Labor markets and organizations: A screening theory of hiring networks and racially homogeneous employment

A

n[eiScCsattap©

K

lpcwct

0

Research in Social Stratification and Mobility 25 (2007) 39–55

Labor markets and organizations: A screening theory of hiringnetworks and racially homogeneous employment

David Jacobs a,∗, Benjamin Cornwell b

a Department of Sociology, The Ohio State University, 300 Bricker Hall,190 N. Oval Mall, Columbus, OH 43210, United States

b University of Chicago, USA

Received 30 August 2005; received in revised form 4 April 2006; accepted 15 May 2006

bstract

This theoretical analysis uses three relationships between individual and organizational performance to help explain when orga-izations will use different labor markets to organize employment and when they will use different hiring networks. Granovetter’sGranovetter, M. (1973). The Strength of Weak Ties. American Journal of Sociology, 78, 1360–1381] distinctions about job seek-rs’ networks are combined with Jacobs’ [Jacobs, D. (1981). Towards a theory of mobility and behavior in organizations: Annquiry into the consequences of relationships between individual performance and organizational success. American Journal ofociology, 87, 684–707] and Marsden and Campbell’s [Marsden, P. V., & Campbell, K. E. (1990). Recruitment and selection pro-esses: The organizational side of job searches. In R. Breiger (Ed.), Social mobility and social structures (pp. 59–79). New York:ambridge University Press] work on employers’ recruitment and screening strategies to develop insights about when weak or

trong ties will be used in the hiring process. Strong ties between a job seeker and an intermediary and between the intermediarynd an employer are possible without also assuming a tie between the job seeker and the employer. This is possible because theie between the job seeker and the intermediary exists in a different kind of network than the tie between the intermediary and
he employer. When employees can only hurt collective efforts and employers are concerned about the quality of informationbout job seekers, administrators will prefer strong ties. These ideas also explain when ethnic homogeneity in hiring will be mostrevalent.
2007 Elsevier Ltd. All rights reserved.

ational

issues, we lack an integrated theory about these criticalprocesses.1 After briefly discussing some necessary

eywords: Theory of Organizational Careers; Individual and Organiz

There is little consensus about the multiple socio-ogical theories of labor markets. In the absence of alausible approach that takes both organizational pro-esses and networks explicitly into account, empirical
ork will remain handicapped. Organizational factors
ontribute to the kinds of labor market networks usedo fill vacancies, yet despite the importance of such

∗ Corresponding author.E-mail address: [email protected] (D. Jacobs).

276-5624/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.rssm.2006.05.003

Success; Networks, minorities, and hiring

1 The lack of an organizational theory about labor markets troubleseconomics as well as sociology. Probably labor market outcomes arethe most problematic anomaly in this discipline. Despite presump-tions that markets always clear, labor markets do not. Even after sharpreductions in the demand for labor, labor prices in wages and salariesrarely fall. In the 1930s when unemployment went beyond 25%, realearnings stayed well above their pre depression levels (Gordon, 1990,p. 190). Pay levels typically persist after declines in the demand for

mailto:[email protected]

dx.doi.org/10.1016/j.rssm.2006.05.003

ial Stra
40 D. Jacobs, B. Cornwell / Research in Soc
elements from prior work (Jacobs, 1981), we present acombined organizational and network framework basedon position characteristics that condition administrativedecisions about hiring, careers, and other labor marketoutcomes.

Although this perspective will be used to explainmany labor market and career outcomes, at the endof this paper we emphasize one extremely importantconsequence. Work on gender and employment has con-sidered organizational level processes (Kirschenman &Neckerman, 1991), but research on minority hiring haslargely neglected these effects (but see Braddock &McPartland, 1987). The theory we develop here suggeststhat organizations do not use the same hiring networkswhen they fill different positions. This divergence createssome crucial obstacles that minorities face when theyseek many jobs. As a result, minority hiring is automat-ically reduced in these positions. These obstacles facedby minorities are based on the screening methods andresulting networks that provide most optimal method fororganizations to select prospective employees when theyfill vacancies for such positions.

Granovetter’s (1973) seminal insights about theeffects of weak tie networks provide a foundation forthis analysis. In contrast to strong ties, actors connectedby weak ties obtain superior information, so job seekerslinked by weak ties should get better jobs. Yet Gra-novetter’s predictions about the efficacy of weak tiesapparently do not hold when some common conditionsare present. Murray, Rankin, and Magill (1981) find thatstrong rather than weak ties are more important whendemand for labor is weak. Other analyses (Bridges &Villemez, 1986; Marsden & Hurlbert, 1988) suggestthat weak ties do not help applicants get better payingjobs, while additional findings (Lin, Walter, & Vaughn,1981a,b) are not always consistent with Granovetter’sexpectations.

Our goal is to develop a combined organiza-
tional and network framework that uses positioncharacteristics to explain some of these anomalies. Fol-lowing Marsden and Campbell (1990), we see the job
labor (and competition often does not erode discriminatory practices)because organizations protect employees from outsiders who are will-ing to take their jobs at lower wages. Most economists overlook sucheffects by assuming that organizations are undifferentiated unitaryactors (Stiglitz, 1991; Williamson, 1975) that must quickly adjustto the inescapable pressures of competition to survive. While someeconomists have provided useful conceptual efforts to deal with thesecritical anomalies (Holmstrom and Tirole, 1989; Lilien and Hall, 1986;Parsons, 1986), these remedies remain peripheral to the neo-classicalparadigm.

tification and Mobility 25 (2007) 39–55

seeker–intermediary–employer triad as produced by dif-ferent social relationships. Labor markets are dualisticsettings in which job seekers’ and employers’ utiliza-tion of networks must be considered simultaneouslybecause the ties between job seekers and interme-diaries often are qualitatively different from the tiesbetween intermediaries and employers. Such ties arebest viewed as existing in separate but co-existing net-works. This distinction will let us relax Granovetter’s“forbidden triad” redundancy argument as it allows sit-uations in which strong ties between A and B as wellas strong ties between B and C could exist without a tieexisting between A and C. Recognition of these sep-arate networks will let us develop what seems to bea new approach that highlights organizational condi-tions that make strong ties more useful than their weakcounterparts.

The argument presented in this paper rests on Jacobs’(1981) classification of three common organizationalpositions. These positions are defined by how they shaperelationships between individual and organizational per-formance. When Granovetter’s contrasts between strongand weak ties are combined with this positional typol-ogy, the conditions that determine when his argumentapplies and when it does not can be spelled out withgreater precision. Granovetter’s claims about the supe-riority of weak ties make prima facie sense from thejob seeker’s standpoint, but weak tie networks are notalways best for employers. Our presentation will showthe conditions when administrators will try to use strong-tie hiring networks. From the employer’s standpoint, anoverriding concern is the quality of information par-ticularly when some organizational vacancies must befilled. These incentives operate coextensively with thejob seeker’s need for a greater quantity of informationabout vacancies.

We begin by presenting (with some refinements)postulates about the effects of different relationshipsbetween individual and organizational performancebased on the typology in Jacobs (1981). To developnew perspectives on hiring networks, the next sectionbriefly summarizes Jacobs’ discussion about howthese considerations affect careers, but it adds newmaterial on the kinds of labor market arrangementsthat organizations use in units dominated by thesepositions. The third section combines these ideas withGranovetter’s network approach to provide a novelargument about when strong and weak ties will be most
useful to employers. We go on to use this frameworkas the basis for new ideas about some organizationalconditions that will determine when minorities faceincreased obstacles when they seek employment.

ial Stra

1ntstutoimptsltmea

1

t(wamoso

Smdmtp

towplwewacbmt(

D. Jacobs, B. Cornwell / Research in Soc

This presentation will focus on effects (see Blalock,969) without making exhaustive efforts to explain alter-ative causes.2 While this analysis relies on a positionalypology, other explanations clearly matter. Althoughuch similar practices may not be optimal, institu-ional theory suggests that organizations sometimes willse identical hiring procedures to fill dissimilar posi-ions because these hiring choices must be legitimatedr because administrators must abide by externallymposed uniform standards. The larger point is that one

odel cannot capture all that happens in such complexrocesses, so, in what follows, ceteris paribus assump-ions will always be in force. This theoretical analysispells out a model of organizational hiring and otherabor market outcomes by focusing single-mindedly onhe effects of relationships between individual perfor-ance in organizational positions and the success of the

mploying organization, but no claim is made that otherpproaches do not provide equally useful insights.3

. First principles: a typology of positions

We take the view that the labor market is dualistic inhat job seekers use their networks to obtain positionsGranovetter, 1973) while organizations use their net-orks to fill them (Marsden & Campbell, 1990). In suchsetting it is important to consider simultaneously the
ethods used by individuals and organizations to locate
ne another. Prior work (Fernandez & Weinberg, 1997)hows that effects that operate at both the individual andrganizational levels condition the importance of social

