30
Research on Employee Recruitment: So Many Studies, So Many Remaining Questions James A. Breaugh and Mary Starke University of Missouri–St. Louis Over the last thirty years, the amount of research on recruitment topics has increased dramatically. Despite this increase, recent reviews of the recruitment literature often have had a somewhat pessimistic tone. Reviewers have concluded that we still do not know a great deal about why recruitment activities have the effects they do. In particular, recent reviews have criticized many of the studies conducted for being poorly designed, narrow in focus, and not grounded in theory. We believe that many of these criticisms are legitimate. We also believe that, in order for future studies to result in a better understanding of the recruitment process, such studies need to be designed with an appre- ciation of the complexity of the recruitment process (i.e., the number of variables involved and the nature of their relationships). In this regard, we offer an organizing framework of the recruitment process. In intro- ducing this framework, we draw upon theories from a variety of re- search domains and give considerable attention to process variables (e.g., applicant attention, message credibility, applicant self-insight) that mediate the relationships between recruitment activities (e.g., re- cruiter behavior) and recruitment outcomes (e.g., the number of appli- cations generated). Having introduced an organizing framework, we selectively review recruitment research, giving particular attention to the topics of recruitment sources, recruiters, and realistic job previews. This review makes apparent a number of important issues that recruit- ment research has yet to address. © 2000 Elsevier Science Inc. All rights reserved. Research interest in the topic of employee recruitment has increased sub- stantially over the last thirty years. As an example of this increasing interest, consider that in the first edition of the Handbook of Industrial and Organizational Psychology, less than one page of coverage was given to the topic of recruitment (Guion, 1976). By the time the second edition of this handbook was published, Direct all correspondence to: James A. Breaugh, School of Business Administration, University of Missouri–St. Louis, St. Louis, MO 63121; Phone: 314-516-6287; Fax: 314-516-6420. Journal of Management 2000, Vol. 26, No. 3, 405– 434 Copyright © 2000 by Elsevier Science Inc. 0149-2063 405

Research on employee recruitment: so many studies, so many remaining questions

Embed Size (px)

Citation preview

  • Research on Employee Recruitment: SoMany Studies, So Many Remaining

    QuestionsJames A. Breaugh and Mary Starke

    University of MissouriSt. Louis

    Over the last thirty years, the amount of research on recruitmenttopics has increased dramatically. Despite this increase, recent reviewsof the recruitment literature often have had a somewhat pessimistictone. Reviewers have concluded that we still do not know a great dealabout why recruitment activities have the effects they do. In particular,recent reviews have criticized many of the studies conducted for beingpoorly designed, narrow in focus, and not grounded in theory. Webelieve that many of these criticisms are legitimate. We also believethat, in order for future studies to result in a better understanding of therecruitment process, such studies need to be designed with an appre-ciation of the complexity of the recruitment process (i.e., the number ofvariables involved and the nature of their relationships). In this regard,we offer an organizing framework of the recruitment process. In intro-ducing this framework, we draw upon theories from a variety of re-search domains and give considerable attention to process variables(e.g., applicant attention, message credibility, applicant self-insight)that mediate the relationships between recruitment activities (e.g., re-cruiter behavior) and recruitment outcomes (e.g., the number of appli-cations generated). Having introduced an organizing framework, weselectively review recruitment research, giving particular attention tothe topics of recruitment sources, recruiters, and realistic job previews.This review makes apparent a number of important issues that recruit-ment research has yet to address. 2000 Elsevier Science Inc. Allrights reserved.

    Research interest in the topic of employee recruitment has increased sub-stantially over the last thirty years. As an example of this increasing interest,consider that in the first edition of the Handbook of Industrial and OrganizationalPsychology, less than one page of coverage was given to the topic of recruitment(Guion, 1976). By the time the second edition of this handbook was published,

    Direct all correspondence to: James A. Breaugh, School of Business Administration, University of MissouriSt.Louis, St. Louis, MO 63121; Phone: 314-516-6287; Fax: 314-516-6420.

