Everyday

 

 

Few times a week

 

 

Once a week

 

 

Less than once a week

 

 

 

 

 

.003

(.171)

 

.143

(.187)

 

-.180

(.199)

 

-.115

(.186)

 

 

.12%

 

 

4.31%

 

 

-3.12%

 

 

-3.87%

 

 

.324*

(.176)

 

.381**

(.178)

 

.107

(.206)

 

.219

(.158)

 

 

14.38%

 

 

12.27%

 

 

1.98%

 

 

7.88%

Frequency on reads a note or letter in English

Everyday

 

 

Few time a week

 

 

Once a week

 

 

Less than once a week

 

 

 

 

.505*

(.270)

 

.280

(.280)

 

.127

(.313)

 

.492*

(.285)

 

 

23.83%

 

 

8.62%

 

 

2.04%

 

 

10.45%

 

 

.244

(.234)

 

.215

(.227)

 

.108

(.261)

 

.148

(.229)

 

 

12.31%

 

 

7.08%

 

 

1.86%

 

 

3.36%

Age learned to speak English

Does not speak English

 

1-10

 

 

11+

 

 

 

 

 

.347

(.487)

 

.114

(.216)

 

.939***

(.360)

 

2.58%

 

 

2.72%

 

 

11.25%

 

.366

(.360)

 

.029

(.147)

 

-.052

(.303)

 

2.91%

 

 

.74%

 

 

-6.66%

Language spoken before school

English Only

 

 

Eng. and Spanish

 

 

Eng. And other

 

 

Spanish only, with other

 

 

 

.099

(.362)

 

.278

(.470)

 

.452*

(.325)

 

-.243

(.427)

 

 

3.15%

 

 

1.28%

 

 

2.33%

 

 

-4.71%

 

 

-.143

(.279)

 

.077

(.351)

 

-.104

(.261)

 

-.041

(.319)

 

 

-4.86%

 

 

.47%

 

 

-.57%

 

 

-.85%

 

* t-value significant to the (α = .10) level

** t-value significant to the (α = .05) level

*** t-value significant to the (α = .01) level

(Standard errors are in parentheses)

 

 

IV.  Conclusions:

 

            From the results above, one can conclude that there is evidence from the English literacy tests in this study to show that English-language proficiency does have an effect an individual’s employment status outcome.  The results in this study also coincide with the results from past studies that show for someone with increased English-language proficiency, the probability that the individual will be self-employed increases.  From the NAAL 2003 tests we see that for both the Prose and Document tests, the self-employment outcome increases as the number of questions answered correctly increases.  Only the results from the Quantitative test show that one is less likely to be self-employed than unemployed, but only by the slightest of probability.  Alternatively, little can be drawn from the results of the marginal effects for the unemployed outcome; only one variable shows a probability change greater than 1%, and its only 1.05%.

            The three variable for frequency of weekly English literacy engagement have either completely different or exactly the same results for both the self-employed and unemployed outcomes.  For the newspaper/magazine frequency category, decreased weekly readership decreased one’s probability of being self-employed and unemployed.  A similar conclusion can be drawn from book readership as well, increased activity increases unemployment, perhaps because one has more free time available, but alternatively here, self-employment rates increase with readership.  This could mean that with this increased activity, individuals are using these literacy materials to improve the proficiency and increased rates of self-employment are the result.  This conclusion agrees with those made in past studies.  The results from letter and note readership say the same thing that increased activity leads to greater chances of self-employment.

From the results of the participants known first language variable, this study finds the greatest positive marginal effect toward being self-employment when the individual learns English at an older age.   The conclusion that not knowing English at all leads to the greatest probability that one will be unemployed can also be made from the regression results.  These two results virtually mirror the conclusions drawn in Dávila and Mora (2005).  Also in connection with the results of Dávila and Mora (2005), proficiency in both English and another language increases rates of self-employment.  This would coincide with the theory that bilingualism benefits the self-employed immigrant living in an ethnic enclave because they are able to serve both customers of their own ethnicity and English only speaking American citizens.  Ultimately, what this study has found is that when an individual who works as an employee is our base employment status, as one’s level of English-language proficiency increases as measured by the variables in the study, the logistic probability that they will be self-employed increases and the logistic probability that they will be unemployed decreases.