2 Space limits will often make exhaustive comparisons impossible.ome propositions may seem idiosyncratic because they will not beatched by a discussion of the same outcome when alternative con-

itions are in force. The reader should assume that if an outcome isentioned when some conditions are discussed, but it is ignored in

he analyses of opposite conditions, the outcome is unimportant or notresent in the latter case.3 We assume that actors satisfice. They try to follow incentives, but

hey lack complete information, and they rarely maximize or focus onne goal to the exclusion of all others. Such an approach only requireseak assumptions about action: actors are likely to substitute acts thatrobably will be more beneficial (or less harmful) for acts that areikely to be less beneficial (or more harmful). As subsequent examplesill show, even the managers of nonprofits existing in uncompetitive

nvironments must try to control harmful conduct, so the frameworke put forward does not require rationality or profit seeking. We only

ssume that most actors faced with uncertainty respond as best theyan to incentives. Actors may be consciously rational, but appositeehavior can just as easily result from less calculative devices suchimicry, unpremeditated trial and error, or the dictates of an organiza-

ional culture that stipulates optimal behavior without giving reasonsElster, 1989).

tification and Mobility 25 (2007) 39–55 41

capital because referrals are more useful at differentstages in the hiring process. Such studies emphasize thecontingent worth of network ties to job candidates. Here,we highlight another contingency by focusing on how theuse of strong or weak tie networks is conditioned by thecharacteristics of the vacancy. We begin by discussingthree foundational relationships between individual andorganizational performance and go on to explain howorganizational attributes shape these relationships.

1.1. Three contrasting links between individual andorganizational performance

1.1.1. A core distinctionWe begin by stipulating three position types clas-

sified by their relationship between individual andorganizational performance outlined in Jacobs (1981).Stinchcombe (1963) and Blalock (1967) provide a startby pointing out that individual performances in someorganizational positions have substantial effects on orga-nizational success, but in other positions individualperformance differences are largely irrelevant. Successin university research positions or in entertainment willhelp the employing organization, but a stellar perfor-mance by an outstanding house painter or primaryteacher can add little to the likelihood that the employingorganization will achieve it’s goals. Therefore, in somepositions an exemplary individual performance addsmuch to an organization’s total accomplishments whilepoor performances will not matter, but in other positionseven extreme individual performance differentials willhave little effect on organizational success (see Goodmanand Pennings, 1977 for ways to measure organizationalperformance). Thus, while there is variance in individ-ual productivity, position effects determine whether thisvariance will matter to organizations.

The scholar who wins a Nobel Prize adds to a univer-sity’s ability to gain resources from its environment, butscholars rarely do much harm to universities. Yet some-one in another position, like the instructor in a juniorcollege whose teaching is just as superb as a Nobel Prizewinner’s research, can have little effect on his or herorganization (see Fig. 1b). The first positions will becalled goal enhancing (Fig. 1a) while the second willbe called largely irrelevant (Fig. 1b). The vertical loca-tion of the line that defines goal enhancing positions inFig. 1a and its steep slope at the top imply that orga-nizations can accept extreme differences in individual
performances in these jobs because the worst will not domuch harm, but the best will add to the probability of col-lective success. Yet in largely irrelevant positions evenextreme variations in individual performances will have

42 D. Jacobs, B. Cornwell / Research in Social Stra

Fig. 1. The three relationships between individual and organizational

a few passengers are killed, the airline need not planfor increased ticket sales. The potential for harm over-whelms all other aspects of goal destructive positions.5

4 Additional examples of goal enhancing jobs include editors in pub-lishing houses (Powell, 1985), record producers, and people who sellexpensive goods. Examples of largely irrelevant jobs include pooltypists, most clerks, house painters (Stinchcombe, 1963), buildingcleaners, and retail salespersons. Public sector and other jobs wheremistakes or corruption can lead to bad publicity, jobs that involvethe operation of expensive but vulnerable equipment particularly ifits improper use endangers others, and assembly line work due tothe intense sequential interdependence between these positions areillustrations of goal destructive jobs.

5 By changing the definition, it is possible to alter the relationshipshown in Fig. 1c so a stellar performance in a goal destructive jobcan be seen as helping an organization. But a downward movementof the vertical location of the horizontal axis hides critical aspectsof reality because it conceals the damage that people in such jobscan do. Consider a statement by John Dunlop on how the capacityfor harm can overwhelm the potential to help collective endeavors.“. . . with many years of experience setting wages and salaries as aarbitrator, mediator, or with wage controls (and as secretary of labor),it is natural to advise students that some employees are more likely tobe paid according to the damage or the harm they could inflict by notperforming their tasks up to standard than by what they contribute tothe value of production. . .. Consider a roller in a steel mill with red hot

performance (relationships are exaggerated for illustrative purposes).(a) Goal enhancing positions, (b) largely irrelevant positions, and (c)goal destructive positions.

little influence on collective success because the slopethat defines this relationship is so modest (see Fig. 1b).

1.1.2. A third relationshipIn many positions, however, an outstanding individual

performance can add little to organizational success, butif individual performance is poor, important goals willbe jeopardized. A mistake by an airline pilot or by anoperator of a nuclear power plant may be disastrous, but


superb performances in either job will not improve theexpected performance of the employing firm. Therefore,in some positions exemplary individual performancescannot contribute much to an organization’s success,but an occasional poor performance may sharply reduceorganizational accomplishments. Positions subject tothese constraints will be called goal destructive (Fig. 1c).Performance variations in goal destructive positions canbe extremely costly as the worst performances can domuch damage but the most exemplary performances canadd little to collective success.4

Of course, magnificent acts in a goal destructive jobmay save an organization from considerable harm, buteven such heroic deeds will not add to collective suc-cess. Suppose by super human effort, an airline pilotor a nuclear plant operator avoids a terrible accident.Neither feat will increase the revenues of the airline orthe utility. Perhaps the most compelling aspect of bothillustrations is an organization’s willingness to publi-cize these splendid acts. Organizations will make everyeffort to conceal such events because exposure will becostly. But sometimes publicity is inescapable. When aheroic pilot manages to land a crippled aircraft so only

metal running through rollers at 30 or 40 miles/hr. An error or lack ofattention could be enormously expensive.” (personal letter 1990; ouritalics). What is at issue is whether our definition with its stress on thepotential for harm is useful. We therefore define the three relationshipsin Fig. 1 both by their slopes and by their vertical location.

D. Jacobs, B. Cornwell / Research in Social Stra

Fd

Tcwto

1r

1

rfcetwiptbwloiefe

pfp

ig. 2. Illustrations of reciprocal, sequential, and pooled interdepen-ence.

hese three relationships, however, must be theoreti-ally grounded. To avoid a seemingly ad hoc analysis,e next spell out some of the most important factors

hat shape these relationships between individual andrganizational success.

.2. Organizational causes of the threeelationships between positions and organizations

.2.1. InterdependenceIf interdependence between positions is sequential

ather than pooled (Thompson, 1967; see Fig. 2) so Aeeds B who feeds C, the chain of productivity onlyan be as strong as its weakest link. If no other pathsxist, positions arranged this way will be goal destruc-ive (Fig. 1c; see footnote 5 for a good example). Yethen positions are pooled because tasks are similar and

ndividual performers are autonomous, the harm from aoor performance will be more limited and goal destruc-ive positions will be unlikely. As tasks do not flowetween multiple actors at the same organizational levelhen interdependence is pooled, incumbents will have

ittle effects on the performance of others. This auton-my means that pooled interdependence almost always
s a necessary (but never a sufficient) condition for goalnhancing positions (Fig. 1a) that allow a stellar per-ormer to add to collective success but limit the negativeffects of poor individual performances.6 Pooled inter-
6 We ignore reciprocal interdependence (the third form of interde-endence stipulated by Thompson (1967) where A feeds B who directlyeeds A again) because the occupants of reciprocally interdependentositions often can at least temporarily overcome weak performances


dependence makes largely irrelevant or goal enhancingpositions more likely, but sequential interdependencewithout alternative paths will automatically produce goaldestructive jobs.

1.2.2. Frequency of successIn some jobs failure is rare. When mean success

rates are 99 percent, the inept can fall well below thismean, but there is a ceiling on the difference betweenthe best performance and an average one. If incumbentshandle vital tasks (Hickson, Hinings, Lee, Schneck &Pennings, 1971), the rare individual performance thatfalls well below the mean should hurt the organization,but it will be difficult to add to collective success. Thedifference between a superb performance and one at themean will be too modest. But in jobs in which successis infrequent, the best performers can go well beyondthe mean. If incumbents in positions in which successis rare handle vital tasks (Hickson et al., 1971), a stel-lar performance can add much to collective success. Insuch jobs in which success is unusual, performance dis-tributions often are skewed. In this situation, becausea few exemplars can contribute so much to collectivesuccess, their stellar performances may overcome thecosts of many unproductive beginners. High successrates limit contributions to collective success (Fig. 1b),but if important workflows pass through such positions,a poor performance can hurt the collectivity (Fig. 1c).Yet in positions in which success is infrequent, a closerelationship between superlative individual accomplish-ments and organizational performance is likely (Fig. 1a)if incumbents control critical tasks.