    Journal of Management2000, Vol. 26, No. 3, 405434

    Copyright 2000 by Elsevier Science Inc. 0149-2063

    405

  • research on recruitment was seen as meriting an entire chapter (Rynes, 1991).Even though numerous recruitment studies had been published since the 1976Handbook chapter appeared, Rynes perceived this body of research to be lackingin several ways. For example, she found most of the research that she reviewedfocused on one of only three topics (i.e., recruitment sources, recruiters, andrealistic job previews) and that the research on each of these topics was devel-oped in isolation from the others (Rynes, 1991: 399). In her chapter, Rynesshowed how this piecemeal approach to research hindered theory developmentand left unanswered several questions that an employer might have. More re-cently, the expanding body of research on recruitment has been seen as sufficientto merit entire books (e.g., Barber, 1998). Given that dozens of studies werepublished between Rynes (Rynes, 1991) and Barbers (Barber, 1998) reviews ofrecruitment research, it is not surprising that Barber felt that understanding incertain areas (e.g., recruitment source effects) had increased. However, Barberalso pointed out that researchers still had failed to address adequately a number ofsignificant issues (e.g., the site visit). She also noted that methodological weak-nesses (e.g., failure to measure key variables) made it difficult to draw clearconclusions from many studies.

    In summary, if one peruses reviews of recruitment research (e.g., Barber,1998; Breaugh, 1992; Rynes, 1991; Wanous, 1992), one finds a mix of optimismand pessimism. Reviewers were pleased that research on recruitment had in-creased, and they were confident that progress in understanding the effects ofrecruitment activities was being made. Yet, they also seemed frustrated that moreprogress in understanding the recruitment process had not occurred.

    Given Barbers excellent review of recruitment research (Barber, 1998), itseemed unnecessary for us to provide a systematic review of the same research inthis article. However, we also are aware that many readers of this journal will nothave read Barbers book. Thus, we felt it was important to provide a selectivereview of the recruitment literature. We encourage readers who are interested inmore detailed coverage of recruitment research to refer to the works of Barber(1998), Breaugh (1992), Rynes (1991), and others cited herein. In addition toproviding the reader with a selective review of recruitment research, a secondobjective in writing this paper was to stimulate the type of research that will helpanswer many of the research questions that remain even after so many recruitmentstudies have been conducted. In attempting to stimulate research, we havehighlighted unresolved issues and several overlooked topics.

    There are a number of possible approaches for structuring a review. Forexample, given that many readers may not be overly familiar with the recruitmentliterature, we could begin with a review of this research. However, we believe thatit makes the most sense to offer an organizing framework prior to reviewingavailable research. This framework will make apparent the complexity of therecruitment process, complexity that has frequently been overlooked (Barber,1998). It also will show the narrowness of the recruitment literature (i.e., thenumber of key issues that have not been examined). On a more positive note, wehope the framework we offer will stimulate future research that will lead to abetter understanding of the recruitment process (i.e., the interaction among re-

    406 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • cruitment variables and the relationship between recruitment variables and othervariables). The organizing framework presented in Figure 1 is based upon modelsthat have been offered by Barber (1998), Breaugh (1992), and Rynes (1991). Weacknowledge that our framework represents a simplified view of reality. Inaddressing the segments of this framework, we will introduce the theoreticalexplanations that have been offered for the relationships between variables. Wewill not discuss specific research findings until the entire framework has beenintroduced.

    Before presenting our organizing framework, it seems appropriate to offer adefinition of what we mean by the term recruitment. Although the term recruit-ment is commonly used, it is not easy to define (see Barber, 1998: 56). For ourpurpose, the definition offered by Barber will suffice: recruitment includes thosepractices and activities carried on by the organization with the primary purpose ofidentifying and attracting potential employees (p. 5).

    An Organizing Framework for Understanding the Recruitment Process

    As noted by Rynes (1991), much recruitment research has focused on theeffects of recruitment sources (e.g., Do individuals referred by current employeeshave a lower turnover rate than persons recruited via newspaper ads?), recruiters(e.g., Do recruiters who offer more information about a job make a betterimpression on job applicants?), and realistic job previews (e.g., Does providingaccurate job information result in a higher level of job satisfaction for newemployees?). As depicted in Figure 1, all of these topics can be consideredrecruitment activities. Although it is understandable why researchers have fre-quently focused on the effects of recruitment activities (e.g., most companiesdocument how applicants heard about job openings), we believe that moreattention should be focused upon the entire recruitment process in order to betterunderstand whether an employers recruitment activities will accomplish itsobjectives.Establishing Recruitment Objectives and Strategy Development

    In discussing the recruitment process, Barber (1998) delineated three phases(i.e., generating applicants, maintaining applicant status, and influencing jobchoice decisions). That is, (a) certain recruitment activities (e.g., advertising on aSpanish-speaking radio station) may influence the number and type of individualswho apply for a position, (b) certain activities (e.g., professional treatment duringa site visit) may affect whether job applicants withdraw during the recruitmentprocess, and (c) certain recruitment actions (e.g., the timeliness of a job offer) mayinfluence whether a job offer is accepted.