 

Policy Implications:

 

The findings of this paper are policy relevant, especially when it comes to the role of self-employment and the possible consequences of a nation-wide English-Only law.  Taken directly from the conclusions made above, these results illustrate how important learning the English language can be for those who may have been denied employment by private employers.  Self-employment is one way in which these individuals are able to sustain themselves and since previous research indicates that the self-employed earn more than employees, it is a way for these individuals to improve their financial, social, and economic standing.

Such a law would also overturn an August 2000 Executive Order signed by President Bill Clinton, which sought to, “improve access to federally conducted and federally assisted programs and activities for persons who, as a result of national origin, are limited in their English proficiency (LEP).”[16]  Beyond its formal definition, this decree promotes the improvement of LEP speakers by making it unlawful for employers to discriminate against these individuals as part of the Civil Rights Act of 1964.  Although not strictly implemented, the order does provide those who know of it with an equalizer when applying for a position they are qualified for.  This is often looked upon negatively by those who are not afforded a similar advantage, which in return may cause these individuals to further press for a national EOL.

The implications of a constitutional amendment that would create a federal English-Only law could be devastating for those who have not learned the language or are not completely proficient.  If in 2003, decreased proficiency increased the probability of that a person is unemployed, any additional measures could be crippling.  Filling out a job application, completing tax forms, or opening a bank account or applying for a loan would be an experiment in futility.  The extra hurdle could negate a segment of the nation’s population from the formal economy at a loss of a substantial amount of money each year.

 

 

Limitations:

 

The AM Software package provided by the U.S. Department of Education’s, National Center for Educational Statistics, has many limitations.  It is limited in its ability to run certain procedures outside of an OLS, Probit, and Logit regression.  The program offers no way to calculate marginal effects, so they have to be done with pad and paper, and this may be the reason for any miscalculations presented in this study, though the calculations were made multiple times to decrease errors.  Also, many relevant variables that could be of importance to the research done in this paper were not included in the first release of the dataset.[17]  This, of course, implies that Omitted variable Bias is present. 

The issue of collinearity addressed in earlier sections still remains a problem depending on the correlation limit that one uses when eliminating variables.  Between the three tests and the measures of weekly literacy activity engagement, correlation between English-language proficiency variables runs as high as 35%.  As stated in the explanation of the model, one goal of this research was to gauge each type of literacy tested, so inclusion of at least one test of each type was necessary to achieve this goal. 

 

Future Research:

            There is much room for future research in the area of Language proficiency and its affect on employment outcome.  For starters, with the data used I this study, one could change the base of the dependent variable to self-employed or unemployed and then would be able to compute the marginal probabilities for those who are employees.  Alternatively, with the increase of native-Spanish speakers coming into the United States, a reverse study on how Spanish-language proficiency impacts employment outcomes would be interesting.  Such a study could hold great importance for southern boarder states and parts of the Southeast. 

Another great addition to this area of interest will come when the complete dataset is released with all of the raw data by the Department of Education.  With it, future studies could separate the participants by states and see what the affect state English-Only Laws possibly have on employment status outcomes.  Along with this study, it would be interesting to see how many employers and employees know of President Clinton’s Executive Order 13116, and if so how it influences hiring practices.  A final study could include a measurement for the frequency of electronic media activity, with variables for television, radio, and computer/internet usage.  The possibilities for study extensions are numerous and would provide key clues as to the true importance of language in America.