1.2.3. IndivisibilitiesWe can add to Jacobs (1981) by noting that organi-

zations often try to divide important tasks among manyworkers to reduce their dependence on a few (Perrow,1986). Workers resist so the division of tasks often resultsfrom political struggles (Braverman, 1974; Sabel, 1982).If tasks are finely divided, the best efforts of the mostable cannot have much effect on the organization evenif the entire task is vital to the collectivity. Yet if tasksare indivisible, so each incumbent must handle a large
fraction of a critical job, the capacity of an individual toadd to organizational success in such positions will bemuch greater. The best retail salesperson can sell only atiny fraction of a large outlet’s total sales. Yet if a costly
by their partners. Compared to pooled or sequential interdependence,reciprocal interdependence therefore does not have clear effects onrelationships between individual and organizational performance.

ial Stra
good is indivisible, such as a ship or a super computer,a superb salesperson can make a significant contributionto firm revenues.

1.2.4. Subunit effectsIf the effects of individuals on an organization must

go through subunits, the relationship between subunitand organizational performance becomes relevant. Con-ditions similar to those shaping the relationships betweenindividual and organizational success also affect thisrelationship. A subunit may be sequentially related toother subunits, so people in such units will be unable toadd much to organizational success, but a weak perfor-mance by a strategically placed incumbent may weakenthe performance of the entire organization (see Fig. 1c).Yet if subunits are in a pooled arrangement so theirperformances are independent, the effects of a poor indi-vidual performance often will be less damaging to theorganization, although a weak individual performancemay reduce the success of that unit.

1.2.5. Credit asymmetriesOrganizations may find that their successes are

immeasurable, but failure can be costly. Many nonprofitscannot receive much credit for their accomplishments,but mistakes or malfeasance can lead to devastatingpublicity. Little can be known about the triumphs ofprisons, mental institutions, or welfare agencies, yetscandal is an ever-present threat. Although they cando much harm, people who work in such organiza-tions will find that it is difficult to make positivecontributions, so their jobs are goal destructive. Hence,the five factors that shape the relationship betweenindividual and organizational performance are: (1)whether position independence are pooled or sequen-tial, (2) whether individual success is unusual, (3)
whether tasks are indivisible, (4) whether the relation-ship between subunit and organization performance isgoal destructive or largely irrelevant, and (5) whethercollective success is immeasurable.7 These factors are
7 Workers often are embedded in interdependent teams so individualperformances are difficult to assess. This fact is completely consistentwith the arguments that will be made about goal destructive and largelyirrelevant positions. But measurement problems are not nearly as severewhen interdependence is pooled and incumbents act independently, soorganizations can compare individual performances. Past contributionsby individuals often are at least roughly measurable in goal enhancingjobs because such positions tend to be in a pooled relationship. Noteas well that the incumbents of goal enhancing must not be rewardedfor short run behavior that jeopardizes organizational goals. Hence,automobile dealers now retain salespersons partly on the basis of cus-tomer ratings in order to reduce short run destructive behavior such as


not examples of technological determinism becausetechnologies and organizational arrangements rarely arechosen due to inescapable requirements for efficiency(Braverman, 1974; Sabel, 1982). Technological choicesalmost always can be handled by various organizationalarrangements, but once either technologies or organiza-tional arrangements are selected, they typically persistwith important repercussions (Mahoney, 2000). No mat-ter what their first causes may have been, these lastingeffects should not be ignored.

2. Careers and labor markets

Studies of hiring and promotions show that individualperformance is almost impossible to predict. Test scoresor success in training programs are poor indicators (Berg,1970; Guion, 1981). Employers use unrefined indicatorssuch as education or race as a coarse filter because bettermethods do not exist (Doeringer & Piore, 1971; Spence,1974). Often the only indicator that approaches evenweak predictive power is past performance in an equiv-alent job. These uncertainties help explain careers, labormarkets, and hiring networks, but these effects dependon the relationship between individual performance andorganizational success. We focus on how labor marketsare used to organize employment and hiring networks,yet we first must revisit and sometimes refine the discus-sion of individual careers in Jacobs (1981) because thislogic provides a basis for new accounts for labor marketprocesses.

2.1. The three relationships, careers, and labormarkets

2.1.1. Goal destructive positionsIn jobs in which a poor performance can do much

harm but a superlative performance will not matter, thepotential costs of promoting an unreliable candidate will
not be trivial. Mistakes can be costly yet detailed knowl-edge about an individual’s daily behavior is the onlytrustworthy indicator of subsequent conduct. One wayto control the potential for harm in goal destructive posi-
lying to buyers or ignoring their requests for help after the transactionis completed. When sales are based on credit backed by the organi-zation, customers who are poor credit risks threaten firm profitability.To reduce the risks of defaults, organizations allow the transaction tobe completed only after credit worthiness is evaluated by actors whoare not supervised by sales managers. Because failures in these con-trol positions will hurt the organization, such jobs are modestly goaldestructive unless the loan amounts in question are high enough tobankrupt the credit carrying organization.

ial Stra

thttwppitfocsawTfwlpca

ouwmepimtw

motSmlhiarcmbtttsch


ions is to minimize risk by promoting only after one hasad substantial experience with a candidate. Recall thathe rate of success in such jobs often will be substan-ial so errors by even the least dependable incumbentsill be infrequent. Organizations with goal destructiveositions often use long probationary periods before theyromote to insure that unreliable people will not be putn positions that will let them do even greater harm tohe collectivity. When errors are costly but a stellar per-ormance does not add much, lateral moves betweenrganizations will be unlikely for the same reasons. Ifritical goal destructive positions are filled from the out-ide, supervisors will have no experience with new hiresnd little information about their dependability, so itill be easy to fail to reject an error prone candidate.herefore (ceteris paribus), when a poor individual per-

ormance can hurt the collectivity but an exemplary oneill not matter, careers in such goal destructive jobs are

ikely to begin at low levels and progress slowly. Airlineilots, police officers, or employees in capital intensiveontinuous process industries like chemicals and steelre good examples.

We can provide a new insight by noting that whenrganizations employ promotion screens to eliminatenreliable individuals before promoting them to jobshere they can do greater harm will organize employ-ent with internal labor markets (Baron, 1984).8 The

xtended probationary periods used to filter out errorrone employees who are mistakenly hired or promotednto goal destructive positions require an internal labor

arket with its internal promotion ladders. This adap-ation with additional features such as pay scales abovehat new hires would accept, training, and long-term

8 Administrative choices about the use of an internal or spot laborarket are partially determined by the organizational positions spelled

ut in our typology, but these choices also depend on other impor-ant considerations such as the resources available to the organization.mall firms facing severe competition cannot afford internal laborarkets, but large firms in uncompetitive industries often use spot

abor markets (Zucker and Rosenstein, 1981). Organizational factorselp determine when such core firms use this arrangement. If thenternal division of labor within subunits is limited because positionsre grouped by function (so multiple actors perform the same tasks)ather than by market (where different interdependent specialists areontained in the same unit; see Mintzberg, 1979 for a discussion),ost workers in a unit will be doing equivalent jobs so they will

e more interchangeable. Functional grouping reduces the costs ofhe absenteeism in spot labor markets. The pooled interdependencehat comes with functional grouping also limits the harm such poorlyreated employees can do to a collectivity. As many workers performimilar tasks when positions are grouped by function, performanceomparisons and piece rates are facilitated, and these arrangementselp control the poor work practices found in spot labor markets.


employment also is prudent as employee good will is cru-cial in jobs where such considerable harm is possible. Ifemployers do not treat these employees well, even elab-orate control systems frequently will be insufficient toprevent costly errors. While there are many other reasonsto use these arrangements, the screening requirementsimposed by goal destructive jobs make it extremelylikely that organizations will have to use internal labormarkets to organize employment such positions.

2.1.2. Goal enhancing positionsWhen the constraints are reversed so an exemplary

individual performance matters a great deal to an orga-nization but a poor one does not, careers differ. Theability to predict success is limited, yet a weak perfor-mance will not be costly in these jobs. Organizationswith goal enhancing positions react by employing manywhile keeping only the best. To increase the probabilityof finding exemplars, administrators give multiple can-didates a chance, but they try to eliminate all but the fewmost stellar performers. Therefore (ceteris paribus), tothe extent that a strong individual performance can helpcollective efforts but a poor one will be relatively cost-less, organizations often will hire many but keep only themost productive (Jacobs, 1981). Academic researchersin status seeking universities, people who sell expen-sive goods, book editors (Powell, 1985), and popularmusic producers (Peterson & Berger, 1971) are goodillustrations.

