7.5 pages body of supporting, cited information, Current issues and future trends related to the topic. Either way, this element should be addressed. Discuss your personal interest and takeaways regarding the topic and research (not the same as the closin

Journal of Social Issues, Vol. 43, No. I, 1987, pp. 5-39

How Minorities Continue to Be Excluded from Equal Employment Opportunities: Research on Labor Market and Institutional Barriers

Jomills Henry Braddock II and James M. McPartland Johns Hopkins University

Barriers to equal occupational opportunities for minorities are examined at three stages ofthe employment process: the job candidate stage, the job entry stage, and the job promotion stage. Using the authors' recent survey of 4078 employers covering a nationally representative sample of jobs, four types of exclusionary barriers are investigated: ' 'segregated networks'' at the candidate stage, ' 'in- formation bias" and "statistical discrimination" at the entry stage, and "closed internal markets" at the promotion stage. Practical implications are drawn for equal employment opportunity policies directed toward occupational processes and employment outcomes.

Employment equity policies have heen the subject of fierce debates for many years in this country. Arguments have ranged widely in areas of political philosophy, constitutional law, and socioeconomic theory (for example, Glazer, 1975; Maguire, 1980). Disagreements have been particularly strong about the preferential affirmative action policies begun in 1965. Rather than review here the various directions of the debates or rehash the opposing sides, this paper

Work on this report was supported by research grants from the Office of Naval Research (N00014-84-K-0007) and from the National Institute of Education, U.S. Department of Education (NIE-G-83-0002). The two authors share equal responsibility for the preparation of this report; the opinions expressed do not necessarily reflect the position or policy ofthe sponsoring agencies and no official endorsement should be inferred. The authors wish to thank George Levinger for his encour- agement and useful suggestions, and Marylee Taylor, Robert Crain, and the 75/ reviewers for helpful criticisms on early drafts. Special thanks to Barbara Colton for her assistance in the preparation of this manuscript.

Correspondence regarding this article should be addressed to either author at the Center for Social Organization for Schools, The Johns Hopkins University, Baltimore, MD 21218.

5

0O22.4537/87/0300.00O5«)5.0O/l © 1987 The Society for Ihe Psychological Study of Social Issues

6 Braddock and McPartland

presents statistics on current labor market processes that can be used to assess the continuing need for strong policies of equal employment opportunities.

Statistics have frequently been used to evaluate the extent of employment discrimination, but they have rarely been used to help identify the specific barriers that may unfairly inhibit the job chances of women or minorities. Thus, we have numerous statistical studies that estimate the size of sex or race gaps in occupational attainments such as income or job level. The authors of these studies usually first try to control statistically on other characteristics of workers that affect occupational success, such as educational attainment or community location, then they interpret any residual race or sex gap as the result of "dis- crimination," or the absence of a residual gap as evidence that discrimination is a thing of the past. Social scientists often disagree about what variables should be measured and controlled in estimating race or sex occupational gaps, and there are many other technical problems in using such residual statistics to estimate discrimination (McPartland & Crain, 1980). But, in the end, this use of statistics does not inform discussions of what particular kinds of policies may be needed to combat discrimination because the specific barriers that may stand in the way of fair employment chances are not assessed directly.

The statistics we present should better inform discussions of particular policy alternatives. Using our recent national survey of 4078 employers, we describe the distribution of actual practices used in recruiting for and filling different kinds of jobs, and we identify the practices that have a differential impact on the probability that minorities will wind up in the job. We also review major theories that have described specific racial-exclusionary processes in em- ployment and we use our data to assess the empirical validity of these ideas.

Following the research results, we draw implications for practical programs and policies. We recommend specific kinds of programs to address the particular employer practices we have empirically identified as unfair employment barriers for minorities. Also, we use our statistical descriptions of the most common employer practices in recruitment, selection, and promotion for different kinds of jobs to comment on the points in the employment process where different broad policy approaches seem most appropriate, including policies of affirmative ac- tion, enforcement of Equal Employment Opportunity (EEO) complaints, and voluntary employer programs.

Race Barriers at Different Employment Stages

Blacks and Hispanics can face special employment difficulties at different stages of the occupational process because they are members of a racial or ethnic minority. Barriers can appear at the job candidate stage when employers are recruiting the pool of candidates for job openings, at tbe job entry stage when an

Minority Exclusion 7

individual is actually selected to fill the vacancy, and at tht job promotion stage when transfers are made within a firm to fill spots at higher levels. We examine each stage in turn by describing the distribution of employer practices for differ- ent kinds of jobs and by analyzing the differential impact on individuals from minority groups of certain employer actions. Evidence is drawn from previous research and from our recent national survey of 4078 employers, which covers public and private sector jobs held by a representative sample of workers from major sex, race, and education subgroups. (The appendix describes the national sample of employers used and the method for defining subcategories of jobs.)

We focus on the barriers faced by race and ethnic minorities that do not derive from educational deficiencies or sex discrimination in occupations. To be sure, those factors produce major income and occupational inequities and require major public programs in their own right (Aaron & Lougy, 1986; Bielby & Baron, 1986; Wilson, 1978). But this paper will focus primarily on issues of fairness for race and ethnic minorities at different employment stages by investi- gating employer practices within subcategories of jobs defined by the sex com- position and educational attainment of their current workers.

The Job Candidate Stage

A qualified person's chances for employment in the most desirable job openings begins with finding out about those vacancies and becoming part of a pool of candidates. To determine whether minorities have a fair chance at the job candidate stage, we need to learn how employers most commonly recruit candi- dates for different kinds of jobs and to assess whether minorities have equal access to these recruitment channels.

Our recent national survey of 4078 employers shows that the type of job to be filled strongly influences the variety of recruitment methods used by em- ployers. At the same time, informal recruitment methods that rely upon social networks of information are among the most frequently used methods for all job types (see Tables A2 and A3, rows 1-10, in the appendix).

Employers are usually not disposed to spend much time or money in recruit- ing for lower level jobs that do not require any college education. The most convenient and inexpensive methods dominate employer practices for these jobs. In order of their frequency of use and value for employers, the most important methods include unsolicited "walk-in" applications, informal referrals from current employees, and referrals from public employment agencies (see also Becker, 1977; Lippman & McCall, 1976; Rosenfeld, 1975). Apparently em- ployers get enough qualified applicants for most lower level jobs by doing nothing more than placing a job opening sign at their establishment, passing the word to their current work force about the vacancy, or making a call to the local

S Braddock and McPartland

public employment agency. Other recruitment methods, such as placing ads in local media, are used less frequently and much less frequently than when recruit- ing to fill higher level jobs. A similar pattern of domination by informal and inexpensive methods emerges from parallel studies of the job search practices of individuals who do not have any college education. These individuals most frequently rely on "word of mouth" job information from friends and relatives and make direct "walk-in" applications for work (Baker et al., 1984; Granovet- ter, 1974, 1982; Mangum, 1982; Pames, Miljus, Spitz, & Associates, 1970; U.S. Department of Labor, 1975, 1976).

Employer recruitment methods vary much more for upper level jobs, but the informal methods remain as major sources of college-educated job candidates. Employers often spend the time and money to seek college-educated job candi- dates from college placement services, media ads, professional organizations, and private employment services, but our recent survey indicates that informal referrals from current employees and unsolicited walk-in applications are also among the most frequent and most important employer methods for creating college-educated candidate pools (Tables A2 and A3, in the appendix).

Thus the use of informal social networks is a principal method through which employers with job openings are brought together with individual job seekers from outside the firm. How do minorities fare at this job candidate stage? Social scientists have long suspected that blacks are denied equal access to the most valuable informal sources of job information. They have reasoned that black job seekers are primarily tied to social networks composed of other blacks who, on the average, are not as well situated to know about many desirable job openings as the members of the social networks used by white job seekers (Crain, 1970; McCall, 1972; Rossi, Berk, Boessel, Eidson, & Groves, 1968; Rossi, Berk, & Eidson, 1974). Thus, an important minority exclusionary barrier, which we will call "social network segregation," has been hypothesized to operate at the job candidate stage.

Several empirical studies support this view, although most previous re- search has not included direct measures of the kinds of informal social networks that link job seekers to job vacancies. One set of studies examined firms with different racial compositions. In 1967, Rossi and his co-workers (1968, 1974) surveyed 434 personnel managers of the largest employers in 15 major cities to investigate factors related to the number of blacks who applied for work and were hired for recent vacancies at three different job levels. The authors argued that the past employment practices of a firm, as measured by the percentage of blacks in their current work force, could be used to indirectly assess the importance of social networks in the job recruitment process. According to the authors, if the current racial composition of a firm is the best predictor of the rate of recent black applications, we would have indirect evidence that the social networks through current black employees provide an important recruitment channel to

Minority Exclusion 9

reach potential new black applicants. They found that the percent black in a firm's current work force is indeed a strong predictor of the probability that blacks had recently applied for work at the firm, after statistically controlling on other characteristics of the firm and the labor market (including the racial com- position of the city in which the firm was located).

An analogous result is reported by Becker (1980), who used the Equal Employment Opportunity Commission national survey of the racial composition of firms to show that the racial composition of an establishment's work force at one occupational level is strongly related to its racial composition at other levels. This finding also supports the view that black employees in a firm provide useful informal links to other blacks in the labor market to become candidates for employment at the firm.