    In planning a strategy for generating applicants, a fundamental question thatshould be addressed is, What type of individual does the organization want torecruit (e.g., What knowledge, skills, and abilities are important? Is a diverseapplicant pool desired?)? Until an employer determines the type of applicants itseeks, it is difficult for it to address several other strategy-related questions:Where should the organization recruit (e.g., colleges versus state employment

    407RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • Figu

    re1.

    AM

    ode

    loft

    heO

    rgan

    izat

    iona

    lRec

    ruitm

    entP

    roce

    ss

    408 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • offices)? What recruitment sources should it use to reach the desired applicantpopulation (e.g., the Web versus job fairs)? When should the employer beginrecruiting (e.g., at the start of a college students senior year versus during thesecond semester)? What message should it convey to potential applicants (e.g., agood deal of job-related information versus information on a few key aspects ofthe job)? In order for an organization to intelligently answer these basic questions,it must have a clear sense of what its recruitment objectives are (Breaugh, 1992).For this reason, we have portrayed the establishment of objectives as the firstphase of the recruitment process (see Figure 1).

    In the past, it appears that many organizations have had the simple recruit-ment goal of attracting a large number of job applicants (Wanous, 1992). For avariety of reasons (e.g., the cost of processing applications), several researchershave questioned the wisdom of simply trying to attract a large number ofapplicants. Instead, it has been suggested (Rynes, 1991) that employers would bewise to consider a wider range of possible recruitment objectives. For example,some employers might be interested in trying to influence one or more post-hireoutcomes by the way they recruit. Such post-hire outcomes include: the jobsatisfaction of new employees, their initial job performance, whether the organi-zation is seen as living up to the psychological contract that has been established,and the first-year retention rate of new hires. In establishing recruitment objec-tives, organizations might also focus their attention on post-hiring outcomes thatcan be measured the day employees begin work (Breaugh, 1992). Such outcomesmight include: the cost of recruiting, the speed with which jobs were filled, thenumber of individuals hired, and/or the diversity of the new employees.

    Although recruitment activities have been linked to some of these post-hireoutcomes, some researchers (e.g., Williams, Labig, & Stone, 1993) have arguedthat in recruiting many employers are not overly concerned with post-hire out-comes. Rather, they are interested in pre-hire outcomes such as the number ofindividuals who apply for a position, the quality of these applicants, their diver-sity, and the number of individuals who accept job offers that are extended. If anemployer is interested in these more proximal outcomes of the recruitmentprocess, its strategy development should be focused on how to accomplish them.

    In summary, we believe that the first stage of the recruitment process shouldbe the establishment of objectives. If clear objectives have not been established,it is difficult to develop a sound recruitment strategy (Rynes & Barber, 1990).Having established a core set of objectives, an organization can more intelligentlyanswer the strategy development questions posed earlier (e.g., What recruitmentsources to use?). Having developed a recruitment strategy, an employer can thenundertake the recruitment activities that are likely to lead to its desired objectives.Intervening/Process Variables

    Although we believe that the recruitment process should proceed in themanner portrayed in Figure 1 (i.e., recruitment objectives3strategydevelopment3recruitment activities3recruitment results), it is useful to addintervening/process variables to this model and to work backwards through themodel. That is, in order to make decisions about what recruitment activities to

    409RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • undertake (i.e., strategy development), an employer needs to understand whycertain recruitment activities may result in certain recruitment outcomes (Rynes,1991). We use the term intervening/process variable as a label for the factors thathave been hypothesized to explain the relationships between recruitment activitiesand outcomes. A number of potential intervening variables are portrayed in Figure1.

    In terms of generating applicants, it is critical that an employers recruitmentactions attract the attention of potential job applicants (Barber, 1998). Researchsuggests that the following attributes are likely to generate attention: (a) messagesthat are vivid in nature (e.g., include pictures) and include concrete language(Tybout & Artz, 1994), (b) messages that convey unexpected information (Kulik& Ambrose, 1993), (c) messages that provide personally relevant information(Chaiken & Stangor, 1987), and (d) messages that are conveyed in face-to-faceconversations (Tybout & Artz, 1994).