As organizations with goal enhancing jobs must con-stantly search for talent but poor performances are notcostly, allowing lateral moves between employers willincrease the probability of finding late bloomers. Theresulting frequent transfers between employers leadsto another refinement overlooked in Jacobs (1981).Althauser and Kalleberg (1981) suggest that some jobsare organized by an occupational internal labor marketwhere vertical mobility frequently occurs due to indi-vidual mobility between employers. Recall that goalenhancing positions often are in a pooled relationship. Inthis arrangement individual performances can be easilycompared because positions are so autonomous. Suchrelatively effortless comparisons and the reduced needsfor mutual adjustments between these autonomous posi-tions often means that a few can supervise the work ofmany. The wide spans of control that result place limitson internal promotion possibilities as few supervisoryopenings exist. If jobs are goal enhancing, occupational
internal labor markets are a common adaptation becauseso much vertical mobility in these jobs occurs by movesbetween organizations. Itinerant academics and salespersons are good illustrations.

ial Stra
2.1.3. Largely irrelevant positionsWhen individual performance differentials have lit-

tle influence on collective success, choosing the mostsuitable employees will not be critical. In such jobsthe inept can be hired and protected (Goode, 1967)because incumbent performances will not matter to theorganization. As the costs of hiring less talented ormore unreliable individuals will be negligible (ceterisparibus), decisions about hiring and careers in these jobscan be based on particularism. One common (but notthe only) manifestation of this particularism involvesdecisions based on loyalty. Organizational life can befiercely political. Loyalty therefore should be an impor-tant criterion because internal political struggles makeallegiance so valuable. When individual performancedoes not matter, gaining fealty from one’s subordinatesby deliberately hiring or promoting dependent incompe-tents will have little effect on collective success. Recallthat largely irrelevant positions often are found in sub-units whose performance also is largely irrelevant to thecollectivity. When subunit performances also are largelyirrelevant, the internal allocation of resources amongunits typically will be settled politically, and this makesloyal subordinates even more valuable. In organizationsthat use internal labor markets where collective successis largely unaffected by individual performance differ-ences, hiring and careers often will be based on politicalconsiderations.

But there are other options. We can add anotherinsight overlooked in Jacobs (1981) by noting that ifindividual performances are largely irrelevant, spot(or secondary) labor markets may be most suitable forthe organization. In contrast to internal labor markets,pay in spot labor markets is set by what unemployedjob seekers would accept, so wages are reduced.Promotions do not exist and training is unusual as moreproficient but unprotected employees have no reasonto sharpen the skills of their potential replacements.The result is an inexpensive and unreliable labor force
(Dickens & Lang, 1985). Yet spot labor markets maybe acceptable as individual performance in these jobsdoes not matter much.9 To control wage costs, it may be
9 Organizations avoid transfers between position types. Employeesin goal destructive jobs in which a poor performance can do greatharm must be risk averse. They are graded on their mistakes and ruleviolations, so it does not pay to innovate. In goal enhancing jobs asuperb individual performance will help but mistakes do not matter, soorganizations encourage innovation. These opposite adaptations meanthat habits learned in one position must be unlearned to do well thenext. Sometimes after long careers people in goal enhancing jobs whono longer wish to meet constant demands for production will become


best to tolerate all the inefficiencies that come with spotlabor markets.10 But if positions let incumbents eitheradd to or subtract from collective efforts, spot labormarkets with their low pay and poorly treated employeeswill not be prudent. Clearly other factors such as anorganization’s ability to pay the higher wages associatedwith internal labor markets also contribute to choicesabout these arrangements. The presence of largelyirrelevant positions therefore is a necessary but not asufficient condition that contributes to administrativedecisions about spot or internal labor markets.

The force of all of these observations may beweakened in larger organizations where it will be moredifficult for an individual to have as much impact oncollective performance. Such organizations generallyare divided into subunits so individual performancesmay add to or subtract from the success of these units. Inthis instance emphasis should shift to subunit success.We now can use this framework to develop a screeningtheory of hiring that captures the dual roles of jobcandidates’ and employers’ networks. Our analysisbuilds on research suggesting that the influence ofnetwork ties in the labor market depends on the workbeing done (Burt, 1997).

3. Getting a job: strong or weak ties?

This framework can provide additional insights aboutthe networks used to fill positions. Granovetter’s (1973)insightful discussion about the strength of weak tiesis primarily concerned with job seeking behavior, so itonly explains one side of the labor market (Montgomery,1992). We expand on Marsden and Campbell (1990)by focusing on the needs employers have to screencandidates and by considering the networks employedby both job seekers and employers. From Granovetter,we know that job seekers often use intermediaries, andfrom Marsden and Campbell we know that employersdo so as well. Organizations often rely on current
employees as informal brokers (Fernandez, Castilla, &Moore, 2000; Kalleberg, Knoke, Marsden, & Spaeth,1996; Marsden, 1994). By developing a frameworkabout these factors, we hope to add new insights about
administrators, but transfers between such dissimilar positions are notcommon.10 The operation of expensive but vulnerable equipment and the

sequential aspect of assembly lines lead to goal destructive jobs wherea few poor performances can be costly. When a capital intensive organi-zation like an automobile manufacturer can make slight improvementsin screening and the dependability of workers, costly equipment fail-ures can be substantially reduced (Horowitz and Sherman, 1980).

ial Stra

tt

3a

3

stssrattuda&L1owai

3

nmcmwCios

oecc(tuSaacsi


he network ties that are best for job seekers and thosehat are best for employers.

.1. Some advantages and disadvantages of weaknd strong ties

.1.1. An advantage of weak tiesGranovetter (1973) argues that weak ties are useful for

ecuring information.11 He reasons that if A is stronglyied to B and B is strongly tied to C, then A ought to betrongly tied to C, and the information A gets from Bhould be similar to the information A gets from C. Thisedundancy will not be present when ties are weak, soctors linked by weak ties will have more informationhan actors with the same number of strong but redundanties. Information about suitable openings is a scarce andseful resource, so this logic should explain how candi-ates get jobs under some conditions. While this initialrgument has received some empirical support (Podolny

Baron, 1997), it can be expanded (Bian & Ang, 1997;in et al., 1981a,b; Montgomery, 1992; Murray et al.,981; Wegener, 1991). We begin by offering some basicbservations about the comparative value of strong andeak ties under various conditions, and then we provideframework for understanding how these ties are used

n labor markets.

.1.2. A disadvantage of weak tiesIf no attempt is made to conceal information, Gra-

ovetter’s reasoning is apposite. Knowledge should beore quickly disbursed to more actors when they are

onnected by weak ties. Yet when the complete truthay result in costs to parties on one side of an exchange,eak ties should diminish the accuracy of information.
andidates for jobs have reasons to divulge only what
s consistent with their interests. If a parties on one sidef an exchange face incentives to avoid being forthrighto useful information is distributed asymmetrically, cor-

11 Granovetter (1973, p. 1361) furnishes a definition. “The strengthf a tie is a (probably linear) combination of the amount of time, themotional intensity, the intimacy (mutual confiding), and the recipro-al services which characterize a tie.” Granovetter (1974) uses a thirdategory called formal methods, which includes ways to find a jobadvertisements or employment agencies) that do not involve personalies. We ignore long weak ties because he finds they are almost neversed. The method used can be decided by the preponderant effect.uppose a close relative tells a job seeker about an opening, but allpplicants must take a test. If most applicants pass the test, but thepplicant with the tie still gets the job due to the relative, the mainause was the tie. If the job seeker with the tie gets the job because/he earned the best score on the test, the tie would have been lessmportant and the job would have been obtained by formal methods.


rections will be less likely in weak tie networks. Actorslinked in this manner will have interactions that are lessintense. When only weak ties are present, claims cannotbe readily verified. This characteristic of weak ties makesthem less attractive to employers particularly when somevacancies must be filled.

3.1.3. An advantage of strong tiesThe opposite is the case when ties are strong. Due to

the redundancies Granovetter stressed, strong ties pro-vide access to intimates who often will have superiorinformation about a candidate. If some intimates aremost loyal to the organization because they work for it,these intermediaries can be used to assess the accuracyof a job seeker’s claims. By definition, people connectedby strong ties have intense interactions of longer dura-tion. As deception will be difficult when ties are strong,such ties can be used to enhance an organization’s infor-mation about candidates. Weak ties therefore should bemost useful when the dissemination of information aboutvacancies is important. Yet, when the accuracy of infor-mation about candidates is most critical, strong tie hiringshould be used. As long as individual performanceshave different effects on organizational success, theseconsiderations will not be equally important. By consid-ering both applicant and employer motivations, we canidentify optimal linkages between job seekers and inter-mediaries as well as those between intermediaries andemployers. We begin by revisiting the weak tie argument.