A second set of earlier studies examined the occupational consequences for blacks of attending segregated or desegregated elementary and secondary schools. If using segregated social networks during the job search process se- riously impedes black employment in desegregated jobs and firms, blacks who graduate from segregated black schools—and thus most likely to have access to segregated networks only—should wind up in racially segregated employment. Braddock, Crain, and McPartland (1984), summarizing the results from five different national surveys conducted since 1970, report that black graduates from desegregated schools are significantly more likely to be employed as adults in desegregated places of work. Although these studies did not measure which graduates used friends to search for jobs, the authors argued that student access to desegregated social networks was a major explanation for the observed rela- tionship between graduating from desegregated schools and entering desegre- gated work environments, especially since they had statistically controlled for differences in racial proportions in local labor markets.

Our recent survey of 4078 employers permits more direct study of how social networks affect minorities' job chances because we have measures of employer recruiting practices, individual job search techniques, and the em- ployment outcomes that result from using different methods.

Table 1 highlights the results of multiple regression analyses that investigate the relationship between employers' use of the social networks provided by their current employees to recruit new workers and the likelihood that a job opening will be filled by white rather than minority workers. The multiple regression analyses for Table 1 also included measures that controlled for the percentage of whites in the local labor market and the job sector, job sex, and education compositions (see appendix for details).

For college-degree jobs (positions usually filled by workers with a college degree), we find the chances are significantly greater that an opening will be filled by whites when social networks are used as a major employer recruitment method. But for middle level and lower level jobs (positions usually filled by

Braddock and McPartland

Table 1. The Effects of Employers' Use of Social Networks on the Probability that a Job is Filled by Whites (by Educational Level

of the Job, Controlling for Five Labor Market Variables)"

College degree jobs (n = 850)

Some college jobs (n = 1048)

High school jobs (n = 2396)

Employers using social networks

.83

.74

.64

Employers not using social networks

.75

.72 (ns)*

.66 (ns)

"The complete equations, with multiple correlation coefficients for all the regression tables, are available upon request from the authors,

''ns, not statistically significant at .05 level.

workers whose highest education level is either some college attainment or only a high school education), there is no sizable or consistent employment benefit to whites or minorities that depends upon whether the employer recruits through social networks.

We believe the racial composition of social networks tied to different jobs is the best explanation for the initial finding that employers' reliance on referrals from their current work force results in greater employment chances for whites only for higher level positions. Accordingly, we now further examine qualitative differences in social networks tied to lower level jobs to draw our final conclu- sions about informal barriers in these cases.

The measure of the frequency of employer reliance on informal networks used in Table 1 is likely to incorporate the qualitative advantages to whites of this recruitment method for college-level jobs because of the racial demography of current employment in these jobs. The current work force in most college-level jobs is predominantly white, so the informal social networks of relatives and friends linked to these jobs will also be predominantly white. Therefore, most college-educated minorities will not have access to the white informal networks tied to these college-level jobs, and will be cut off from the candidate pools when informal word-of-mouth referrals from current employees is a primary recruit- ment method for these jobs, as shown in Table 1.

However, the overall frequency of use of social networks is not a good measure of informal recruitment barriers at lower level jobs, because despite smaller overall differences in the racial representations in lower level em- ployment within the same education category of work, white social networks may be tied to higher quality jobs than minority social networks. In other words, we could find no racial differences in overall employment rates for lower level jobs that depend upon the employers' use of word-of-mouth referrals because

Minority Exclusion 11

both whites and minorities frequently find jobs through these methods—al- though whites find better jobs than minorities in this way. As Lin (1982) has pointed out, in studying social networks in employment, we need to pay attention to how networks differ in their instrumental value due to how they provide access to different resources and positions. When whites are currently employed in better jobs than blacks of the same education level and informal networks of information about job openings follow racial lines, we need to study not just the frequency of use but the qualitative worth of different social networks to evaluate racial barriers for lower level jobs.

We used job information from our recent survey of employers in combina- tion with the National Longitudinal Survey of individuals that parallels our sample to study the details of social networks that black high school graduates used to search for jobs. To compensate for the lack of direct measures of the racial composition ofthe social networks used by black job seekers, we used the racial characteristics of the high school from which each individual graduated to identify his or her social networks as segregated or desegregated. Table 2 pre- sents selected average job outcomes for black high school graduates who used segregated social networks, did not use any social networks, or used desegre- gated social networks. It shows that black high school graduates who used desegregated social networks to find their jobs are in the highest paying positions in firms and in jobs with the highest percentage of white co-workers. Those who used segregated black social networks are on the average in the lowest paying positions in firms and in jobs with the lowest percent of white co-workers. Black high school graduates who did not use social networks to find their job fall in- between the other groups in pay level and desegregation of co-workers. Thus the value of social networks for finding good jobs by black high school graduates depends upon the kind of social networks used: segregated networks lead to poor-paying, more segregated jobs (it is better on the average to depend on some other job search technique), and desegregated networks lead to better paying, less segregated work.

Table 2. Job Characteristics of Black High School Graduates Who Used Different Types of Networks in Their Job Search (Private Sector)

Job outcome

Percent white of fellow workers

Percent white in the firm

Hourly wage

Used segregated networks

.462 (75) .523 (70)

$5.69 (78)

Did not use networks

.504 (277) .596 (252) $5.74 (287)

Used desegregated networks

.560 (42) .694 (41)

$6.45 (41)

12 Braddock and McPartland

Based on the Table 1 results for higher level jobs, the Table 2 results for lower level jobs, and previous research consistent with these findings, we con- clude that segregated social networks constitute an important racial barrier at the job candidate stage. Minorities often miss the chance to be hired for good jobs because they do not have equal access to one of the most important employer recruitment channels that create the actual pool of candidates for the job open- ings. We find this problem continues to exist at all job levels, but it may be especially important for those lower level jobs where employers' use of informal methods dominates their recruitment practices. For higher level jobs, employers are more often willing to use a variety of recruitment methods, including the more expensive and time-consuming formal practices that do not seem to have the same racial biases as social network recruitment. Still, informal methods are a major source of job candidates for jobs at all levels, and minorities continue to have unequal access to good jobs because of the frequently segregated nature of these channels.

The Job Entry Stage

An employer selecting whom to hire from a pool of job candidates recruited from outside the firm usually has a mental list of the most important traits needed to perform the job and some information about each candidate with which to judge these traits. The selection process is much more difficult when the job applicants have never worked for the firm, because no direct information will exist within the firm on how each candidate performs on a job and gets along with supervisors and fellow workers. In this case, an employer must rely either on references about each job candidate from other employers and educators, or on data that can be obtained through the firm's use of tests, interviews, and assessments in its own direct screening. Employers differ widely in the exten- siveness of the infomiation they use in choosing new employees from outside the firm.

Most employers first establish a minimum education level for eligibility for each job. Educational diplomas or degrees are used as an initial screening device for different jobs because employers believe a particular educational credential provides a "signal" of the minimum kinds of worker traits possessed by the individual who earned it (Spence, 1973). Although some economists question whether better educated individuals are actually more productive in all job situa- tions (Berg, 1981), most employers assume that individuals who have gone farther in school are more likely to have desirable skills related to academic or learning tasks on the job and successful functioning in an organizational environ- ment. In any case, requiring a minimum education level is an easy and inexpen- sive way to limit the field of job candidates, and it is usually the basis for the employer's first cut in the hiring process. Some have argued that requiring a

Minority Exclusion 13

candidate to have a general educational credential such as the high school diplo- ma, is discriminatory in cases where the credential has not been proved to predict specific traits needed in the job, especially since minorities in most localities are significantly more likely to have dropped out of school before achieving the required credential (U.S. Equal Employment Opportunity Commission, 1974). But even after an initial screening of candidates by education level has been made, other frequent selection practices have been hypothesized as unfairly excluding minority applicants from job opportunities.

The amount of information beyond the applicants' educational level used in the selection process will usually depend on the importance of the job in the firm and the difficulty of finding candidates with the desired job traits. Certain com- mon combinations of job traits sought by employers and information used in screening candidates can cause serious problems for qualified minority job candi- dates. Job entry barriers for minorities often occur because employers do not use the kinds of additional screening infomiation that would give minority applicants an equal chance to demonstrate their qualifications on the high-priority job traits.

Job Traits in Demand

Our recent national survey of employers shows that attitudinal traits are at least as important as educational training in hiring decisions for many jobs, especially jobs filled by high school graduates (see Tables A4 and A5 in the appendix). For example, dependability in coming to work regularly and on time, proper attitudes about work and supervisors, and the ability to get along well with work-team members consistently top employers' lists of qualities they seek in filling lower level entry positions (see also Committee on Economic Develop- ment, 1985; National Academy of Sciences, 1984; U.S. Department of Educa- tion, 1986). In our survey, employers usually report they do not need high levels of reading and math competencies for these positions, but they do expect basic literacy and computation skills and the ability to'leam new things quickly on the job. Employers seem generally satisfied with the basic academic skills of most high school graduate job applicants, but less than satisfied with their work attitudes and on-the-job learning abilities (Crain, 1984; McPartland, Dawkins, & Braddock, 1986).

Good attitudinal traits are also a high priority for upper level jobs, but other factors emerge: more advanced levels of language or computation skills and specialized knowledge become in high demand, as well as the ability to deal with complex situations and to learn new things quickly. Besides knowledge acquired from specialized college courses, employers look for good judgment and lead- ership potential among applicants who have college training and credentials (Tables A4 and A5).