References

 

Borjas, George J. “The Self-Employment Experience of Immigrants” NBER Working Paper No. 1942

 

Borjas, George J. “The Self-Employment Experience of Immigrants” Journal of Human Resources vol. 21, no. 4 (Fall 1986) Pages 485-506

 

Bregger, John E. “Measuring self-employment in the United States” Monthly Labor Review, U.S. Department of Labor, Bureau of Labor Statistics. January/February 1996. Pages 3-9

 

Dávila, Alberto and Mora, Marie T. “English-Language Skills and the Earnings of Self-Employed Immigrants in the United States: A Note” Industrial Relations vol. 43, no. 2 (April 2004) Pages 386-391

 

Fairlie, Robert W. “Self-employment, entrepreneurship, and the NLSY79” Monthly Labor Review, U.S. Department of Labor, Bureau of Labor Statistics. February 2005.  Pages 40-47

 

Fairlie, Robert W. and Meyer, Bruce D. “Ethnic and racial Self-Employment Differences and Possible Explanations” The Journal of Human Resources vol. 31, no. 4 (Fall 1996) Pages 757-793

 

Lofstrom, Magnus. “Labor market assimilation and the self-employment decision of immigrant entrepreneurs” Journal of Population Economics (15) 2005. Pages 83-114

 

Mora, Marie T. and Dávila, Alberto. “Ethnic group size, linguistic isolation, and immigrant entrepreneurship in the USAEntrepreneurship and Regional Development (17) September 2005. Pages 389-404

 

Woodruff, Christopher and Fairlie, Robert W. “Mexican Entrepreneurship” A Comparison of Self-Employment in Mexico and the United States” The Institute for the Study of Labor (IZA). IZA Discussion Paper No. 2039, March 2006.

 

Yuengert, Andrew M. “Testing Hypothesis of Immigrant Self-Employment” The Journal of Human Resources vol. 30, no. 1 (Winter 1995) Pages 194-204

________________________________________________________________________

 

American Civil Liberties Union. English Only Briefing Paper Number 6, 1996. Accessed through the ACLU official website: http://www.aclu.org/library/pbp6.html  Gathered April 27, 2007.

 

Hyon B. Shin with Rosalind Bruno. “Language Use and English-Speaking Ability: 2000” U.S. Census 2000 Brief, United States Census Bureau Issued October 2003. Pages 1-11

 

Roosevelt, Theodore. Personal Works, Vol. XXIV (Charles Scribner & Sons, New York) 1926.

 

U.S. English, Inc. www.us-english.org/ Gathered April 19, 2007.

_______________________________________________________________________

 

National Center for Educational Statistics, United States Department of Education     National Assessment of Adult Literacy of 2003, Public-Use Data

 

National Center for Educational Statistics, United States Department of Education     National Assessment of Adult Literacy of 2003, Public Use Handbook.

 

The American Institutes for Research. AM Statistical Software, © 1998, Version 0.06.03 updated 2003. Developed by the National Center for Educational Statistics for the U.S. Department of Education. Jon Cohen, software manager.

 

 

 

 

 

 

 


Variable


Appendix A: Descriptive Statistics

 
Variable Description

Minimum

Maximum

Mean

St. Dev.

N

Missing

EMPTYP

Employment status

0

2

0.453

0.771

14729

28

DAGEC

age

1

4

2.706

1.097

14729

0

DRACE

race/ethnicity

1

4

1.699

0.925

14729

2

DSEX

gender

0

1

0.462

0.499

14729

0

DCBIRTH

country of birth

1

2

1.161

0.368

14729

0

DARRIVE

age of arrival in the US

1

3

1.255

0.616

14729

19

D1STLNAG

language spoken before starting school

1

5

1.592

1.211

14729

1165

DENGAGE

age learned to speak english

1

4

1.408

0.792

14729

12

DEDATTN

educational attainment

1

9

5.033

2.441

14729

1171

DEDBFUS

education before coming to the US

1

3

 

 