3.2. Ties from different networks

The crux of Granovetter’s argument is that strong tiesbetween A and B and between B and C lead to at least aweak tie between A and C. But for this to be true we mustassume that the ties in question are all the same type. Yet,when thinking about how jobs are filled, we must con-sider the simultaneous effects of different kinds of ties.This is the case, in part, because the nodes at issue oftenare different entities. The job seeker is a person, but theintermediary could be an agency or a current employee,while the employer typically is an organization. Tiesbetween the intermediary and the employer are based onsuch things as prior success in recruitment. If the inter-mediary is a current employee, the quality of the employ-ment relationship should contribute to the strength of thetie. Yet, the strength of ties between the job seekers andintermediary are determined by such things as the dura-
tion and intensity of their relationship. In labor markets,the “glue” that binds these various entities together is notthe same. These observations are important because theycall into question the applicability of the information

48 D. Jacobs, B. Cornwell / Research in Social Stratification and Mobility 25 (2007) 39–55

elationsn the em

strong reasons to help fill vacancies in their work groupwith dependable loyal new colleagues who will be silentabout errors or infractions. No one wants an unreliable

12 What is at issue is a beginner’s reliability. Bad news travels well, sothe elimination of an experienced candidate who already has committeda major error in a goal destructive job often will not be problematic.It will be much more difficult to predict which beginners will be most

Fig. 3. Two hypothetical co-existing networks; friends and employee rlife. The actors in the friendship network are the same as the actors ioperate simultaneously.

redundancy argument in labor markets when the actorsdiffer. Redundancy should be important when the tiesunder consideration are of the same type. In a networkof friendships, for instance, a link from A to C throughlinks between A and B and between B and C makes sensebecause friendships are built on similar activities, but thisis not the case for employee-employer networks.

Essentially, we are talking about the existence ofseparate networks that operate simultaneously. Fig. 3illustrates what these two networks might look like.Weak ties are represented by dashed lines between actorsand strong ties are solid. The networks are layeredto show that they operate simultaneously but are ondifferent planes because they are relevant to differentrealms. While the intermediary engages the job seekerin one network, for instance, the intermediary engagesthe employer in a different network. Actors have accessto both networks, but need not be connected to the sameactors in both. As a result, these links need not consti-tute a redundancy. This idea of a multi-layered ecologyof networks will let us make hypotheses about strongties existing between A and B and between B and Cwithout also assuming the existence of a tie between Aand C. By allowing triads to be made up of differentkinds of social relationships, we can relax Granovet-ter’s “forbidden triad” redundancy argument and developpropositions about when strong ties might be preferred.

3.3. When will strong or weak ties be useful?Networks and the three relationships

3.3.1. Goal destructive positionsWhen a poor individual performance can be costly but

an exemplary one will not matter, unreliable employees

. Note: The shaded parallelograms represent different realms of socialployee relations network. The arrows signify that the two networks

who are likely to make costly errors must be avoided.Yet information about reliability usually will be inade-quate. Applicants have reasons to be less than candid andeven the most mistake-prone incumbent will make infre-quent errors. In this situation strong ties will be far moreeffective than their weak counterparts because such tiesreduce uncertainties about applicant reliability. Admin-istrators who must fill goal destructive positions will pre-fer direct strong ties to suitable candidates without inter-mediaries, but their number strong ties to suitable appli-cants often will be modest. In the absence of many directstrong ties with applicants, we can expect that (ceterisparibus), administrators will attempt to recruit throughstrong tie networks between current employees and can-didates when goal destructive jobs must be filled.12

Clearly this proposition will have the most force whenintermediary and employer interests overlap. Yet currentemployees in goal destructive jobs and their supervisorshave similar if not identical interests in the character ofnew hires. Because they can make such costly errors,employees in goal destructive jobs are evaluated (andpunished) for mistakes and rule violations. They have

reliable, and here strong ties should be invaluable. The candidate witha history of minor mistakes or rule violations will be a poor risk in agoal destructive position, yet only informants employed by and loyalto the organization who are connected to candidates by strong ties willbe a source of accurate information about these attributes.

ial Stra

ctgessaEriwhtc(petjn

fivatrj(rap(eofbItsp

tmpni(wstst

employment to many, they keep only the best. If a weakperformer is mistakenly selected for a goal-enhancingjob, this error can be corrected with minimal cost tothe organization. Organizations therefore should cast a

13 Organizations under these constraints may have to substitute oneremedy for another. Administrators who can use long probationaryperiods to gauge the reliability of candidates for goal destructive jobsshould not have as great a need to hire through strong ties. An absenceof strong ties for any reason will be less costly if an organization can


oworker who can jeopardize their safety or their reputa-ion, but these considerations will be absolutely crucial inoal destructive jobs. The best solution for both employ-rs and the currently employed intermediaries who areo often used to help fill goal destructive jobs involvestrong ties between these intermediaries and candidatess such ties engender accurate information and trust.ncouraging employees in goal destructive jobs to

ecruit intimates also enhances productivity by facilitat-ng informal training and it provides a perk to employeeshose good will is crucial because they can do so mucharm. While employed intermediaries may overvaluehe reliability of kin and close friends, other importantonsiderations outlined by Kahneman and Tversky1979) produce risk averse behavior. Such offsettingsychological inclinations should not be sufficient toliminate the powerful and largely overlapping organiza-ional and employee inducements to fill goal destructiveobs with the most reliable applicants via strong tieetworks.

Empirical results are supportive. Granovetter (1974)nds that successful candidates often learn aboutacancies from current employees, while Kirschenmannd Neckerman (1991) report that many firms giveheir employees bonuses for successful referrals. Closeelatives often work together in goal destructive railroadobs (Gamst, 1980) and in such jobs in deep sea fishingNorr & Norr, 1978). These workers give dependableelatives and friends accurate advice about how topply for work-group openings due to the knowledgerovided by these close ties about who can be trustedfor similar claims see Fernandez et al., 2000; Kallebergt al., 1996; Marsden, 1994). While there are manyther reasons for wanting one’s relatives and closeriends to be hired, these motives are strengthenedy the incentives imposed by goal destructive jobs.nducements operating at both the administrative andhe work-group level combine to help applicants withtrong ties to current employees get goal destructiveositions.

Administrators may use non-employees with strongies to candidates as intermediaries, but this arrange-

ent involves risks due to the uncertainties created byotentially conflicting loyalties. Intermediaries who areot employed by the organization may place candidatenterests above those of organization. Yet both Duncan1984) and Anderson (1990) give examples of sponsorsho, while not employed by the hiring organization,

trive first to maintain their credibility with administra-ors. Although they have strong ties to the applicants theyponsor, these intermediaries are most likely to disregardhese loyalties and to be candid about candidate flaws.


Organizations seeking to fill jobs where the potential forharm is so great have strong reasons to pay close atten-tion to the accuracy of a non-employed sponsor’s priorrecommendations. While the literature suggests that thisform of strong tie recruitment is common, it probablyis not as widespread as hiring based on strong ties tocurrent employees due to the greater correspondence inthe interests of employees and their employer when goaldestructive jobs are filled.

An exception to all of these hypotheses about whatorganizations do to avoid unreliable employees in goaldestructive jobs is present when labor markets are tight.If applicants are scarce, it may not be possible to useeither long probationary periods or strong ties to screenthe few available candidates. When the railroads firstfilled their many goal destructive jobs, they faced tightlabor markets. These shortages interfered with effortsto use screening procedures to eliminate unreliable newhires, so costly accidents increased (Kingsford, 1970).When labor markets are tight, organizations must useother methods such as hiring through employment agen-cies or by advertisements. But since less informationabout candidates will be available, weak ties or substi-tutes (agencies, advertisements, or walk-in hiring) onlywill be used as a last resort when organizations fill goaldestructive vacancies.13

3.3.2. Goal enhancing positionsWhen stellar employee performances can help but

serious harm is unlikely, information about applicantswill be less critical. While additional information aboutapplicants provided by strong ties always is better, thespread of information about these vacancies will be moreimportant than information about candidates. Recallthat the continued employment of new entrants in goalenhancing jobs is based on their performance duringtrial periods. Although organizations give probationary

initially place new employees in jobs where they can do less harmwhile they are carefully evaluated for promotions into goal destructivepositions. Piloting commercial airlines is a good example because firstofficer candidates typically must spend years as closely supervisedco-pilots.

ial Stra
net as widely as possible to maximize the probability offinding the rare individuals who are capable of exem-plary performances. The far more numerous weak tiesspread information better than strong ties, so (ceterisparibus), weak ties should be used to fill goal-enhancingpositions. For example, jobs in sports and entertainmentare typically filled by weak ties or by formal meth-ods such as advertisements or employment agencies(Blalock, 1967). While administrators always will prefermore information to less, this preference should not over-whelm the powerful incentives to use the more extensiveweak ties to locate the few stellar performers who willbe so useful in these jobs.