The average employer seems to perceive important racial and ethnic group

I't Braddocii and McPartland

differences on these priority job traits. When generalizing about white and mi- nority group workers of the same sex and same educational level, many em- ployers see blacks as higher risk employees, both in their attitudes about work and in their previous training in useful skills for the job. In an earlier survey of personnel officers conducted in the late 1960s, significant proportions agreed with derogatory statements about blacks' attitudes toward work when consider- ing members of this racial minority group as potential employees (Rossi et al., 1974, pp. 278-279). Our recent national survey of employers provides evidence that employers are more likely to avoid hiring minorities in those jobs that emphasize academic achievement and thinking skills. After describing this re- sult, we present further evidence on whether the observed racial employment patterns go beyond measured individual differences in the job requirements given high priority.

Table 3 highlights the results from multiple regression analyses that investi- gate the relation between the percent white hired in a job and the importance rating that employers give to selected worker characteristics for the job. These regression analyses included measures to control for percent white in the local labor market, job sector, and job sex and education compositions (see appendix for details).

Whites are significantly more likely to be found in lower level jobs (most often filled by workers whose education went no further than high school) that require both basic and advanced skills in reading and math, as well as in jobs that value quick learning and good judgment in complex situations. Whites are also favored in lower level jobs that emphasize certain interpersonal attributes, such as

Table 3. Probabilities that a Job is Filled by Whites When Selected Worker Traits Are Important (by Educational Level of the Job, Controlling for Six Job Conditions)

High school jobs College-degree jobs (n = 2396) (n = 850)

Trait is not Trait is very Trait is not Trait is very Worker trait important important important important

Basic adult literacy .59 .68 .83 .82 ns" Advanced reader .63 .73 .74 .80 Basic arithmetic .55 .71 .77 .79 ns Excellent at math .64 .74 .77 .80 ns Quick learner .56 .68 .65 .81 Good judgment .55 .69 .69 .79 Client relations .63 .70 .78 .78 ns Good team member .56 .67 .77 .79 ns Can supervise .63 .70 .78 .79 ns

°ns, probabilities in adjacent columns are not statistically significant at .05 level.

Minority Exclusion 15

being skilled at client or customer relations, being able to get along with people as good team members, and providing direction or leadership in supervision.'

For upper level jobs (most often filled by college graduates) statistically significant racial differences in hiring patterns disapppear for most of the job traits. But employers continue to show a significant hiring preference for whites in upper level jobs that emphasize the most advanced academic and reasoning skills, including advanced reading, quick learning, and good judgment.

These results do not indicate discriminatory behavior, if employers are making hiring decisions based on actual individual differences on the desired job traits among the job candidates, and if the minority candidates often fail to measure up in these individual assessments. For example, even though racial gaps have been closing in recent years on tests of academic skills, segregated schools with unequal resources for large proportions of minority students to- gether with other disadvantages in learning environments continue to produce sizable average differences between whites and minorities on these tests. Thus, it is conceivable that the racial differences in hiring for jobs that emphasize aca- demic skills could primarily refiect the results of employers' assessments of the individuals who appear as candidates for these jobs. But another possible process has been identified in employer selection that overlooks individual differences to produce an unfair racial exclusionary barrier.

When employers use negative group images rather than direct assessments of individuals in their selection process, the exclusionary barrier of "statistical discrimination" is said to exist (Aigner & Cain, 1977; Thurow, 1969, 1975). Employers will consider a group identifier such as sex or race in hiring decisions when they believe the traits on which subgroups differ predict job performance, and when they are unable or unwilling to determine individual differences within subgroups on these traits. Thus, when information about individual differences is lacking, employers who use a group identifier in selection will expect to have a better statistical chance of getting a desirable worker because of their perception of average group differences on job-relevant traits. Usually the use of race or ethnic identifiers in job selection means that a white will be chosen over a black or other minority applicant (Thurow, 1969, 1975).

If a qualified minority job candidate cannot escape a negative racial group profile when being judged for employment, that individual is denied an equal employment opportunity. This exclusionary barrier can come into play whether the employer perceptions are based on actual group differences or on entirely uninformed group stereotypes. But to a minority member who has individual

'Our data did not allow a test of racial hiring diffferences in jobs requiring dependability and proper attitudes. Almost all employers rated these traits as very important, which left no variance on these items to analyze.

16 Braddock and McPartland

qualifications well above the minority group average and who is denied em- ployment without those individual traits being considered, it will matter little whether the employer's group perceptions that cost him or her the job are true, partially true, or not true at all.

We now look for evidence of the existence of statistical discrimination in two ways. We use data that measure both individual differences and racial hiring rates in jobs that emphasize selected worker traits, to test whether the observed hiring results can be accounted for by individual differences alone. Then we examine the infonnation employers actually use in selecting among candidates for different jobs, to study the frequency of the conditions for statistical discrimination.

We can make some direct tests of the hypothesis of statistical discrimination because our employer survey infonnation about the requirements and hiring outcomes in a national sample of jobs can be linked to a sample of individuals in each of the sample Jobs, and we have measurements of each individual's race, sex, educational attainment, and academic test score performance. Thus, we can investigate whether the jobs that whites and blacks hold are equally likely to emphasize academic skills in reading or math or general learning skills, after taking into account individual differences in test score performance on the same job requirements.

Table 4 presents the results of multiple regression analyses conducted on three subsamples of individual workers. The dependent variable is their em-

Table 4. How White Workers' Jobs Differ from Black Workers' Jobs in the Importance Rating Given by Employers to Selected Job Traits (Controlling on Individual Differences in the Same Trait

and Three Other Worker Characteristics, by Education Level of the Job)°

Job trait being rated (dependent variable)

Quick learner

Basic adult literacy

Advanced readers

Basic arithmetic

Excellent at math

Good judgment

High school jobs (n = 2396)

.109 (.031)

- . 0 3 5 ns (.035) .113

(.043) .102

(.041) .172

(.045) .093

(.035)

Some college jobs (n = 1048)

.065 ns* (.041)

- . 0 5 5 ns (.037) .004 ns

(.060) .156

(.053) .019 ns

(.068) .087

(.040)

College degree jobs (n = 850)

- . 0 2 0 ns (.052)

- . 0 3 1 ns (.048) .028 ns

(.060) - . 0 8 5 ns (.059)

- . 1 4 9 ns (.079)

- . 0 3 8 ns (.033)

"Unstandardized regression coefficient (standard error in parentheses), ''ns. not statistically significant at .05 level.

Minority Exclusion ^'

ployer's rating of the importance of a selected job trait, and the independent variables are the individuals' race, test score on the same job trait, educational attainment, sex, and job sector. The three subsamples are defined by the educa- tion of the majority of the workers in each individual's job. Each value shown in Table 4 is a regression coefficient for the individual race measure: A significant positive value indicates that white workers are more often found in jobs rated high on the selected trait, even after individual differences in the same trait are controlled statistically. This condition would be produced from statistical dis- crimination practices by employers—it means that equally qualified blacks have not been hired with the same frequency as whites in jobs that emphasize the selected trait. A significant negative value indicates that black workers are over- represented in jobs rated highly on the selected trait, given the same individual qualifications. This condition would be consistent with certain "affirmative action" programs that establish an acceptable job trait criterion level for hiring, above which all candidates would be qualified, and then hire some qualified blacks even though their individual scores are below some white candidates not hired.

Table 4 provides consistent evidence of the existence of statistical discrimi- nation for those lower level (high school) jobs when academic and learning traits are highly valued. Occasional statistically significant positive values are also found for middle-level (some-college) jobs, suggesting that statistical discrimi- nation may also occur in these cases. For the highest level (college-degree) jobs, no values are statistically significant, but most have negative signs. Thus, we find no evidence for statistical discrimination in jobs usually filled by college graduates, and there is a hint that hiring policies may admit some blacks whose tested level of academic skills is not at the same point as whites in the same jobs.

We have only been able to study statistical discrimination for a limited set of academic job traits on which individual data was available. There are numerous other traits that are also important for hiring decisions on which qualified blacks may also be unable to escape group stereotypes in employers' selection deci- sions. These include work attitude dimensions and other characteristics that research has shown are highly valued and where racial group stereotypes are often held by employers. In the case of academic job traits, we conclude from Table 4 that statistical discrimination is often a significant problem for blacks who have not completed a college degree.

We can learn about possible exclusionary barriers at the job entry stage not only by studying how employers react to a candidate's race when different job traits are in demand, but also by describing how employers actually use informa- tion in their selection processes and establish the conditions for equal or unequal employment opportunities.

^̂ Braddock and McPartland

Information Used in Selection

Our recent survey of employers shows the types of information that are used most frequently and are most influential in employers' hiring decisions for jobs that recruit candidates at different education levels (see Tables A2 and A3, rows 11-18 in the appendix). We find that job level influences both the type of information used and the general effort that employers make to gather outside data.

For middle-level and upper-level jobs that require some college or a college degree, employers are often interested in the specialized knowledge that further education produces. They use screening information about the type and reputa- tion of the applicants' college program, the applicants' grades in college, and recommendations from college officials. But even more important than informa- tion about educational training in the final decision of whom to hire for upper level jobs are references or recommendations from previous employers. Em- ployers want to know not only whether a candidate has the proper educational training, but also how the candidate has worked out in other actual job situations.

For most lower level jobs, employers rarely use detailed, specific informa- tion about an individual's education or skills to choose among applicants who have graduated from high school. In fact, the final screening process is often quick and superficial. Our research, consistent with previous studies, shows that only two sources of information are both frequently used and highly valued in hiring decisions for lower level jobs: (1) impressions gained from the job applica- tion form or during the personal interview with the candidate, and (2) recommen- dations from previous employers when available (Bishop, 1986; Hollenbeck, 1984; McPartland, Braddock, & Dawkins, 1986).