14729

6

DLFORCE

labor force participation

1

5

2.632

1.791

14729

1177

DMED

mother's educational attainment

1

8

3.053

2.180

14729

1658

DFED

father's educational attainment

1

8

3.177

2.356

14729

2510**

DMARITAL

marital status

1

3

1.956

0.729

14729

46

DHEALTH

self-assessment of overall health

1

5

2.370

1.106

14729

25

DWORKHIS

work history

1

3

1.267

0.533

14729

1184

DLIVEUS

years living in the US

1

3

1.289

0.674

14729

1163

DRDENGPR

frequency of reading newspapers/magazines in english

1

5

2.092

1.285

14729

13

DRDENGBK

frequency of reading books in english

1

5

2.667

1.480

14729

13

DRDENGLN

frequency of reading letters/notes in english

1

5

2.136

1.370

14729

15

DOC1

document test 1

0

3

1.053

.358

14729

0

DOC2

document test 2

0

3

1.001

.446

14729

0

DOC3

document test 3

0

3

.944

.263

14729

0

PROSE1

prose test 1

0

3

.870

.644

14729

0

PROSE2

prose test 2

0

3

.968

.435

14729

0

QUANT1

quantitative test 2

0

3

.926

.298

14729

0

QUANT2

quantatative test 2

0

3

.845

.615

14729

0

** For Father’s Educational Attainment, the number missing is 2,510 therefore the total number of observations in the study is 12,219






Appendix B: Hypothesized Effects

 

Variable

Description

Units of Observation

Estimated Relationship

EMPTYP

Type of Employment/ Employment Status

0 = “Worker”/ Employee

1 = Self-employed

2 = Unemployed

 

This is the dependent variable of the model.

PERSONAL CHARACTERISTICS

DAGE

Age

1 = 16-24 year olds*

2 = 25-39 year olds

3 = 40-49 year olds

4 = 50-64 year olds

The older that one becomes, the more likely the probability that they will make the decision to become “self-employed”.  Older Americans have business experience and often have the finances to take the entrepreneurial track.

 

*Denotes that this age group has been eliminated from the regression.

 

DMARITAL

Marital Status

1 = Single, Never Married

2 = Married

3 = Separated, Divorced, or Widowed

It seems highly logical that an individual without the added burden of a loved one would be more willing to take on “self-employment”.  However, married couples may have more financing to use as initial start-up capital and would see the partnership increasing their likelihood of success.

DRACE

Race

0 = White

1 = Black

2 = Hispanic

I believe that whites will have a higher probability of being both “worker” and “self-employed” than both Blacks and Hispanics.  Because of language differences, and that is what the study looks at, it seems that Blacks will have higher representation in both of the employed outputs than Hispanics

DHEALTH

Overall Health

1 = Excellent*

2 = Very Good

3 = Good

4 = Fair

5 = Poor

The greater one’s overall health is, the more likely that they will not be “unemployed”.

 

* Excellent is used as the base Health dummy variable.

 

DARRIVE

Age of Arrival

1 = U.S. born

2 = 0-18 years old

3 = 19 + years

The probability of being “self-employed” will increase as the age of arrival lowers.  Older arrivals will not have the knowledge needed to be “self-employed” in the United States.  Older adults will most likely migrate toward “worker” jobs based on their level of education.

DLIVEUS

Years Living in the United States

1 = U.S. born

2 = 0-5 years

3 = 6 + years

Once again, it would seem that the longer one has lived in the United States, the higher likelihood that they would be “self-employed” compared to an individual who has been in the states for a shorter period of time.  Those who have been here longer would have learned U.S. business practices, and for lending purposes, could have established credit.  If one has just arrived to the U.S., it would be easy to assume that there is a high probability that they are “unemployed” or “worker” because the reason they moved to the U.S. was for employment purposes.

 

EDUCATIONAL CHARACTERISTICS

DEDBFUS

Education before coming to the United States

1 = Did not attend school/Primary

2 = Elementary

3 = Secondary +

This variable allows us to gauge education level before entering the United States.  As educational attainment increases before entry into the U.S., the probability that they will be employed either as a “worker” or “self-employed” increases.