    In addition to attracting attention, recruitment communications need to beunderstandable and viewed as credible by the individuals whom the organizationis interested in recruiting (Breaugh & Billings, 1988). In terms of a message beingunderstood, such mundane factors as using an appropriate level of expression andchoosing the correct language (e.g., Spanish) are clearly necessary (Jablin, Put-nam, Roberts, & Porter, 1987). The medium used to deliver a message can alsoinfluence comprehension. For example, Stiff (1994) suggested that, in contrast toa verbal message, a written message may increase understanding given that themessage can be reread and studied. In contrast, Lengel and Daft (1988) suggestedthat in-person communication may lead to greater understanding, especially whenthe message is somewhat complex. Lengel and Daft argued that in-person com-munication is a richer medium in that tone of voice, non-verbal cues, and othervariables exist that are lacking in a written message.

    With regard to message credibility, research (Stiff, 1994) has consistentlyshown that communicator expertise and trustworthiness lead to a messages beingbelieved. In terms of expertise, generally those who are closest to the worksituation (e.g., job incumbents) are seen as being an informed source of job-related information (Fisher, Ilgen, & Hoyer, 1979). With regard to trustworthi-ness, research has shown that receiving information that is different than thatexpected from the message source results in credibility (Stiff, 1994). For example,in persuasion research, it has frequently been found that communicators whoconvey information that detracts from their position are rated as more trustworthythan communicators who cite only information that supports their position(Chaiken & Stangor, 1987). In contrast, communicators who are likely to derivepersonal benefit from misleading the communication recipient frequently lackcredibility (Stiff, 1994). Research (Harkins & Petty, 1981) also has shown thatreceiving a consistent message from multiple sources results in an attribution ofcredibility.

    In order to be effective, a recruitment message also needs to generate initialinterest from potential job applicants. Such interest is more likely to be forth-coming if a job opening (e.g., the job itself, the organization, the location) isviewed positively (Barber & Roehling, 1993). Interest is also more likely to be

    410 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • forthcoming if an applicant perceives that he/she has a reasonable likelihood ofreceiving a job offer (Rynes, 1991).

    As noted by Barber (1998), the recruitment process does not end with thegeneration of job applicants. An organization also should be concerned aboutmaintaining the interest of job candidates and influencing their job choice deci-sions. With regard to intervening variables, many of the same ones that affect thegeneration of job applicants also are important for these later two stages of therecruitment process. For example, individuals are likely to remain job candidatesif they continue to see a job opening as attractive and if they perceive themselvesas likely to receive job offers as they move through the recruitment process(Rynes, Bretz, & Gerhart, 1991). Similarly, the attractiveness of a job openingshould have a major influence on a job applicants decision whether to accept ajob offer (Turban, Eyring, & Campion, 1993).

    In attempting to understand the effects of recruitment activities, two keyintervening variables are the accuracy of applicants job expectations (Wanous,1992) and their degree of self-insight (Breaugh, 1992). If an employer is inter-ested in influencing such post-hire outcomes as job satisfaction, initial jobperformance, perceptions of violations of the psychological contract, and first-year retention rate, it should be particularly concerned with whether job applicantshave an accurate perception of what a job entails and whether they have insightinto their own capabilities (e.g., knowledge, skills) and desires (e.g., what theywant from a position and an employer). For example, given that the labor marketallows job candidates who do not perceive a good person-job fit to self-select outof job consideration, applicants who have an accurate perception of a job situationand have self-insight should, if hired, find their jobs satisfying (otherwise, theywould be unlikely to accept job offers). Being satisfied, new employees should beless likely to quit during the initial employment period (Wanous, 1992). Self-selection also should favorably influence the initial performance of new hires(e.g., Downs, Farr, & Colbeck, 1978). This assertion is based upon the assumptionthat, if a person does not perceive he or she has the ability to perform well, theperson will be less likely to accept the job offer.

    Regardless of whether the labor market allows applicants to self-select out ofjob consideration, having realistic job expectations should influence new employ-ees perceptions of psychological contract violations (Rousseau, 1995). For ex-ample, if an applicant has no better options, even after receiving information thatmakes a job appear undesirable, he or she may accept a job offer. However, if theemployer has been honest with the individual during the recruitment process, eventhough he or she may not like the new position, the person should not feel that theorganization has failed to live up to its side of the employment contract. It also hasbeen argued (Wanous, 1992) that first-year retention rate may be positivelyaffected by individuals having had accurate job information during the recruit-ment process, even when the labor market does not allow for self-selection. Thisargument is based on the assumption that, having made an informed job choicedecision, new employees will feel an obligation to remain in the job for at leasta few months. In summary, in order to understand why certain recruitment

    411RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • activities may impact certain recruitment outcomes, it is necessary to understandthe role that several intervening/process variables play.