It may be surprising that other infonnation such as school records or tests of candidates are not used in the hiring process, but employers often have good reasons for not trying to get better information with which to screen their appli- cants for lower level entry jobs.

Employers often have little time to gather outside infonnation on job appli- cants at this level because openings often come without much notice (due to unexpected resignations or moves of current employees) and vacancies must be filled quickly to maintain routine work flows. Employers who need to move rapidly cannot wait for schools to provide transcripts or recommendations, and in any case most schools are not well equipped to provide records on graduates to employers (Bishop, 1986; Hollenbeck, 1984). Except for some clerical posi- tions, written tests are infrequently used to screen for most jobs at this level (Friedman & Williams, 1982) because they can be costly and there can be uncertainties about their legal standing for hiring decisions (Tenopyr, 1981). Moreover, most school records or test information will pertain to academic and learning skills rather than to the attitudinal traits given highest priority by em-

Minority Exclusion "

ployers for most of these jobs. Woixies about legal obligations in hiring pro- cesses may also hinder the value of checking references by telephone, because previous employers who are asked to serve as references may often provide only dates of service with no qualitative assessments, to avoid potential involvement in legal proceedings (Bishop, 1986). More generally, employers may simply be unwilling to invest much in screening for low-level positions because they feel new hires may not stay long in these spots and they can find equally good replacements from walk-in applicants who meet their established minimum edu- cation level for the job (Berg, 1981; Kalleberg & Sorensen, 1979).

Even when outside infonnation is actually used in the selection process, another type of exclusionary barrier has been cited that we can also comment on with our data. This barrier, which can be called "infonnation bias," will occur when employers select among candidates by using specific infonnation that minorities cannot provide with the same frequency or credibility. It can be argued that minorities' concentration in racially segregated neighborhoods and schools, and in economically depressed local labor markets, creates a racial bias in the information employers most frequently use to fill certain kinds of jobs.

Minorities may be at a special disadvantage when employers are interested in a candidate's previous employment experiences, or in references and recommen- dations for a candidate from school or employment officials. Because of the higher youth unemployment rates in minority communities, minority job seekers are less frequently able to list previous work experience on their job applications or to describe previous jobs during an employment interview. Because both em- ployment application forms and interviews are especially important in the selec- tion process for lower level jobs, the disadvantages that minority youth experience from poor employment opportunities in their own communities can carry over into infonnation bias in later job screening processes that use previous work experience for selection among applicants.

Another type of information bias can occur when the recommendations or references provided by minority applicants carry less weight with employers than the recommendations or references provided by white candidates. Due to segre- gation of schools and communities, white employers may be less familiar with a black school, a black clergy, or a black firm that a minority individual may use for sponsorship of his or her job candidacy, or white employers may feel more suspect of infonnation provided by minorities due to stigma or stereotypes at- tached to minority sources. In a separate study conducted with our employer survey, it was found that employers gave special credibility and weight to minor- ity graduates of suburban schools if they were asked when they might hire minority high school graduates in their firms (Braddock, Crain, McPartland, & Dawkins, 1986). This result supports the argument that segregation introduces information bias into the screening process by assigning different credibility to employment sponsors of minority and of white applicants.

20 Braddock and McPartland

The Job Promotion Stage

Some job vacancies are filled from within the establishment by finding suitable individuals from the current work force. Our employer survey covers a national sample of all types of jobs, including jobs filled by new hires from the outside, jobs filled from within by internal promotions or transfers, and jobs that have been filled from both within and outside the firm. We use the survey data to compare promotion opportunities for minorities, and to investigate previously cited exclusionary barriers against minority chances for advancement.

Finding Candidates for Internal Promotion

Employers who plan to fill a job vacancy from within the organization do not necessarily begin by recruiting a pool of candidates as they usually would when hiring from the outside. Internal promotions that do resemble the wide- spread recruitment used in outside hires are those for which a general announce- ment of job openings is made to current employees by posting a job vacancy notice and inviting applications. More often, specific current employees are in line for certain job openings because of the way a firm internally organizes its jobs. In many of these cases, a career ladder is established in the firm so that lower level positions are the training grounds for the next level, and the workers currently on these lower rungs automatically become the candidates for promo- tion for relevant vacancies.

If minority workers do not enter the firm in jobs that have training oppor- tunities and are tied to upward career ladders, they will be excluded at the outset from chances for career advancement within the firm. This kind of exclusionary barrier can be called a dosed internal labor market. Minorities may have particu- lar difficulties in being initially hired into those entry jobs that provide training and advancement opportunities because, as we have seen, employers often tend to downgrade minorities' abilities as quick learners, a trait that seems to be most valued for entry positions with growth potential. To directly test whether minor- ities are excluded at the outset from promotion possibilities, we analyzed data from our employer survey on internal recruitment methods.

On our survey of employers, we asked how often the following methods were used for different types of jobs: (a) inform current employees ofthe opening by posting or circulating a written vacancy notice, (b) go directly to a specific current employee to encourage that person to apply, or (c) offer the job directly to a specific current employee. The first method opens the application oppor- tunity to all interested current employees. This method is more likely to be used by public employers than by private employers (approximately 80% vs 50% of the time) and is more likely to be used in larger establishments. The other

Minority Exclusion 21

Table 5. Probability Job is Filled by Whites when Different Internal Recruitment Methods Are Used (Controlling on Job Sector, Firm Size, and Three Other Labor Market Conditions, n = 2284 jobs)

Internal recruitment method Method used Method not used

Go directly to specific employees for applicants .71 .67 Offer job to specific current employee .72 .68 Post or circulate a written vacancy notice .67 .72

methods give some current employees the inside track for internal promotion opportunities, and are more likely to be used by private than by public employers (approximately 35% vs 10% of the time) and by smaller establishments.

Table 5 shows the results of multiple regression analyses that estimate the relation between each employer's internal recruitment method and the percent of white workers selected for the job, with statistical controls on the job sector, establishment size, racial composition of the local labor market, and sex and education level of job incumbents. It shows statistically significant differences that favor white chances of being hired through internal transfers when em- ployers go directly to specific employees to find applicants and when employers offer the job directly to a specific current employee. On the other hand, the probability that a minority worker will fill the job is significantly greater for jobs for which employers post or circulate a written vacancy notice.

These results indicate that minorities are more often deprived of the oppor- tunity to apply for openings within their firm when these employment oppor- tunities are withheld from the public channels of infonnation or are wired to particular individuals who are favored for promotion or transfer. Our data do not show why employers use these exclusionary methods, so both intentional avoid- ance of potential minority candidates and unintentional consequences of internal career ladders are possible reasons (Feagin & Feagin, 1978; Femandez, 1981; Sorensen, 1983, 1984). That is, white managers may give favorable treatment to candidates of their own race, or whites may be in line to fill vacancies because they dominate the lower career-ladder positions that lead to later promotions. In any case, minorities appear to lose many opportunities to become candidates from within the firm for job openings when informal or exclusive channels are used in the internal recruitment process.

Criteria for Promotion

Given that minorities are frequently at an initial disadvantage in getting into the candidate pools for many internal promotions, is there evidence for additional structural barriers when the final selection for promotion is made? Are there forces that make discrimination less likely when an employer selects from an

22 Braddock and McPartland

internal candidate pool of current workers than from an external pool of outside candidates? Are there other forces that favor more equal selection opportunities in the cases of hiring from the outside? Sorensen (1984) has argued that internal labor markets are less subject to the economic market forces that can make discrimination costly to employers, and that discrimination diminishes when employers compete in open markets for outside workers to fill their jobs. On the other hand, because employers possess direct information on their current em- ployees' actual job performance, they should be less likely to practice statistical discrimination—i.e., to judge individual minorities by characteristics of their group—when internal selections are made. Several researchers have argued that the exclusion of women and minorities from positions that are likely to lead to promotions is a major explanation for sex and race gaps in occupational attain- ment; these arguments are almost always based on inferences from studies of general attainment models rather than from direct investigations of personnel practices (Baron, 1984, pp. 40-41). Our employer survey presents some re- search opportunities to look more closely at this issue.

We asked employers what kinds of information they use when filling a job opening from within their firm, using a question that closely parallelled (with some additional categories) the question asked about selection information for external hires (see Tables A2 and A3, rows 19-28). Some types of information (especially seniority and union membership) pertained more to lower level jobs; and other types of information (such as type of education) applied more to higher level jobs. The most important information overall was the job performance of individuals within the firm as indicated by production or sales records, perfor- mance ratings, and recommendations by supervisors or colleagues in the firm. In other words, when choosing among internal candidates for a job vacancy, the overriding factor is how well an individual has proved himself or herself by behavior within the firm. To be sure, to the extent that subjective evaluations are included in the performance ratings of individuals, prejudice can still distort the record of minorities (Butler, 1976; Feagin & Feagin, 1978). Also, some skills required for the promotion position may not be needed in the lower level jobs of the internal candidates, which still permits group stereotypes to influence selec- tion decisions. Pettigrew and Martin (1987, this issue) discuss other powerful interpersonal processes that weaken the chances of a minority member being selected for promotion even when the individual is initially situated in a job that could lead to promotion. Nevertheless, minority members admitted to an internal candidate pool should experience less selection discrimination than those in external candidate pools, due to the availability of direct information about how they have actually performed within the firm.