 

DEDATTN

Educational Attainment

1 = less than/some High School

2 = GED

3 = High School graduate

4 = Trade or Vocational School

5 = Some College

6 = Associate’s/ 2-year degree

7 = College Graduate

8 = Graduate Studies                                                                  /Degree

As level of educational attainment increases, it seems more likely that an individual would have “worker” status because an individual who has sought a higher level of education would most likely be doing so to market themselves to other employers and not for “self-employment”.  One who is self-employed may have attended college, but if they knew that they wanted to be self-employed, it would be financially efficient to only take courses that meet their specific needs.  It would be expected that few individuals that have achieved anything greater than GED or High School graduate level would be “unemployed” unless there was a shortage of employment opportunities in the market.

DMED

Mother’s Educational Attainment

1 = less than/some High School

2 = GED

3 = High School graduate

4 = Trade or Vocational School

5 = Some College

6 = Associate’s/ 2-year degree

7 = College Graduate

8 = Graduate Studies /Degree

 

I am uncertain of what the effect of mother’s educational attainment will be.  Both for this variable and the following variable, father’s educational attainment, the result will be interesting, although I do not think t will be significant.

DFED

Father’s Educational Attainment

1 = less than/some High School

2 = GED

3 = High School graduate

4 = Trade or Vocational School

5 = Some College

6 = Associate’s/ 2-year degree

7 = College Graduate

8 = Graduate Studies / Degree

 

Again, just as for mother’s educational attainment, I am uncertain of what effect this variable will have on the individuals employment status.

ENGLISH LANGUAGE PROFICIENCY

DOC1

DOC2

DOC3

PROSE1

PROSE2

QUANT1

QUANT2

English Language Test Results

Three tests types:

(Prose, Document, and Quantitative)

 

Document Tests:

- 3 tests

- 26 questions per test

 

Prose Tests:

- 2 tests

- 20 questions per test

 

Quantitative Tests:

- 2 tests

- 19 questions per test

 

Scoring Rubric for all tests:

0 = Incorrect

1 = Correct

2 = Omitted

3 = Not Reached

 

As stated in the introduction, my hypothesis is that as the level of English proficiency decreases based on these assessment scores, the less likely an individual will be “self-employed”.  Lower proficiency, in this case a lower number of correct answers,  will subjugate that individual to a low wage “worker” position and increases the probability that they may be “unemployed” as well.

DRDENGPR

Frequency of reading newspapers and/or magazines in English

5 = never

4 = less than once a week

3 = once a week

2 = few times a week

1 = everyday

These three variables provide the study with insight as to how often the participant engaged in English Language proficiency building activities.  As one variable in the set that describes an individual’s level of English language proficiency, the general hypothesis that as frequency increases it becomes more likely that one will be “self-employed”.

DRDENGBK

Frequency of reading books in English

5 = never

4 = less than once a week

3 = once a week

2 = few times a week

1 = everyday

DRDENGLN

Frequency of reading letters/notes in English

5 = never

4 = less than once a week

3 = once a week

2 = few times a week

1 = everyday

DENGAGE

Age learned to speak English

1 = only speak English

2 = 0-10 years old

3 = 11 + years old

4 = do not speak English

The younger that an individual learned to speak English the more likely they will achieve increased educational attainment over those who learned English later.  This educational advantage will decrease the probability of the individual being “unemployed”.  Not being able to speak English at all will increase the likelihood that one is “unemployed” or has “worker” status, although at the lowest level.

1STLAN

Language spoken before starting school

1 = English only

2 = English and Spanish

3 = English and other language (not Spanish)

4 = Spanish only or with other language

5 = Other only

Not knowing English before starting school is another barrier that an individual would have to break through, even though I do not see this as a problem for a young student in primary school.  I see the real effect for older students who do not know English.  It seems that those who know English are more likely to be “self-employed” than those who did not at the start of schooling.  For those who did not know English before starting school, the probability of being “unemployed” should be higher.