3.3.3. Irrelevant positionsWhen vacancies exist in largely irrelevant positions,

both job finding methods will be used. Remember thatwhen individual performance differentials do not mat-ter much and these jobs are in an internal labor market,particularism will be more important than locating themost capable or the least error prone candidates. Asindividual performances at best have modest effects, inthese jobs we can expect that candidates with close tiesto supervisors or to current employees will be hiredfirst. Particularism, of course, is facilitated by strongties between intimates or close friends. It follows that(ceteris paribus), when candidates seek largely irrele-vant jobs in internal labor markets, they also should bemore likely to get positions via strong ties to currentemployees.

But strong tie hiring will not be as likely in orga-nizations dominated by largely irrelevant positions thatuse spot labor markets. When employment is orga-nized by a spot labor market, promotions are rare andpay is low. The meager organizational rewards avail-able to employees in spot labor markets therefore willnot be worth fighting for. Hence (ceteris paribus), tothe extent that organizational positions are largely irrel-evant and employment is governed by a spot labormarket, neither loyalty nor employee differences shouldmatter, and weak ties or alternative methods such asadvertisements and walk-ins should be the most com-mon way to find such jobs. Research by Bian (1997)supports this proposition. In short, the interdependentmatches between the kind of employee the organiza-tion wishes to recruit, the kind of labor market andthe characteristics of the positions organizational hiringagents seek to fill is endogenous. Position characteris-
tics, labor markets, and the employee characteristics thatare most desirable interact reciprocally to create opti-mal arrangements in these three mutually interdependentoutcomes.

4. Competition, legal mandates, and raciallyhomogeneous hiring

The prior observations concerning the types ofnetwork ties used by employers also help explain whenracial and ethnic minorities will be least likely to behired. Economists argue that competition will reduceor eliminate discrimination because organizations thatfavor inferior non-minority over superior minority can-didates will have a less productive work force. Yet thisrelationship between competition and reduced discrim-ination is atypical because unusual job characteristicsdetermine when this argument will be accurate. Bytaking organizational positions into account, we showhow network screening methods affect minority hiring.As with our other propositions, the usual ceteris paribuscaveat holds. Ethnic groups, for example, may try torestrict access to jobs, and some employers excludeminorities based on discriminatory tastes, but spaceconstraints mean we cannot discuss these alternativeexplanations.

4.1. Goal destructive positions and minorityemployment

As we have seen, when an individual’s performanceonly can hurt the collectivity, organizations will try toeliminate error prone candidates by recruiting throughstrong ties. Yet such ties rarely go across racial orethnic boundaries. Whites may have strong ties withother whites but their relationships with people in othergroups are more distant (Feld & Carter, 1998; Korte& Milgram, 1970). As potentially negative informa-tion about applicants and trust is so vital, once goaldestructive jobs are filled by one racial or ethnic group,the undisrupted operation of competitive markets willnot lead to an increased tendency to recruit applicantsfrom another group due to the networks used to fillthese jobs (for examples see Tilly, 1998 on opportunityhoarding). Without interference, recruitment processesfor goal destructive positions therefore should continueto be intra-ethnic and intra-racial. While hiring diversityfaces many other obstacles, these handicaps are partic-ularly severe when organizations fill goal destructivevacancies.

Underground mining is a good illustration. Mistakescan be terribly costly to both the organization andworkers, so these dangerous jobs are among the best
examples of goal destructive positions. Accordingto Boswell (1986), extreme ethnic homogeneity waspresent because neophytes typically were recruitedby working miners (Stinchcombe, 1990). Within any

ial Stra

sptjcTgdswrsfi

tSCtatsdlltctidrgpt

4

obmnprlpstpstct

does not mean that the normal operations of unfetteredfree markets will eliminate labor market discriminationagainst minorities. Despite assumptions to the contraryby economists who erroneously view organizations as

14 This network theory does not require minorities to be inferior tonon-minorities. Explanations involving statistical discrimination reston the assumption that while some minorities may be superior tosome non-minorities, the minority mean is lower than the non-minoritymean. According to this network theory, the performance of minori-ties could be equal or even superior to non-minorities and minoritieswould still be unduly handicapped by their inferior connections whenthey seek most jobs.15 As is true of most similar analyses, this approach only partially


ingle mine, only one ethnic group held the higheraying below ground jobs in part because strongie employment networks for these goal destructiveobs did not go across ethnic groups. Yet the ethnicomposition above ground jobs was more mixed.he improbable degree of ethnic and racial homo-eneity in the goal destructive jobs in police and fireepartments – before the courts became involved –uggests these agencies also used selection systems thatere less than completely open despite civil service

egulations. Ethnically homogeneous hiring thereforehould persist when goal destructive vacancies arelled.

Statistical findings provide additional support forhese claims (McPherson & Smith-Lovin, 1987;now, Zurcher, & Ekland-Olson, 1980). Marsden andampbell (1990) suggest that informal recruitment

hrough current employees makes diverse hiring unlikelys these intermediaries tend to disseminate informa-ion to socially similar persons. This condition cannowball by producing increased group similaritiesue to what McPherson, Popielarz, & Drobnic (1992)abel a directional selection process that leads to aess heterogeneous labor force. “[Weak ties] act ashe harbinger of social change in groups. [Such ties]ast more widely through the social world than strongies. Without weak ties, groups would be static, lockednto place by the local density of strong, redun-ant, closed social networks.” (p. 166). As long asecruitment through strong ties is more likely whenoal destructive vacancies are filled, competitive hiringrocesses will not reduce employment discrimina-ion.

.2. Irrelevant positions and minority employment

When individual performance has only trivial effectsn organizational success, productivity differentialsetween individuals or differences in their propensity toake errors will not be critical. In this situation it will

ot be costly to overlook talented minorities and fill suchositions with less productive members of a majorityacial or ethnic group. If organizations dominated byargely irrelevant positions use an internal labor market,articularistic considerations often will dominate, butuch particulistic hiring typically will go through strongies between intimates. These conditions thereforeroduce ethnically homogeneous hiring because the
trong ties used to fill these jobs again are unlikelyo reach across racial or ethnic boundaries. Theseonsiderations suggest that minorities are less likelyo have the connections to get largely irrelevant jobs

when these positions are governed by an internal labormarket.14

When organizations with largely irrelevant positionsuse a spot rather than an internal labor market, strongties will not be as important, so their contacts shouldnot hurt minorities as much. Yet minority successessecuring such positions hardly are an equal employmenttriumph. If qualified ethic or racial minorities arelikely to be confined to low paying spot labor marketpositions where advancement and training are nil, thismeans that minorities still are experiencing unequalemployment outcomes.15 It probably does not matterif largely irrelevant positions within organizations aregoverned by an internal or a spot labor market. Ineither case, the interaction of organizational and labormarket processes is likely to produce inferior minorityoutcomes, and interfirm competition is unlikely to erodethis unfortunate result.

4.3. Goal enhancing positions and minorityemployment

Blalock (1967) argues that minorities first succeed injobs in which strong individual performances are mostimportant to organizations. The resulting competition fortalent forces organizations to hire superior minorities.Exemplary performances matter a great deal to organi-zations involved in entertainment, professional sports,or sales, so these jobs provided the first upward mobilitypaths for various minorities (Blalock, 1967). Yet thesepositions are not common. While occupational discrimi-nation will be least likely under these circumstances, that

explains inferior labor market outcomes for women by providinginsights about when spot labor markets will be used. Dickens andLang (1987) and Coverman (1986) find qualified women are morelikely to be confined to spot labor markets, so the model explains someconditions when women will do worse than men.

52 D. Jacobs, B. Cornwell / Research in Social Stratification and Mobility 25 (2007) 39–55

es of tie
Fig. 4. Relationship between position type and typ
undifferentiated individuals, goal-enhancing positionsare too unusual.16

4.3.1. Competitive effectsIf organizational environments are competitive, the

predictive power of these propositions about goaldestructive and goal enhancing positions ought to beenhanced. Smaller differences in individual perfor-mances should matter more when competition is intense.Organizations in competitive environments thereforeshould be less likely to exclude talented minoritiesfrom goal enhancing positions. Yet intense competitionbetween organizations should make racially homoge-neous hiring even more likely when organizations fillgoal destructive jobs. In a competitive environment anorganization’s survival can hang by a thread becauseerrors in goal destructive jobs may give competitors irre-
versible advantages (Hirschman, 1971). Hence, Murrayet al. (1981) find that strong rather than weak ties aremore important when demand for labor is weak. Whilefierce competition may enhance a superior minority indi-
16 Jobs where an individual may help or hurt have not been discussed.Our focus has been on positions that limit the effects of strong orweak performances because these pure cases have such clear effects,but this means the analysis is not exhaustive. Jobs where individualperformances can both add or subtract from collective success maylead to intermediate outcomes, although an ability to do harm usuallywill be given greater weight. Important actors see executives as havingthe capacity to help or hurt their organizations. Jackall (1988) andGranovetter and Tilly (1987) suggest they are recruited with strongties, in part because it is best to be risk averse. When an organizationis performing adequately, rather than using weak ties and casting anet widely to find the most innovative leader, recruiters instead usestrong ties to find someone of good character who is dependable. Whenincumbents in any organizational position can both help or hurt, as longas the unit in question is reasonably successful, organizations shouldsatisfice by seeking candidates with the same risk averse, strong tiesused to fill goal destructive jobs. Of course, political considerationssuch as trust and loyalty and the power executives have over othersalso encourage the use of strong ties. And because moral characteris so important in executive positions, strong ties will be more usefulbecause they provide information about this crucial consideration.

s that are beneficial to employers and job seekers.

vidual’s likelihood of getting an unusual goal enhancingposition, competition gives administrators stronger rea-sons to use strong ties to find out all they can aboutcandidates when goal destructive jobs must be filled. Andminority candidates, of course, will typically be handi-capped when organizations hire through strong ties.