Because our employer survey covered both a sample of jobs usually filled from the inside and jobs usually filled from the outside, we can contrast racial

Minority Exclusion 23

Table 6. Relationship Between Job Hourly Wage Rate and Percent Black Workers in the Job (Controlling for Educational Levels of Workers in the Job, By Job Sector and Sex)"

Private

Private

Public

Public

sector male jobs

sector female jobs

sector male jobs

sector female jobs

Jobs filled from inside

- $ . 9 7 (.52)

n = 681

- $ . 5 1 (.35)

n = 572

$.25 (.83)

n = 135

- S . 6 4 (.43)

n = 213

ns*

ns

ns

ns

Jobs filled from outside

- $ 2 . 3 3 (.38)

n = 976

- $ . 7 8 (.21)

n = 1110

- $ 1 . 2 5 (.58)

n = 275

- $ . 7 9 (.27)

« = 445

"Unstandardized regression coefficients (standard error in parentheses; n = number of jobs).

*ns, not statistically significant at .05 level.

differentials in the two sets of jobs. Table 6 summarizes the results of these analyses, which examined how the beginning hourly wage rate of jobs is related to the percent black in the job, controlling for the distribution of educational attainments of workers in the job. This relationship between wage rate and percent black was estimated separately for jobs primarily filled from within the firm and jobs primarily filled from without, in different labor market sectors (male jobs and female jobs in the private and public sectors). In every com- parison between internally and externally filled jobs, the difference in wage rates between jobs due to whether blacks or whites had been selected was smaller for jobs filled from within the firm. Thus blacks who make it into the candidate pool for internal selection seem to face less discrimination in achieving good jobs (at least good-paying jobs) than blacks who are job applicants from outside the firm, controlling on education differences among the candidates.

In a study of a large public bureaucracy using different data and methods, Rosenbaum (1984) also identified racial gaps at the point of occupational entry as a more serious problem of discrimination than racial differences in occupational status after entry.

The research results do not mean there are no serious problems due to a candidate's race during internal selections. For example, we find negative salary differences associated with percent black for internally filled jobs in three of our four subgroups, and Rosenbaum's research also consistently found negative salary gaps for long-term minority workers in his public sector research. But, it

24 Braddock and McPartland

does appear in our study that when minority workers are given a chance to prove themselves on internal jobs with growth potential, they have fewer problems with discrimination than when they must rely on the selection information used for external hires. In our view, the most serious problem then becomes the lack of equal opportunity for minorities to enter those jobs that have the best training and advancement possibilities and that form the candidate pools for internal selection.

Practical Implications

Equal employment opportunity policies can be directed toward employment processes or toward employment outcomes. Policies to improve employment processes are aimed at specific employer practices in recruitment, selection, and promotion that create unfair barriers for minorities. Policies about outcomes focus on the degree to which the actual racial or ethnic distribution of employees in a firm matches the distribution of each group in the local labor market with the required job traits. Outcome-based policies often use affirmative action goals and timetables to work toward a better race or ethnic match of a firm's work force and the local labor market demography. Either type of policy can be either voluntary or enforced, depending upon the degree to which employers' actions are monitored and responded to by outside agencies. We briefly review specific policies of each type that have been proposed, and we comment on their necessi- ty and efficacy in light of our research results.

Implications for Fair Employment Processes

Employer activities can be identified at each stage of the employment pro- cess that would make equal treatment more likely for all qualified potential candidates. Some of these ideas for improving the employment process go back to the 1960s and 1970s, and can be found in government guidelines (U.S. Equal Employment Opportunity Commission, 1974) and in academic and professional books on the topic (Alvarez, 1979; Feagin & Feagin, 1978; Fernandez, 1975, 1981). Some of these ideas have been developed recently to provide useful new directions.

At the recruitment stage, employers can avoid word-of-mouth or walk-in methods, clearly indicate the firm's EEO policy in advertisements and advertise in media specifically directed toward minorities, emphasize the firm's EEO policy with private employment agencies and list jobs at all levels with public employment agencies, and use community agencies that specialize in providing minority job candidates. Also, employers can develop closer working ties with high schools and community colleges in order to work with school officials who can locate minority candidates and to use part-time, work-study, and summer job

Minority Exclusion 25

programs that will introduce potential long-term minority employees into the firm.

At the job selection stage, employers can use objective rather than subjec- tive screening methods and ensure that these include only job-related and vali- dated selection standards that do not require greater educational credentials or competencies than are actually needed to perform the job adequately. Detailed guidelines on the proper design, content, and use of application forms, inter- views, and screening tests for selection have been developed over the years.

Recent proposals suggest providing more complete accessible information on young adult applicants at the job selection stage. Schools could develop portable records of academic and nonacademic accomplishments that their grad- uates can carry with them as job applicants. These records, called "career passports" (Chamer, 1984) or "job search portfolios" (Bishop, 1986), include official information on a student's behaviors and accomplishments in school that can be used as indicators of job-relevant attitudes and skills in the job selection process. For example, a record of good school attendance would indicate to an employer that the applicant would not have absenteeism problems as an em- ployee. A record of membership or leadership in extracurricular school activities would imply that the individual would fit well into the work team. A transcript of academic courses and grades in this folder might help an employer appreciate the specialized knowledge a job applicant would bring to the firm, and written recommendations from school officials and instructors could draw attention to other competencies and positive attributes of the candidate. But this information must be available at the time of the screening process to be useful to the job applicant, so collecting it into a portable file that the job seeker brings directly to the employer when applying for the position is essential. If schools can help their graduates assemble such files, it should be especially useful to minority job candidates, who may face unique barriers when extensive objective selection information is not available.

At the promotion stage, employers can post and publicize all job openings to be filled internally and emphasize objective measurable performance factors in selection. Also, employers need to recognize that the problem of minority under- representation in higher level positions may begin at the job entry stage, because minorities' chances for advancement often depend upon receiving equal oppor- tunity for training within the firm or beginning in a job tied to an upward career track.

Although all these suggestions are certainly worth pursuing to improve equal employment opportunities, our investigations of how specific employer practices are related to the probability of minority employment in jobs at differ- ent levels did not produce strong evidence that current variations in most em- ployer practices had much impact. We separately studied each ofthe 28 practices listed in Tables A2 and A3 in the appendix. After controlling on the sex and

26 Braddock and McParttand

Table 7. The Effects of Employers' Use of Community Agencies in Recruitment on the Probability that a Job is Filled by Whites (by

Educational Level of the Job, Controlling for Five Labor Market Variables)

College-degree jobs (« = 850)

Some-college jobs (« = 1048)

High school jobs in = 2396)

Employers using community agencies

.72

.64

.61

Employers not using community agencies

.81

.73

.67

education composition of jobs, few statistically significant and consistent find- ings identified particular employer recruitment, selection, or promotion practices that produced underrepresentation or overrepresentation of minorities. With the exception of the results reported previously concerning social networks for entry jobs and the identification of internal candidates for job promotions, plus one other major exception, few associations between specific employer practices and job racial composition were uncovered.-^

The other exception involved the use of community groups in employer recruitment for outside candidates. Table 7 shows how minorities' chances of being hired improve when employers use community agencies to recruit appli- cants, even after the race composition of the local labor market and other job characteristics are taken into account. Although this recruitment method is used much less frequently than other methods (Table 2A line 7 shows that less than 15% of employers report that they used the method), our finding has important practical implications. When employers are committed to recruiting minority job candidates or are required to do so by affirmative action regulations, they can get practical help from a community agency such as a local branch of the Urban League or Opportunities Industrialization Centers that specialize in providing minority job applicants. Likewise, local agencies that become known as inexpen- sive and reliable placement services can help individual minority job seekers locate employment opportunities that they would not find in other ways.

Thus, except when employers are motivated to use specialized avenues to accommodate minority candidates, we do not find strong consistent evidence that

^We do not include the tables in this paper that show the absence of consistent significant associations for most employer practices. These tables will be made available on request to other researchers.

Minority Exctusion 27

the current range of variations in most employer recruitment, selection, and promotion practices is related to differences in minorities' chances of em- ployment. Under the present conditions that have produced this range of varia- tion in employer practices, we were unable to fmd convincing evidence that most of the longstanding practical suggestions for simple or straightforward adjust- ments of current employer practices have resulted in reliable and sizable im- provements in equal employment opportunities.

Implications for Outcome-Based Policies

We interpret the preceding results concerning the relationships between the frequency of specific employer practices and minority representation in different jobs as one important set of reasons that outcome-based affirmative action policies are required in employment. Although there are viable methods for recruiting and hiring more minority qualified applicants when an employer is so inclined—such as using relevant community agencies—these methods are not frequently imple- mented and most other employer practices do not penetrate exclusionary barriers under the present conditions of weak employer regulations and incentives.

The need for strong outcome-based policies is best understood when we also consider the specific nature of the current barriers to equal employment oppor- tunities and the absence of voluntary incentives for employers to confront them. Our research indicates that exclusionary barriers (1) continue to restrict equal employment even in the absence of intentional discrimination, (2) are imbedded in the structure of labor markets and major institutions of society, and (3) are reinforced by the usual unregulated incentive systems for employers.