~ All data points come from the National Center for Educational Statistics, United States Department of Education, National Assessment of Adult Literacy, 2003.

 

 

Appendix C: Complete Results Table

 

 

Variable

Self-Employed

M.E. (Self)

Unemployed

M.E. (Un.)

Intercept β0

1.89

 

2.02

 

 

Personal Characteristics

Age

25-39

 

40-49

 

50-65

 

.746***   (.248)

 

1.03***   (.251)

 

1.565*** (.248)

 

29.83%

 

30.91%

 

73.71%

 

-.001     (.137)

 

.328**   (.158)

 

.451*** (.172)

 

.0004%

 

10.52%

 

22.70%

Race

White

 

Black

 

Hispanic

 

-.033   (.195)

 

-.33     (.233)

 

-.345    (.351)

 

-1.53%

 

-10.16%

 

-9.29%

 

-.032     (.154)

 

.409***  (.150)

 

.018       (.220)

 

-1.58%

 

13.46%

 

.52%

Region where one lives

Northeast

 

South

 

West

 

-.039    (.158)

 

.136     (.127)

 

.263     (.164)

 

-1.17%

 

6.37%

 

6.54%

 

-.027     (.110)

 

.026      (.105)

 

.315***  (.113)

 

-.86%

 

1.30%

 

8.38%

Marital Status

Single/ Never Married

 

Married/ Living as Married

 

-.336**   (.147)

 

-.07        (.117)

 

-13.05%

 

-3.29%

 

 

-.085    (.099)

 

.081      (.084)

 

-3.53%

 

4.07%

Health Assessment

Very Good

 

Good

 

Fair

 

Poor

 

-.237*    (.121)

 

-.358**   (.158)

 

-.106      (.198)

 

.021        (.362)

 

-10.18%

 

-12.48%

 

-2.10%

 

.17%

 

.123*     (.070)

 

.006       (.097)

 

.209       (.157)

 

-.014      (.336)

 

5.64%

 

.22%

 

4.43%

 

-.12%

 

Educational Attainment Characteristics

Personal Educational Attainment

Less than/ Some H.S.

 

GED/ H.S. equivalency

 

H.S. Graduate

 

Vocational/Trade Certification

 

Some College

 

Associates/2 yr. degree

 

College Graduate

 

Graduate studies/ degree

 

 

-.397     (.359)

 

-.638*   (.374)

 

-.604*    (.361)

 

-.281     (.373)

 

 

-.608      (.387)

 

-.565      (.390)

 

-.434      (.403)

 

-.243      (.427)

 

 

-12.32%

 

-5.18%

 

-18.94%

 

-2.85%

 

 

-11.96%

 

-11.20%

 

-8.16%

 

-4.10%

 

 

-.754***  (.264)

 

-.733**   (.339)

 

-.211     (.269)

 

-.278     (.272)

 

 

-.055     (.248)

 

.089       (.228)

 

.209       (.255)

 

.959***  (.280)

 

 

-25.00%

 

-6.36%

 

-7.07%

 

-3.02%

 

 

-1.16%

 

1.88%

 

4.20%

 

17.28%

Mother’s Educational Attainment

Less than/ Some H.S.

 

GED/ H.S. equivalency

 

H.S. Graduate

 

Vocational/Trade Certification

 

Some College

 

Associates/2 yr. degree

 

College Graduate

 

 

 

.122       (.285)

 

.088       (.411)

 

.042       (.286)

 

.187       (.323)

 

 

.132       (.314)

 

.271       (.313)

 

.365       (.284)

 

 

5.49%

 

.56%

 

1.66%

 

1.32%

 

 

1.14%

 

3.11%

 

5.37%

 

 

-.193     (.202)

 

-.453**  (.220)

 

-.221      (.190)

 

-.397*    (.230)

 

 

.006       (.247)

 

-.359      (.223)

 

-.491*** (.191)

 

 

-9.27%

 

-3.09%

 

-9.34%

 

-3.01%

 

 

.0006%

 

-4.41%

 

-7.71%

Father’s Educational Attainment

Less than/ Some H.S.