This argument about competitive effects stressed byeconomists can be summarized by noting that even if weignore problems enforcing laws against discriminationand look only at the interaction between competi-tion, organizational processes, and hiring networks, ourscreening framework suggests that homogeneous hir-ing will be most persistent in goal destructive jobs, andthis effect will be magnified in competitive environ-ments. Equal hiring instead will be most likely whengoal-enhancing jobs are filled, but these positions unfor-tunately are not common.

5. Summary

These arguments about the usefulness of strong ver-sus weak ties are summarized in Fig. 4. Recall that ourframework hinges on two attributes: (1) position charac-teristics, and (2) the existence of multiple networks. Mostnetwork-oriented labor market research emphasizes theimplications of weak ties for the quantity of informationreceived by job seekers. By considering the relationshipsbetween individual and organizational success, we showwhen employers will have incentives to use strong tiesto recruit employees. This analysis also uncovers theresidual benefits for job seekers who make contact withemployers via strong ties in these situations. The para-dox of labor market information flow concerning goaldestructive jobs is that while weak ties are beneficialto job seekers due to their ability to provide increasedinformation, strong ties are beneficial to employers due
to their ability to generate information that is more accu-rate. And because employers have the final say aboutwho is hired, contacts with employers through strongties should enhance applicant success in goal destructive

ial Stra

pvtei

ost&Hr(tBtme(fiiltidrodbfi

spsrfiutbzsmo

atpnicd


ositions. A similar process guides employment in irrele-ant positions where particularism dominates. Weak tiesherefore are beneficial only to candidates who seek goal-nhancing jobs or those who seek largely irrelevant jobsn spot labor markets.

These conditional refinements may explain why testsf Granovetter’s seminal ideas have not always beenupportive. Results suggest that weak ties do not seemo help candidates get jobs with better pay (Bridges

Villemez, 1986; Graaf & Flap, 1988; Marsden &urlbert, 1988). While Lin et al. (1981a,b) find some

esults supporting Granovetter, other findings do notbut see Granovetter, 1983). Yet in all of this work lit-le or no attempt is made to consider position attributes.urt (1997) considers the contingent effect of network

ies when job seekers compete for positions sought byany others, but the only study that looks at a pure

xample of a one of the positions enumerated in Fig. 1Murray et al., 1981 focus on a goal enhancing job)nds results consistent with our arguments. If the util-

ty of strong or weak ties depends on how positionsimit the relationship between individual and organiza-ional performance, inconsistent results can be expectedf researchers analyze an indeterminate mix of these threeifferent organizational positions. If our analysis is cor-ect, researchers should be careful to take the kinds ofrganizational vacancies into account when they con-uct quantitative analyses. If their analyses instead areased on an uncertain sample of these different positions,ndings will be contradictory and ungeneralizable.

More generally, the novel features of this analy-is begin to provide insights about how organizationalrocesses affect labor markets and hiring. The firstection provided a foundation by presenting the threeelationships between individual and organizational per-ormance and their effects on careers and mobility, andt added fresh material on the kinds of labor marketssed to organize employment when these three posi-ions are present. The next section extended the theoryy providing new treatments about how this organi-ational framework accounts for hiring through eithertrong or weak ties. Finally, these insights about employ-ent networks were used to clarify the different effects

f competition on minority hiring.It is distressing that minority employment difficulties

ppear to rest on such influential and persistent explana-ory factors. Although this analysis may appear to beessimistic, it still may be valuable. We probably do
ot live in the best of all possible worlds where thenevitable effects of competitive markets will automati-ally eliminate racially or ethnically based employmentifferentials. The more important point is that attempts

to produce equal opportunity employment are unlikely tobe successful if we ignore or underestimate the strengthof these countervailing factors that make homogeneoushiring so persistent.

Acknowledgements

David Britt, Samuel Cohn, Robert Kaufman, RobertO’Brien, and Leo Rigsby and the members of the strati-fication work group at Ohio State helped by commentingon prior drafts. We also are indebted to Joan Acker andJoe Stone for their useful advice.

Appendix A

The optimal way to gauge the presence of the threeposition types enumerated in Fig. 1 and their effectsinvolves a combination of quasi-experimental meth-ods and nonlinear regression because, in this instance,behavioral measures will be superior to those based onperceptions (see Alchian, 1950 for reasons). These tech-niques are too complicated to explain in a brief appendix(but see Jacobs, 1981 for a longer attempt). The degreeto which a position fits one of the ideal types in Fig. 1 canbe more crudely measured by asking administrators howmuch individual performances in a particular organiza-tional position add to or subtract from the performanceof their unit. The resulting position classifications thencan be used to predict the outcomes stipulated by thehypotheses in this paper.

For example, positions uniformly characterized byadministrators as goal destructive could be assigned adummy variable code of one while other positions wouldbe scored zero. This dummy and additional explana-tory variables, such as how tight labor markets wereat the time when position incumbents were hired, canbe used in regression equations computed on a sampleof employees to predict the primary method they usedto get their jobs. Stinchcombe (1990) states that job-holders typically overestimate their importance to theirorganization, so the relationships between individual andorganizational performance should not be gauged by ask-ing employees how much they think their performancematters to their organization.

This test would involve several stages. First, admin-istrators would be asked to score positions on the degreeto which incumbents could hurt collective performancewith a poor individual performance, and how much
they can help collective performance. Second, using theresults from a survey of employees, the position desig-nations would be used to predict the predominant jobfinding method they used to get their positions.

ial Stra
If the administrative answer to both questions onwhether an individual performance can help or hurt col-lective performance uniformly is “relatively little”, theposition would be scored as largely irrelevant. If supervi-sors almost always see poor performances and mistakesas extremely dangerous and/or highly destructive to theirreputations, but they are unable to give plausible exam-ples when an exemplary individual performance addedsignificantly to collective achievements, the positionwould be scored as goal destructive and this dummyshould predict hiring with strong ties. Relatively uni-form answers that exemplary individual performancesin some positions increase the probability that collec-tive goals would be achieved, but poor performances donot matter much would lead a position to be scored asgoal enhancing. A score of one on this dummy should beindependently associated with finding jobs through weakties or with formal methods such as responses to helpwanted advertisements, or through employment agen-cies in comprehensive regression equations designedto predict the job finding methods most often used byemployees.

References

Alchian, A. A. (1950). Uncertainty, evolution, and economic theory.The Journal of Political Economy, 58, 211–221.

Althauser, R., & Kalleberg, A. (1981). Firms, occupations, and thestructure of labor markets: A conceptual analysis. In I. Berg (Ed.),Sociological perspectives on labor markets (pp. 119–149). NewYork: Academic.

Anderson, E. (1990). Street wise. Chicago: University of ChicagoPress.

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

Berg, I. (1970). Education and jobs. New York: Praeger.Bian, Y. (1997). Bringing strong ties back in: indirect ties, network

bridges, and job searches in China. American Sociological Review,62, 366–385.

Bian, Y., & Ang, S. (1997). Guanxi networks and job mobility in Chinaand Singapore. Social Forces, 75, 981–1005.

Blalock, H. M., Jr. (1967). Toward a theory of minority group relations.New York: Wiley.

Blalock, H. M., Jr. (1969). Theory construction. Englewood Cliffs, NJ:Prentice-Hall.

Boswell, T. (1986). A split labor market analysis of discriminationagainst Chinese immigrants, 1850–1882. American SociologicalReview, 51, 352–371.

Braddock, J. H., II, & McPartland, J. M. (1987). How minoritiescontinue to be excluded from equal employment opportunities:Research on labor market and institutional barriers. Journal of
Social Issues, 43, 5–39.
Braverman, H. (1974). Labor and monopoly capital. New York:Monthly Review Press.

Bridges, W., & Villemez, W. (1986). Informal hiring and income inthe labor market. American Sociological Review, 51, 574–582.


Burt, R. S. (1997). The contingent value of social capital. Administra-tive Science Quarterly, 42, 339–365.