We find many minorities continue to face the exclusionary barriers of segre- gated social networks, infonnation bias, and statistical discrimination in finding entry positions, and these barriers contribute to the problems of closed internal markets frequently faced by minorities within the firm. Minorities face special difficulties in the employment process not only because they are victims of past discrimination in educational and occupational opportunities, but also because of the specific barriers that qualified individuals often encounter at present because of their membership in a race or ethnic minority group. These barriers continue to exclude minorities unfairly even when there is no intention by employers to treat minorities any differently from other potential employees.

We find these barriers are kept in place in part because they are tied to the racial segregation of schools and neighborhoods that persists in modem society and to the white perceptions of racial group differences that derive from unequal educational and employment opportunities of the past and present. Continued segregation supports the exclusionary barrier of social networks in finding job opportunities, because the most serious inequalities occur when networks operate

28 Braddock and McPartland

along racial lines. Segregation also can produce racial bias of information used in selection because white employers will be less familiar and less impressed with the references from segregated sources often used by minority candidates. Simi- larly, the practice of statistical discrimination, which introduces group percep- tions of racial group differences into individual hiring decisions, is based on broad societal factors. Under current conditions, employers often use convenient group images, which are the product of past and present racial inequalities in education and other institutions, to make employment decisions in the absence of clear information about individuals.

Not only are the continuing barriers sustained by major institutions of American society, but there are few strong incentives for employers to overcome these barriers. Indeed, the desire for cost-efficiency contributes strongly to keep- ing these barriers in effect. We find employers have strong incentives to use the simplest and least expensive methods for recruitment and hiring that will yield an effective work force. But the use of simple inexpensive methods often creates the primary conditions for racial exclusionary barriers in employment, such as the use of racial group indicators rather than individual traits in statistical discrimina- tion, and the use of informal recruitment and selection methods involving segre- gated networks and biased information.

It will often cost more for employers to find minority applicants and to obtain selection information that gives each individual a fair chance. But em- ployers are unlikely to assume even modest added costs. Employers do have a strong desire to avoid errors in selecting individuals who will fail as employees, so they will invest in practices to avoid doing so. On the other hand, employers will usually experience no real losses when they discard candidates who would have been equally acceptable to those they actually hired, so employers are not often willing to invest their resources to be more fair to all potential candidates. Thus public policy cannot rely on the usual incentives of employers to penetrate exclusionary barriers and ensure that the rejection of an individual's job can- didacy or the unequal access to pools of job applicants is not related to a person's race or ethnicity.

Because employers need to be strongly motivated to use the specialized methods that can produce qualified minority employees, we believe strong out- come-based policies are required. Because the barriers that unfairly exclude minorities continue to exist and are deeply ingrained in present American institu- tions, we also conclude that effective public regulatory actions in employment will be needed as long as racial segregation and racial stereotypes are so deeply embedded in major institutions of our society. And because employers usually do not have strong business incentives to surmount racial exclusionary barriers and in many cases follow incentives that produce likely conditions for some of these barriers, we also recommend outcome-based public policies that can overcome these labor market factors by requiring fair treatment in employment.

Minority Exclusion 29

References

Aaron, H. J., & Lougy, C M . (1986). The comparable worth controversy. Washington, D.C.: Brookings,

Aigner, D. J,, & Cain, G. G, (1977), Statistical theories of discrimination in labor markets. Indus- trial and Labor Relations Review, 30, 175-187,

Alvarez, R, (1979), Institutional discrimination in organizations and their environments. In R, Alvarez, K, G, Lutterman, & Associates (Eds,), Discrimination in organizations (pp, l-4a), San Francisco: Jossey-Bass,

Baker, P,, Carpenter, S,, Crowley, J, E,, D'Amico, R,, Kim, C , Morgan, W,, & Weilgosz, J, (1984), Pathways to the future: A longitudinal study of young Americans (Voi. 4), Columbus, OH: Center for Human Resource Research, Ohio State University,

Baron, J, N, (1984), Organizational perspectives on stratification. Annual Review of Sociology, 10, 37-69,

Becker, H, J, (1977), How young people find career-entry jobs: A review ofthe literature (Report No, 241), Baltimore, MD: Johns Hopkins University Center for Social Organization of Schools,

Becker, H, J, (1980), Racial segregation among places of employment. Social Forces, 58, 761-776, Berg, I, (Ed,), (1981), Sociological perspectives on labor markets. New York: Academic Press, Bielby, W, T,, & Baron, J, N, (1986), Men and women at work: Sex segregation and statistical

discrimination, American Journal of Sociology, 91. 759-799, Bishop, J, (1986), Preparing youth for employment: Does learning basic skills pay off? Paper

presented at American Educational Research Association meetings, San Francisco, CA, Braddock, J, H,, Crain, R, L,, & McPartland, J, M, (1984), A long-term view of school desegrega-

tion: Some recent studies of graduates as adults. Phi Delta Kappan, 66, 259-264, Braddock, J, H,, Crain, R, L,, McPartland, J, M,, & Dawkins, R, L, (1986), Applicant race and job

placement decisions: A national survey experiment. International Journal of Sociology and Social Policy, 6, 3-24,

Butler, J, S, (1976), Inequality in the military: An examination of promotion time for black and white enlisted men, American Sociological Review, 41, 807-818,

Chamer, I, (1984), The career passport: Documenting youth experience and making the job connec- tion. Washington, D,C,: National Institute for Work and Leaming,

Committee on Economic Development, (1985), Investing in our children. New York: Author, Crain, R, L, (1970), School integration and occupational achievement of Negroes, American Journal

of Sociology, 75, 593-606, Crain, R, L, (1984), The quality of American high school graduates: What personnel officers say and

do about it (Report No, 354), Baltimore, MD: Johns Hopkins University Center for Social Organization of Schools,

Feagin, J, R,, & Feagin, C, B, (1978), Discrimination American style: Institutional racism and sexism. Englewood Cliffs, NJ: Prentice-Hall,

Femandez, J, P, (1975), Black managers in white corporations. New York: Wiley, Femandez, J, P, (1981), Racism and sexism in corporate life. Lexington, MA: Lexington Books, Friedman, T,, & Williams, E, (1982), Current use of tests for employment. In A, Wigdor & W,

Gamer (Eds,), Ability testing. Washington, D,C,: National Academy Press, Glazer, N, (1975), Affirmative discrimination. New York: Basic Books, Granovetter, M, S, (1974), Getting a job: A study of contacts and careers. Cambridge, MA: Harvard

University Press, Granovetter, M, (1982), The strength of weak ties: A network theory revisited. In P, V, Marsden, &

N, Lin (Eds,), Social structure and network analysis (pp, 105-130), Beverly Hills, CA: Sage, Hollenbeck, K, (1984), Hiring decisions: An analysis of Columbus employer assessments of youthful

job applicants. Columbus, OH: Ohio State University, National Center for Research in Vocational Education,

Kalleberg, A, L,, & Sorensen, A, B, (1979), The sociology of labor markets. Annual Review of Sociology, 5, 351-379,

Lin, N, (1982), Social resources and instrumental action. In P, V, Marsden & N, Lin (Eds,), Social structure and network analysis (pp, 131-145), Beverly Hills, CA: Sage,

30 Braddock and McPartland

Lippman, S,, & McCall, J, (1976), The economics of job search: A survey. Economic Inquiry, 14, 155-189,

Maguire, D, C, (1980), A new American justice. New York: Doubleday, Mangum, S, I, (1982), Job search: A review ofthe literature. Report submitted to U,S, Department

of Labor, Employment and Training Administration, Office of Research and Development, McCall, J, J, (1972), The simple mathematics of information, job search and prejudice. In A, H,

Pascal (Ed,), Racial discrimination in economic life (pp, 205-224), Lexington, MA: D, C, Heath,

McPartland, J, M,, & Crain, R, L, (1980), Racial discrimination, segregation and processes of social mobility. In V, Corello (Ed,), Poverty and public policy (pp, 97-125), Boston: G, K, Hall,

McPartland, J, M,, Braddock, J, H,, II, & Dawkins, R, L, (1986), Education and work: How American firms use schools and education information at different stages ofthe employment process. Baltimore, MD: Johns Hopkins University Center for Social Organization of Schools,

McPartland, J, M,, Dawkins, R, L,, & Braddock, J, H, (1986), The school's role in the transition from education to work: Current conditions and future prospects (Report No, 362), Bal- timore, MD: Johns Hopkins University Center for Social Organization of Schools,

National Academy of Science, National Academy of Engineering, Institute of Medicine, and Com- mittee on Science, Engineering and Public Policy (1984), High schools and the changing workplace: The employer's view. Washington, D,C,: National Academy Press,

Pames, H, S,, Miljus, R, C , Spitz, R, S,, & Associates, (1970), Career thresholds. Volume I. Washington, D,C,: U,S, Department of Labor Manpower Research,

Rosenbaum, J, E, (1984), Career mobility in a corporate hierarchy. New York: Academic Press, Rosenfeld, C, (1975), Jobseeking methods used by American workers, Monthh Labor Review,

98(8), 39-42, Rosenfeld, R, A, (1980), Race and sex differences in career dynamics, American Sociological

Review, 45, 583-609, Rossi, P, H,, Berk, R, A,, Boessel, D, P,, Eidson, B, K,, & Groves, W, E, (1968), Between white

and black: The faces of American institutions in the ghetto. In National Advisory Commission on Civil Disorders (Ed,), Supplemental studies (pp, 69-153), Washington, D,C,: U,S, Gov- emment Printing Office,

Rossi, P, H,, Berk, R, A,, & Eidson, B, K, (1974), The roots of urban discontent: Public policy, municipal institutions and the ghetto. New York: Wiley,