 

GED/ H.S. equivalency

 

H.S. Graduate

 

Vocational/Trade Certification

 

Some College

 

Associates/2 yr. degree

 

College Graduate

 

 

 

.001       (.188)

 

-.078      (.325)

 

-.057      (.155)

 

.100       (.256)

 

 

.065       (.279)

 

-.213      (.224)

 

-.237      (.234)

 

 

.0005%

 

-.46%

 

-2.11%

 

.67%

 

 

.49%

 

-2.02%

 

-3.96%

 

 

.177       (.157)

 

-.050      (.235)

 

.231        (.176)

 

-.125       (.230)

 

 

.066        (.218)

 

.136        (.204)

 

.038        (.155)

 

 

8.59%

 

-.31%

 

9.15%

 

-.90%

 

 

.54%

 

1.38%

 

.68%

 

English language Proficiency Characteristics

Document Test 3

.198      (.248)

1.98%

.009        (.208)

.001%

Prose Test 1

.141       (.010)

3.01%

-.098*     (.063)

-1.05

Quantitative Test 2

 

-.190***    (.071)

-.47%

.007      (.069)

.19%

Frequency one reads a the newspaper or a magazine in English

Everyday

 

Few times a week

 

Once a week

 

Less than once a week

 

 

 

 

-.310       (.248)

 

-.354        (.287)

 

-.183         (.264)

 

-.334*       (.266)

 

 

 

-.14%

 

-12.64%

 

-3.65%

 

-5.38%

 

 

 

.566**       (.284)

 

.475*        (.286)

 

.375         (.313)

 

.090          (.286)

 

 

 

28.33%

 

18.13%

 

8.00%

 

1.55%

Frequency one reads a book in English

Everyday

 

Few times a week

 

Once a week

 

Less than once a week

 

 

 

.003          (.171)

 

.143          (.187)

 

-.180         (.199)

 

-.115          (.186)

 

 

.12%

 

4.31%

 

-3.12%

 

-3.87%

 

 

.324*          (.176)

 

.381**        (.178)

 

.107           (.206)

 

.219           (.158)

 

 

14.38%

 

12.27%

 

1.98%

 

7.88%

Frequency on reads a note or letter in English

Everyday

 

Few time a week

 

Once a week

 

Less than once a week

 

 

 

.505*         (.270)

 

.280           (.280)

 

.127          (.313)

 

.492*         (.285)

 

 

23.83%

 

8.62%

 

2.04%

 

10.45%

 

 

.244           (.234)

 

.215          (.227)

 

.108          (.261)

 

.148          (.229)

 

 

12.31%

 

7.08%

 

1.86%

 

3.36%

Age learned to speak English

Does not speak English

 

1-10

 

11+

 

 

 

.347           (.487)

 

.114          (.216)

 

.939***     (.360)

 

 

2.58%

 

2.72%

 

11.25%

 

 

.366       (.360)

 

.029        (.147)

 

-.052         (.303)

 

 

2.91%

 

.74%

 

-6.66%

Language spoken before school

English Only

 

Eng. and Spanish

 

Eng. And other

 

Spanish only, with other

 

 

 

.099          (.362)

 

.278          (.470)

 

.452*          (.325)

 

-.243           (.427)

 

 

3.15%

 

1.28%

 

2.33%

 

-4.71%

 

 

-.143           (.279)

 

.077           (.351)

 

-.104           (.261)

 

-.041          (.319)

 

 

-4.86%

 

.47%

 

-.57%

 

-.85%

 

* t-value significant to the (α = .10) level

** t-value significant to the (α = .05) level

*** t-value significant to the (α = .01) level

(Standard errors are in parentheses)


[1] American Civil Liberties Union. English Only Briefing Paper Number 6, 1996. Accessed through the ACLU official website: http://www.aclu.org/library/pbp6.html  Gathered April 27, 2007.