    Recruitment Activities that are Likely to Influence Key Intervening/Process Variables

    To date, most of the research conducted has addressed recruitment sources,recruiters, and realistic job previews (Barber, 1998). Given this fact, we will focusmost of our attention on these three activities, their hypothesized relationship withintervening variables, and how such process variables influence several outcomevariables. As noted earlier, in introducing our framework of the recruitmentprocess, we are focusing upon theoretical rationales that have been offered for therelationships between recruitment activities and the intervening variables.

    Recruitment sources. Although employers use a variety of recruitmentsources, little recent data exist on the frequency of usage. The most commonlycited study of source usage was conducted by the Bureau of National Affairs(1988). This study examined the frequency of source usage across five job types(office/clerical, production/service, professional/technical, commission sales, andmanagers/supervisors). Although the results of the BNA study are too complex tofully discuss here, a few general trends are worth noting. Regardless of job type,newspaper ads, employee referrals, direct applications (i.e., individuals who applyfor a job without having received information that a job opening existed), andrecruiting at schools were commonly used sources. A few sources were closelytied to the type of job opening. For example, state employment service officesonly were commonly used for filling office and production jobs. The NationalOrganizations Study, which involved a national probability sample of employers,reported findings similar to those of the BNA survey (Kalleberg, Knoke, Marsden,& Spaeth, 1996). Reports from a national probability sample of employees(Vecchio, 1995) concerning how they found their current positions further supportthe general finding that firms rely heavily upon employee referrals, newspaperads, and direct applications.

    With regard to recruitment sources, two theoretical explanations (i.e., therealistic information hypothesis and the individual difference hypothesis) for whysources may be differentially associated with recruitment outcomes have attractedthe most attention (Barber, 1998). The realistic information hypothesis proposesthat persons recruited via certain sources are likely to have more accurateinformation about what a job entails (Rynes, 1991). Possessing such informationis thought to enable an applicant to make a more informed decision about whetherto pursue a job. For example, with regard to employee referrals, it has beensuggested (Breaugh, 1992) that before referring a person for a job with their firm,current employees would provide the individual with realistic information aboutwhat a job involved. Concerning direct applications, it has been suggested (e.g.,Breaugh & Mann, 1984) that they may possess more realistic information con-cerning a job opening as a result of having done more research on potentialemployers than applicants coming from other sources. Conversely, it has beenargued that individuals recruited by other sources (e.g., school placement offices,newspaper ads) may lack realistic job information (Williams et al., 1993). Lacking

    412 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • such information, these individuals are thought to be less likely to be able toself-select out of consideration for jobs that may not be a good fit in terms of theirskills or interests. If hired, these less informed individuals are more likely to beunhappy with their job choice and may be more likely to resign.

    The second explanation commonly offered for recruitment source differ-ences is known as the individual difference hypothesis (Rynes, 1991). Thisexplanation is based on the premise that sources differ in the types of individualsthey reach, and that these differences result in different outcomes. For example,Ullman (1966) speculated that persons recruited via employee referrals may bemore capable than individuals recruited from such sources as newspapers. Thisassertion is based on the assumption that current employees would screen indi-viduals before making job referrals because they may see their own reputation asbeing affected by the quality of the referrals. Other researchers have speculatedthat different recruitment sources reach applicant groups that may differ oncharacteristics such as motivation or perceived job mobility (Barber, 1998).Unfortunately, researchers have not specifically described which sources shouldbe linked to which individual difference variables (e.g., Would direct applicantshave a higher motivational level to find a job because applying for job openingsthat may or may not exist takes more effort than simply responding to advertisedpositions?).