Sorensen, A, B, (1983), Processes of allocation to open and closed positions in social structure, Zeitschrift fur Soiiologie, 12, 203-224,

Sorensen, A, B, (1984), Inequality of opportunity in internal and competitive labor markets. Paper presented at Office of Naval Research/Chief of Navy Education and Training Symposium on Minorities and High-Technology Organizational Success, Pensacola, FL,

Spence, M, A, (1973), Job market signaling. Quarterly Journal of Economics, 87, 355-374, Tenopyr, M, L, (1981), The realities of employment testing, American Psychologist, 36, 1120-1127, Thurow, L, (1975), Generating inequality. New York: Basic Books, Thurow, L, (1969), Poverty and discrimination. Washington, D,C,: The Brookings Institution, U,S, Department of Labor (1975), Job-seeking methods used by American workers (Bulletin 1886),

Washington, D,C,: U,S, Government Printing Office, U,S, Department of Labor (1976), Recruitment, job search and the United States Employment

Service. R&D Monograph 43, Washington, D,C,: U,S, Govemment Printing Office, U,S, Equal Employment Opportunity Commission, (1974), Affirmative action and equal em-

ployment: A guidebook for employers. Washington, D,C,: U,S, Govemment Printing Office, U,S, Department of Education, (1986), What works: Research about teaching and learning. Wash-

ington, D,C,: U,S, Govemment Printing Office, Wilson, W, J, (1978), The declining significance of race. Chicago: University of Chicago Press,

Minority Exclusion 31

Appendix

We developed a sampling plan and instrument design to provide more direct tests of how certain employment practices may affect the occupational chances of minorities. The sampling plan used a stratification approach that would yield large samples of jobs typically filled by each of the three major race-ethnic subgroups in our nation (whites, blacks, and Hispanics), The survey instrument asked questions of employment officials that focused on a specific job title and description, to identify the major recruitment, selection, training, and promotion practices involved.

The Sample

We defined the sampling strata and directory for the selection of jobs by using a nationally representative sample of young adult workers covering large numbers of each race-ethnic target group that provided information on each individual's job, employment location, sex, race-ethnicity, age, and educational attainment. This initial sample of workers was the 1976 and 1979 follow-up surveys of the National Longitudinal Survey of the High School Graduating Class of 1972 (NLS), available from the U,S. Department of Education National Center for Education Statistics, We used NLS to establish the sampling frame of jobs within six strata defined by the NLS respondents' sex and race/ethnicity. Because enough time had passed since the high school graduation of NLS re- spondents to permit most individuals who had gone to college to complete their degree (four years for the 1976 job and seven years for the 1979 job), the NLS data file offered a large nationally representative sample of jobs recently held by young adults with different amounts of completed education within each sex/- race-ethnicity stratum. We used telephone directory services to find the mailing addresses and phone numbers of NLS employers, derived from the information on the NLS questionnaires providing employers' names and respondents' resi- dential locations. A brief telephone call was made to each identified employer to check the NLS sample job title and job duties at that place of work and to request participation in the survey. This process produced usable addresses for 90.2% of the initial sample frame of employers and jobs, for a sample of 5493. Through a series of mail and telephone surveys in 1983, we received completed question- naires from 4078 employers—74.2% of the sample—for whom accurate ad- dresses and job descriptions had been obtained. Because we had established sampling strata to provide representation in each major sex/race-ethnicity group, our achieved sample included jobs held by 1960 white NLS respondents, 1518 black NLS respondents, and 600 Hispanic respondents. Thus our sample is both a sample of individuals and a sample of jobs. Depending upon the analyses, weights were calculated to accurately reflect either the sampling proportions used

32 Braddock and McPartland

in the original NLS sampling frame of individuals or in our sampling strata of jobs.

Table Al presents a comparison of 1983 U.S. Census national distributions of job characteristics and our 1983 weighted sample of jobs, to investigate the representativeness of the sample used in this paper. The actual achieved sample size in major job categories is also shown in Table A l , to reflect the actual sampling variation available for studies of relationships between job charac- teristics. With occasional exceptions. Table Al provides reassurance that our 1983 sample of 4078 jobs is an adequate representation of jobs in the various sectors, industries, demographic categories, and labor market locations of our nation.

Because our initial source for the sampling frame of jobs was a national sample of young adult workers who had at least graduated from high school, we expected some bias in our achieved sample toward higher level jobs held by younger workers. On the other hand, because many job titles filled by young workers are usually also held in the firm by other workers from throughout the age and educational attainment distributions, we expected our sampling approach to yield large numbers of cases and the full range of variability for all categories and segments of the American occupational structure. Table Al shows some sample bias, reflecting more jobs held by younger workers who had at least achieved a high school education, but a sufficient sample base across all major job segments minimizes the likelihood that estimates of relationships among job attributes would be misleading. In particular. Table Al shows the weighted sample to be a good representation of the national distributions of (a) job sector; (b) industry; (c) occupation, except that the sample underrepresents low-level factory jobs (operators, fabricators, and laborers) and overrepresents high level and supervisory positions (managerial and professional specialties); (d) job loca- tion and size of establishment; and (e) demographic characteristics of job incum- bents, except that the sample underrepesents jobs held by workers aged 40 or over, overrepresents jobs held by workers in the age range 26-39, and overrepre- sents jobs held by workers with some college. Because the actual achieved sample includes large numbers of cases in the job categories that are propor- tionally under- or overrepresented, we believe estimates of relationships between job attributes will be accurately estimated by our sample, although caution for possible bias needs to accompany point estimates of averages, percents, and standard deviations based on our sample.

Measures and Method

We sent an 18-page questionnaire to each employer in our sample. Most of the questions focused on the specific sample job that had been identified by an individual NLS respondent. Some of these questions covered the demographic

Minority Exclusion 33

Table A l . A Comparison of the 1983 U,S, Census and the Weighted Sample of Employers on Selected Job Characteristics and the Achieved Sample Size for Different Job Categories

Job characteristic

Sector (percent distribution) Public Private

Industry (percent distribution) Agriculture and mining Construction Manufacturing Transportation Trade Finance Services Public administration

Occupation (percent distribution) Managerial and professional specialty Technical, sales, and administrative support Service occupations Precision production, craft, and repair Operators, fabricators, and laborers Farming, forestry, and fishing

Location and size of establishment Region (percent distribution)

Northeast Midwest South West

Size of establishment (percent distribution) Under 20 employees 20-99 employees 100-249 employees 250 or more employees

Race, sex, education, and age of job incumbents* Percent male Percent female

Percent white Percent black Percent Hispanic Percent other ethnicity

Percent high school graduate or less Percent some college Percent 4 year college or more

Percent age 25 or younger Percent age 26-39 Percent age 40 or older

U,S, Census 1983 employed civilian

labor force"

17,6 82,4

4,4 6,1

19,8 6,9

21,0 6,4

30,7 4,7

23,4 31,0 13,7 12,2 16,0 3,7

21,3 25,3 33,4 20,1

26,8 28,5 14,4 30,4

56,3 43,7

82,9 9,3 5,2 2,5

56,1 18,4 25,4

22,8 36,6 40,7

Weighted sample of 1983 jobs

19,4 80,6

2,3 4,5

18,8 7,2

20,5 6,6

34,3 5,6

34,9 34,7 10,1 12,5 4,8 3,0

23,0 30,1 30,2 16,7

23,2 26,4 12,8 38,2

48,4 51,6

82,6 10,7 4,7 1,7

49,4 25,7 24,9

25,0 47,0 22,7

Actual sample size (number

of jobs)

978 3100

80 145 739 291

1429 305

1340 259

1228 1499 422 551 229 133

660 870

1791 750

827 946 601

1704

2016 2062

2716 584 242

32

2396 1048 850

823 1874 714

"Source: Statistical Abstract ofthe United States, 105th Edition (1985), Tables 654, 657, 658, 659, 667, 674, 676, 678, and 690,

*The actual sample size of jobs for the fmal set of characteristics is the number of sample jobs where at least 50% of incumbents have the particular race, sex, education, or age trait under consideration.

34 Braddocii and McPartland

distribution of current workers in the sample job, including their sex, age, race- ethnicity, and educational attainment. We also asked about specific employer practices used to recruit candidates, and about information used to hire from within the firm and from outside to fill openings in the sample job. We asked employers to indicate how frequently each practice was used and to rank the most important practices for the final determination of who would fill the job. We also asked each employer to rate the importance of 16 specific worker qualifications for successfully filling the sample job, and to estimate the percent of recent openings that were filled by promotions or external hires and the usual starting salary in the sample job. We also asked some questions about the establishment as a whole, including the size and race and sex distribution of the total work force, and the policies, if any, concerning affirmative action.

Data Analyses

Three types of analyses were conducted for this paper: (a) descriptive tab- ulations of the distribution of employer practices shown in Tables A2-A5; (b) estimates of the relationships between job characteristics, shown in Tables 1,3, 5-7; and (c) estimates of the relationships between individual worker charac- teristics and job outcomes, shown in Tables 2 and 4,

In each type of analysis, we use demographic characteristics of current workers in each job to create different job categories. We categorize "male jobs" or "female jobs" depending upon whether males constitute at least 50% ofthe current job incumbents or females constitute at least 50% ofthe current job incumbents. Similarly, we categorize jobs by their education level either as "high school jobs", "some-college jobs" or "college-degree jobs," depending upon which level of education has been completed by 50% or more of the current employees in each job. Employment sector (private or public employer) is an additional variable on which we categorize jobs.