[2] Roosevelt, Theodore. Personal Works vol.xxiv (Charles Scribner & Sons, New York) 1926. Page 554

[3] American Civil Liberties Union. English Only Briefing Paper Number 6, 1996.  Officially titled H.R. 123, the bill has never come to vote in the U.S. Senate, or has been endorsed by either of the two sitting presidents since 1996.

[4] Hyon B. Shin with Rosalind Bruno. “Language Use and English-Speaking Ability: 2000” U.S. Census 2000 Brief, United States Census Bureau Issued October 2003. Page 1

[5] Hyon B. Shin with Rosalind Bruno. “Language Use and English-Speaking Ability: 2000” Page 2

[6] States with EOL’s include: Alabama, Arizona, Arkansas, California, Colorado, Florida, Georgia, Illinois, Indiana, Iowa, Kentucky, Mississippi, Missouri, Montana, Nebraska, New Hampshire, North Carolina, North Dakota, South Carolina, South Dakota, Tennessee, Utah, Virginia, and Wyoming. Gathered from U.S. English, Inc. www.us-english.org Included in Appendix C is a detailed map of the states with English-Only Laws.

[7] An ethnic enclave is informally defined as a neighborhood, district, or suburb which retains some cultural distinction from a larger surrounding area.  These areas are mostly populated by immigrants who voluntarily cluster together.  Usually, there is a geographic concentration of businesses, residents, and community institutions of a single ethnic group.

[8] Woodruff, Christopher and Fairlie, Robert W. “Mexican Entrepreneurship” A Comparison of Self-Employment in Mexico and the United States” The Institute for the Study of Labor (IZA). IZA Discussion Paper No. 2039, March 2006. Page 19

[9] Dávila, Alberto and Mora, Marie T. “English-Language Skills and the Earnings of Self-Employed Immigrants in the United States: A Note” Industrial Relations vol. 43, no. 2 (April 2004) Page 390

[10] Note: The previous three studies include the 1985 Young Adult Literacy Assessment, the 1990 Workplace Literacy Survey, and the NALS of 1992.

[11] National Center for Educational Statistics, United States Department of Education  National Assessment of Adult Literacy of 2003, Public Use Handbook. The six objectives are to describe the level of literacy demonstrated by the adult population as a whole, characterize adults’ literacy skills in terms of demographic and background information, profile the literacy skills of the nation’s work force, compare assessment result with those of previous assessments, interpret the findings to create curriculum concerning adult education and training, and increase understanding of the skills and knowledge associated with living in a technological society.

[12] Definition can be first found in the 1985 Young Adult Literacy Assessment and in subsequent literacy thereafter.

[13] U.S. Department of Education, National Center for Educational Statistics. The National Assessment of Adult Literacy.  NAAL of 2003 Public-Use Handbook Pages 5-6

[14] The model above will be run in AM Software, a program designed by the U.S. Department of Education’s, National Center for Educational Statistics, and is the program in which any analysis of the NAAL 2003 must be completed.

[15] As defined in the U.S. Census, the country is divided into four regions: Northeast, Midwest, South, and West.  The West region includes Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

[16] United States Department of Justice. http://www.usdoj.gov/crt/cor/Pubs/eolep.htm  Executive Order 13166: Improving access to services for persons with limited English proficiency (LEP), signed by William J. Clinton on August 11, 2000.

[17] Later releases of all NCES datasets are available in transferable form and can be used in other non-NCES statistical programs.  The length of time it takes for additional releases is up to the discretion of the U.S. Dept. of Education.  Note: the NAAL 1993 was not released in its raw from until summer 1998.