    Given the organizing framework we presented, it should be apparent that webelieve that more theory development needs to take place with regard to recruit-ment sources and their relationship with process variables. For example, as Barber(1998) noted, recruitment scholars have largely ignored the question of how toattract attention to recruitment materials (p. 39). In terms of attracting theattention of potential job applicants, one may be able to build a theoretical casethat the use of direct mail (a rarely examined recruitment source) or employeereferrals (which involve actively seeking out individuals who may be interested ina job opening) would be attention-getting recruitment sources compared to suchpassive sources as newspaper advertising (a potential job candidate may not belooking for a position). Based upon psychological research (Stiff, 1994), onemight hypothesize that job advertisements placed in atypical locations in anewspaper may be attention-getting (Breaugh, 1992). With regard to messagecomprehension and credibility, theory development also is lacking. One mighthypothesize that employee referrals would be a particularly credible source ifmade by job incumbents as opposed to supervisors who may be perceived byapplicants as under pressure to fill vacant positions. Given that most employeereferrals involve two-way interactions, referrals may be a source that facilitatesmessage comprehension (Lengel & Daft, 1988). Although we could offer furtherspeculation about what intervening processes explain why recruitment sourceshave the effects on outcome variables (e.g., job survival) that have been docu-mented (see Wanous & Colella, 1989), such theory development is beyond thescope of this paper.

    In summary, in terms of the relationships between recruitment sources andkey intervening variables (e.g., applicant attention), it suffices to say that muchtheoretical work remains. In the future, researchers need to move beyond simply

    413RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • considering recruitment source effects that are associated with realistic job ex-pectations and individual differences. Once theoretical linkages have been spec-ified between various recruitment sources and key intervening variables, it is morelikely that the type of theory-driven empirical research called for by Rynes (1991),Barber (1998), and others will be conducted.

    Recruiters. Researchers have offered a number of explanations for whyrecruiters may have an effect on job candidates. The first of these has to do withrecruiter informativeness. Researchers (e.g., Powell, 1991) have theorized thatsome recruiters provide more information and more specific information toapplicants than other recruiters. Given that the information provided is of personalrelevance to the job candidate and is believed, interacting with more informativerecruiters should have numerous beneficial effects (e.g., enhanced ability ofapplicants to remove themselves from consideration for jobs that ultimately wouldnot be satisfying). In terms of recruiter informativeness, it has been suggested(Breaugh, 1992) that an individuals prospective supervisor or coworkers shouldbe particularly informative in contrast to other recruiters (e.g., a person from thehuman resources department) who may not possess as much information.

    Researchers (e.g., Maurer, Howe, & Lee, 1992) also have hypothesized thatrecruiter credibility helps explain the differential effects of recruiters on appli-cants. For example, Fisher et al. (1979) hypothesized that, in comparison to jobincumbents, corporate recruiters would lack credibility. Fisher et al. based theirassertion on the assumption that corporate recruiters would be viewed by appli-cants as lacking expertise concerning what a job involves and would be perceivedas having a vested interested in filling open positions.

    It also has been suggested (Rynes, 1991) that recruiters may have an impactbecause job candidates view them as signals of unknown organizational attributes.For example, researchers (e.g., Connerley & Rynes, 1997) have suggested thatrecruiter personableness may be important because it signals how the personmay be treated if hired or how likely the person is to receive a job offer. In asimilar vein, interacting with female or minority recruiters may signal to anapplicant that an employer values diversity. Such a value system may make a jobmore attractive to certain job candidates (Highhouse, Stierwalt, Bachiochi, Elder,& Fisher, 1999).

    Although researchers have theorized about the positive effects of recruiterswho are informative, credible, personable, and demographically diverse, research-ers have not focused heavily on the underlying reasons why different types ofrecruiters may influence the attention potential applicants give to a job opening orthe interest applicants have in a position. Given the significance of attractingapplicant attention and generating their interest, clearly these two issues meritfuture attention. A topic related to the informativeness of a recruiter is the degreeto which the message conveyed by a recruiter involves sharing realistic asopposed to only positive information. Because the topic of providing realisticinformation is generally covered in the context of realistic job previews, we willdiscuss this issue in the next section.

    In summary, a strong case can be made that the choice of a recruiter canmake a difference (Rynes et al., 1991). However, at present, although researchers

    414 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • have provided some theoretical foundation for why recruiters may have beneficialeffects, more theory development needs to take place.

    Realistic job previews. According to Phillips (1998), no recruitment issuehas generated more attention than realistic job previews (RJPs), the presentationby an organization of both favorable and unfavorable job-related information tojob candidates (p. 673). Given the attention that RJPs have attracted, it seemsreasonable to ask, Is possessing inaccurate job expectations really a majorissue? Unlike recruiting sources, national sample data on realistic expectationshave not been gathered. However, data from individual studies suggest that unmetexpectations are a frequent occurrence. In a review of several RJP studies,Wanous (1992) suggested that unmet expectations on the part of new employeesare frequently reported. In the context of psychological contracts, Rousseau(1995) provided ample evidence that new hires often see their organizations as nothaving lived up to agreed-upon arrangements.