Descriptive Findings

Table A2 shows the percentage of employers who report they frequently use each recruitment, selection, or promotion practice for jobs, within three broad categories of the education level of workers in the job. For ease of presentation, percentages are shown for private sector jobs filled primarily by males. Adjust- ment factors are shown to indicate approximately what would be added or sub- tracted to obtain percentages for the public sector or for female jobs. These adjustment factors are the unstandardized regression coefficients from a multiple regression equation where a particular employer practice is the dependent vari- able with three independent variables to measure the education level of the job.

Minority Exclusion 35

the sex composition of the job, and the job sector (each with possible values of zero and one to match the categorical presentation of Table A2),

Tables A3-A5 follow the same format as Table A2 to present, respectively, the percent of employers who rate each practice as most important in finding the actual person who is given the job, the percent of employers who rate each worker trait as extremely important, and the percent of employers who chose each trait as most important.

Table A2. Percentage of Employers Using Different Recruitment, Selection, and Promotion Methods, by Educational Level of Job, with Adjusted Factors for Sector and Sex Composition of Job

Employer practice

Recruitment method 1, Friends of employees 2, School placement service 3, Professional organizations 4, Civil Service 5, Public employment service 6, Private employment service 7, Community agencies 8, Media ads 9, Walk-ins 10, Union referral

Hiring information 11, Employer recommendations 12, Test results 13, Education level or type 14, Education grades 15, Education recommendations 16, Education reputation 17, Union membership 18, License or certification

Promotion information 19, Production record 20, Seniority 21, Internal recommendations 22, Test results 23, Education level or type 24, Education grades 25, Education recommendations 26, Education reputation 27, Union membership 28, License or certification

Education level

High school

38 14 4 3

32 3

13 26 59 10

65 22 22 4

10 8

10 9

50 57 49 16 16 2 6 6

20 9

Some college

37 27

g 4

24 10 12 34 52 6

68 26 39 15 19 20 4

12

60 38 65 21 26 8

10 g

11 7

of job

College degree

3g 44 17 3

16 16 g

34 51 5

74 19 75 32 35 37

2 14

59 18 76 12 57 12 17 25

1 12

Adjustment

Sector (public)

- 1 6 + 2 + 3

+ 34 +4 - 5

+ 10 - 4

- 1 5 - 1

0 + 22 + 21

+ 3 +5 - 2 - 1 + g

0 - 5 + 6

+ 20 + 22 + 3 +7

0 - 5 +9

factor" for

Job sex (female)

+ 1 + 3 - 1 - 3 - 1

0 - 1 + 5 + 8 - 4

+2 +9 + 6 - 1 + 5 + 2 - 4 + 5

+ 11 - 3 + 6 + 6

+ 10 + 3 + 6 + 3 - 9 + 5

"Percentages shown are for private sector, male jobs. Add appropriate adjustment factor(s) to obtain other combinations of sector and job sex.

36 Braddock and McPartland

Table A3. Percentage of Employers Who Cite Each Recruitment Selection and Promotion Method as Most Important for Their Decision (by Education Level of Job, with Adjustment Factors for Sector

and Sex Composition of Job)

Employer practice

Recruitment method 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 10a.

Friends of employees School placement service Professional organizations Civil Service Public employment service Private employment service Community agencies Media ads Walk-ins Union referral Other (miscellaneous)

Hiring information 11. 12. 13. 14. 15. 16. 17. 18. 18a.

Employer recommendations Test results Education level or type Education grades Education recommendations Education reputation Union membership License or certification Other (interview)

Promotion information 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 28a.

Production record Seniority Internal recommendations Test results Education level or type Education grades Education recommendations Education reputation Union membership License or certification Other (performance ratings)

Education level

High school

24 4 1 3

16 2 0

13 26 4 6

50 10 6 5 5 0 4 2

19

25 21 21

5 4 0 0 0 2 1

22

Some college

18 9 2 3 6 4 1

24 15 2

14

42 14 9 4 7 0 1 3

20

22 7

24 8 9 0 1 0 1 2

25

of job

College degree

17 24 5 3 6 8 0

17 11 0 8

31 6

29 2 5 0 0 2

25

23 2

27 4 7 0 1 0 0 2

33

Adjustment

Sector (public)

- 1 0 0 0

+22 0

- 3 + 1 - 8 - 5 - 1 +5

- 1 3 + 14

+5 - 3 - 2

0 - 1 + 1 - 1

- 6 - 6 - 3

+ 11 +6

0 0 0 0

+ 1 - 2

factor" for

Job sex (female)

- 3 + 1 + 1 - 1 + 1

0 + 1 + 6 +2 - 2 - 3

- 1 + 1 - 2

0 +3

0 - 2 + 4 - 2

+ 7 - 5 - 5

0 + 2

0 0 0

- 1 +4 - 3

"Percentages shown are for private sector, male jobs. Add appropriate adjustment factor(s) to obtain other combinations of sector and job sex.

Relations Among Job Characteristics (Tables 1, 3, 5, 6, 7)

Tables 1 , 3 , 5 , and 7 are derived from multiple regression analyses of jobs, where the dependent variable is the percent white of current workers in each job, and the independent variables include five labor market variables (region, per-

Minority Exclusion 37

Table A4. Percentage of Employers Who Rate Various Worker Qualifications as Extremely Important, by Educational Level of the Job (Sample Size = 4078)

Worker qualifications

Methodical Manual dexterity Quick learner Basic adult literacy Advanced readers Perform basic arithmetic Excellent at math Specialized knowledge Client relations Permanence Growth potential Good team members Proper attitudes Dependable Good judgment Can supervise

High school

48 61 47 50 13 44

8 30 32 36 22 68 82 96 50 20

Education level

Some college

45 44 64 72 28 71 16 34 48 44 28 79 84 95 72 32

of job

College degree

31 22 74 88 52 80 36 47 60 44 46 85 80 95 88 43

Adjustment factor" for

Sector (public)

- 3 - 1 1

- 5 +4 +8

- 1 1 - 3 + 4 + 1 - 9 - 4 - 2

0 - 1

0 +2

Job sex (female)

+ 14 - 6 + 6

+ 14 + 3 + 6

0 - 2

+ 12 - 3 - 6 + 7 +4 + 2 +5 - 3

"Percentages shown are for private sector, male jobs. Add appropriate adjustment factor(s) to obtain other combinations of sector and job sex.

Table A5. Percent of Employers Who Select Each Worker Qualification as the Most Important in Filling a Job at Different Educational Levels

Worker qualification

Methodical Manual dexterity Quick learner Reading ability Math ability Specialized knowledge Client relations Permanence Growth potential Good team member Proper attitudes Dependable Good judgment Can supervise Other (Sample size)

High school

5,2 9,6 9,1 3,7 3,5

14,1 7,8 0,9 0,5 5,0

11,9 21,3 2,9 1,3 3,1

(1112)

Education level of job

Some college

3,4 4,9

11,4 3,2 4,1

23,1 7,5 1,0 1,7 5,3

11,2 12,1 6,6 1,5 3,2

(412)

College degree

1,7 1,0 9,9 2,7 2,2

37,2 8,9 0,5 3,2 3,9 7,6 4,4

10,6 2,5 3,7

(406)

38 Braddock and McPartland

cent white in the local labor market, private or public employment sector, per- cent male of current workers in the job, and percent of current workers whose education went no further than high school) plus one other variable of interest. The final variable in Table 1 is employer's use of social networks; in Table 3, it is employer's rating of a selected worker trait; in Table 5, it is employer's use of community agencies in recruitment. Following the estimation of the above multi- ple regression equations, we derive the probabilities shown in each table by substituting the population mean into the equation for the five labor market variables and substituting either the highest or the lowest possible values for the final variable of interest.

Table 6 reports results from multiple regression analyses of two subsamples of jobs; those jobs that are filled from within the firm at least 50% of the time, and those jobs that are filled by outside hires at least 50% of the time. Multiple regression analyses use job hourly pay rate as the dependent variable and percent black workers in the job (regression coefficient shown in Table 6), percent workers with a college degree in the job, percent male workers in the job, and public or private sector.

Relationships Between Individual Traits and Job Traits (Tables 2 and 4)

Table 2 is a tabulation of average job outcomes in the private sector for black high school graduates for different types of networks of friends and ac- quaintances used to find the job. These categories include "did not use net- works," and "used segregated networks," defined by those who graduated from segregated schools and used social networks to find their jobs; and "used deseg- regated networks," defined by those who graduated from desegregated high schools and used social networks to find their job.

Table 4 shows the results of multiple regression analyses where the depen- dent variable is the employer's rating ofthe importance of a selected worker trait on the job and the independent variables are the race of an individual in the job (coefficient shown), the individual's sex, the individual's educational attain- ment, the job sector, and the individual's score on a test of the selected trait.

JOMILLS HENRY BRADDOCK II is Principal Research Scientist at the Center for Social Organization of Schools and Professor of Sociology at Johns Hopkins University. His broad research interests encompass issues of inequality and social justice including extensive examination ofthe long-term effects of school desegregation on black young adults and racial differences in career attainment. His related works include Institutional Discrimination: A Study of Managerial Recruitment in Professional Football.

Minority Exclusion 39

JAMES M. MCPARTLAND is Co-Director ofthe Center for Social Organization of Schools and Professor of Sociology at Johns Hopkins University. His research interests have focused on school effects on student development, school to work transitions, and social processes of race and sex equity.