Collins, R. (1979). The credential society. New York: Academic Press.Coverman, S. (1986). Occupational segmentation and sex differences

in earnings. In R. V. Robinson (Ed.), Research in social stratifica-tion and mobility (pp. 139–172). Greenwich: JAI Press.

de Graaf, N. D., & Flap, H. D. (1988). ‘With a little help from myfriends’: Social resources as an explanation of occupational statusand income in West Germany, The Netherlands, and the UnitedStates. Social Forces, 67, 452–472.

Dickens, W. T., & Lang, K. (1985). A test of dual labor market theory.American Economic Review, 75, 792–885.

Dickens, W. T., & Lang, K. (1987). Where have all the good jobs gone?Deindustrialization and labor market segmentation. In K. Lang & J.Leonard (Eds.), Unemployment and the structure of labor markets(pp. 90–102). Basil Blackwell.

Doeringer, P., & Piore, M. (1971). Internal labor markets and man-power analysis. Lexington: Heath.

Duncan, J. G. (1984). Years of poverty years of plenty. Ann Arbor:University of Michigan Press.

Elster, J. (1989). Nuts and bolts for the social sciences. Cambridge:Cambridge University Press.

Feld, S. L., & Carter, W. C. (1998). When desegregation reduces inter-racial contact: A class size paradox for weak ties. American Jorunalof Sociology, 104, 1165–1186.

Fernandez, R. M., Castilla, E. J., & Moore, P. (2000). Social capitalat work: Networks and employment at a phone center. AmericanJournal of Sociology, 105, 1288–1356.

Fernandez, R. M., & Weinberg, N. (1997). Sifting and sorting: Personalcontacts and hiring in a retail bank. American Sociological Review,62, 883–902.

Gamst, F. C. (1980). The hogshead: An industrial ethnology of thelocomotive engineer. New York: Holt/Rinehart/Winston.

Goode, W. J. (1967). The protection of the inept. American Sociolog-ical Review, 32, 5–19.

Goodman, P. S., & Pennings, J. M. (1977). New perspectives on orga-nizational effectiveness. San Francisco: Jossey-Bass.

Gordon, R. J. (1990). Macroeconomics. Harper Collins.Granovetter, M. (1973). The strength of weak ties. American Journal

of Sociology, 78, 1360–1381.Granovetter, M. (1974). Getting a job. Cambridge: Harvard.Granovetter, M. (1983). The strength of weak ties: A network theory

revisited. In R. Collins (Ed.), Sociological theory 1983. JosseyBass. Chapter 7.

Granovetter, M., & Tilly, C. (1987). Inequality and labor process. InS. Neil (Ed.), The handbook of sociology (pp. 175–221). NewburyPark, CA: Sage.

Guion, R. M. (1981). Choosing members for organizations. In P. C.Nystrom & W. H. Starbuck (Eds.), Handbook of organizationaldesign (pp. 358–381). New York: Oxford.

Hickson, D. C. R., Hinings, C. A., Lee, R. E. S., & Pennings, J.M. (1971). A strategic contingencies theory of intraorganizationalpower. Administrative Science Quarterly, 16, 216–229.

Hirschman, A. O. (1971). Exit, voice, and loyalty. Cambridge: Harvard.Holmstrom, R. B., & Tirole, J. (1989). The theory of the firm. In

R. Schmalensee & R. D. Willig (Eds.), Handbook of industrialorganization (p. 61182). New York: North-Holland.

Horowitz, S., & Sherman, A. (1980). A measure of the relation-ship between human capital and productivity. Journal of HumanResources, 15, 67–77.

Jackall, R. (1988). Moral mazes: The world of corporate managers.Oxford.

ial Stra

J

K

K

K

K

K

L

L

L

M

M

M

M

M

M

M

M


acobs, D. (1981). Towards a theory of mobility and behavior inorganizations: An inquiry into the consequences of relationshipsbetween individual performance and organizational success. Amer-ican Journal of Sociology, 87, 684–707.

ahneman, D., & Tversky, A. (1979). Prospect theory: An analysis ofdecision under risk. Econometrica, 47, 263–291.

alleberg, A. L., Knoke, D., Marsden, P. V., & Spaeth, J. L. (1996).Organizations in America: Analyzing their structures and humanresource practices. Thousand Oaks, CA: Sage Publications.

ingsford, P. W. (1970). Victorian railway men. London: Frank CassLtd..

irschenman, J., & Neckerman, K. M. (1991). ‘We’d love to hire them,but’. . .: The meaning of race for employers. In J. Christopher & P.E. Peterson (Eds.), The urban underclass (pp. 203–232). BrookingsInstitute.

orte, C., & Milgram, S. (1970). Acquaintance networks betweenracial groups: Application of the small world method. Journal ofPersonality and Social Psychology, 15, 101–108.

ilien, D. M., & Hall, R. E. (1986). Cyclical fluctuations in the labormarket. In A. Orley & R. Layard (Eds.), Handbook of labor eco-nomics (pp. 1001–1035). New York: North-Holland.

in, N., Walter, E., & Vaughn, J. (1981a). Social resources and thestrength of ties. American Sociological Review, 46, 393–405.

in, N., Walter, E., & Vaughn, J. (1981b). Social resources and occu-pational status attainment. Social Forces, 59, 1163–1181.

acy S M.W. (1991). Chains of cooperation: Threshold effects incollective action. American Sociological Review, 56, 730–747.

ahoney, J. (2000). Path dependence in historical sociology. Theoryand Society, 29, 507–548.

arsden, P. V. (1994). The hiring process: Recruitment methods. Amer-ican Behavioral Scientist, 37, 979–991.

arsden, P. V., & Campbell, K. E. (1990). Recruitment and selectionprocesses: The organizational side of job searches. In R. Breiger(Ed.), Social mobility and social structures (pp. 59–79). New York:Cambridge University Press.

arsden, P. V., & Hurlbert, J. S. (1988). Social resources and mobil-ity outcomes: A replication and extension. Social Forces, 66,1038–1059.

cPherson, J. M., Popielarz, P. A., & Drobnic, S. (1992). Socialnetworks and organizational dynamics. American SociologicalReview, 57, 153–170.

cPherson, J. M., & Smith-Lovin, L. (1987). Homophily in voluntaryorganizations: Status distance and the composition of face-to-facegroups. American Sociological Review, 52, 370–379.

intzberg, H. (1979). The structuring of organizations. NJ: PrenticeHall.


Montgomery, J. D. (1992). Job search and network composition:Implications of the strength-of-weak-ties hypothesis. AmericanSociological Review, 57, 586–596.

Murray, S., Rankin, J., & Magill, D. (1981). Strong ties and job infor-mation. Sociology of Work and Occupations, 9, 119–136.

Norr, J. L., & Norr, K. L. (1978). Work organization in modern fishing.Human Organization, 37, 163–171.

Parsons, D. O. (1986). The employment relationship: Job attachment,work effort, and the nature of contracts. In A. Orley & R. Layard(Eds.), Handbook of labor economics (pp. 789–848). New York:North-Holland.

Perrow, C. (1986). Complex organizations: A critical essay. RandomHouse.

Peterson, R. A., & Berger, D. G. (1971). Entrepreneurship in organi-zations: Evidence from the popular music industry. AdministrativeScience Quarterly, 16, 97–106.

Podolny, J. M., & Baron, J. N. (1997). Doing well and doing goodresources and relationships: Social networks and mobility in theworkplace. American Sociological Review, 62, 673–693.

Powell, W. (1985). Getting into print. University of Chicago Press.Sabel, C. (1982). Work and politics: The division of labor in industry.

New York: Cambridge.Snow, D. A., Zurcher, L. A., Jr., & Ekland-Olson, S. (1980). Social

networks and social movements: A microstructural approach todifferential recruitment. American Sociological Review, 45, 787–801.

Spence, A. M. (1974). Market signaling. Cambridge: Harvard.Stiglitz, J. E. (1991). Symposium on organizations and economics.

Journal of Economic Perspectives, 3, 15–24.Stinchcombe, A. (1963). Some empirical consequences of the

Davis–Moore theory of stratification. American SociologicalReview, 28, 805–808.

Stinchcombe, A. (1990). Information in organizations. Berkeley: Cal-ifornia.

Thompson, J. D. (1967). Organizations in action. New York: McGraw-Hill.

Tilly, C. (1998). Durable inequality. Berkeley, CA: University of Cal-ifornia Press.

Wegener, B. (1991). Job mobility and social ties: Social resources,prior job, and status attainment. American Sociological Review,56, 60–71.

Williamson, O. E. (1975). Markets and hierarchies. Glencoe: FreePress.

Zucker, L., & Rosenstein, C. (1981). Taxonomies of institutional struc-ture: Dual economy reconsidered. American Sociological Review,46, 86984.

Documents

Labor markets and organizations: A screening theory of hiring networks and racially homogeneous employment