    Compared to the topics of recruitment sources and recruiter effects, there hasbeen considerable theory development with regard to RJPs. For example, Breaugh(1992); Wanous (1992); Fedor, Buckley, and Davis (1997); Hom, Griffeth, Palich,and Bracker (1998); and Saks and Cronshaw (1990) have all offered models of theprocess by which RJPs affect such outcome variables as job survival, workattitudes, and job performance (an adaptation of Breaughs model is portrayed inFigure 2). Given the numerous models in existence, it is not possible to discuss allof their subtle nuances. However, it is worthwhile to discuss key conceptualvariables that most of the models have in common.

    Most RJP models hypothesize that providing realistic job information toapplicants results in their having their job expectations met. It is further hypoth-esized that providing an RJP influences role clarity (which is thought to affect jobperformance) and individuals perceptions that the organization was honest withthem. Given that candidates perceive alternatives to accepting an undesirable job,providing an RJP is hypothesized to result in applicant self-selection, which, inturn, is hypothesized to result in a higher level of job satisfaction, a lower levelof voluntary turnover, and a higher level of performance.

    Although the various models of RJP effectiveness include a number ofvariables, many models have failed to address some key process variables. Forexample, as pointed out by Phillips (1998), all theories about the effectiveness ofrealistic job previews share the underlying assumption that their message isreceived and processed by applicants (p. 673). Given the significance that wehave attached to job applicant attention and comprehension, it should not besurprising that we think that more theory development should address these twovariables. From our review of the research literature, it appears that Colarelli(1984) is the only researcher who has given much attention to many of the processvariables we highlighted.

    In establishing a theoretical framework for his study, Colarelli addressedprocess variables only in the context of providing an RJP via a brochure versusproviding an RJP by means of a face-to-face conversation with a job incumbent.Colarelli hypothesized that, in comparison to receiving an RJP in written form,RJPs received via a face-to-face interaction would attract greater attention, be

    415RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • Figure 2. The Formation of Job Expectations and their Influence on ImportantEmployee Attitudes and Behaviors: A Process Model (adapted from Breaugh, 1992)

    416 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • more understandable, and be more credible. Although he did not address why hethought a face-to-face conversation would attract attention, Colarelli suggestedthat a conversation aided comprehension due to the fact that an applicant couldask questions to clarify ambiguous information. He argued that job incumbentshave considerable credibility given that they are generally seen as more knowl-edgeable than other sources of information (Colarelli, 1984: 634) and becausethey are generally seen as more trustworthy.

    Over the years, a number of researchers have suggested some theoreticalrefinements to the basic processes that have been hypothesized for explaining RJPeffects. We will highlight three of these suggestions. In an important (butsomewhat overlooked) paper, Greenhaus, Seidel, and Marinis (1983) questionedthe emphasis that researchers have placed on met expectations. Greenhaus et al.argued that, if a persons job expectations were negative and met (e.g., the personwas not able to turn down an undesirable job), one would not generally predict jobsatisfaction. Alternatively, a person could be pleasantly surprised about certainaspects of a job upon beginning work (i.e., not all unmet expectations areunpleasant). Instead of focusing so heavily on met expectations, Greenhaus andhis colleagues argued that more attention should be paid to whether a job providedfor value attainment (i.e., a match between a persons job values and jobexperiences). These researchers not only proposed value attainment as an inter-vening variable between having realistic expectations and experiencing satisfac-tion, they tested their assertion. In two studies, they found that value attainmentaccounted for significantly more variance in facet satisfaction than did realisticexpectations (p. 394).

    In order to better understand the effects of RJPs, Fedor et al. (1997) arguedfor more attention being given to the attributions new employees make for anydiscrepancies that occur between what they expected and what they experienced.For example, they suggested that new employees who perceive that an organiza-tion misled them about job features would likely react more negatively than ifthey attributed unmet expectations to the fact that the job had changed since theposition was described to them or to some other reason.

    Researchers also have suggested that more attention needs to be given tocharacteristics of the RJP recipients. For example, Meglino, DeNisi, Youngblood,and Williams (1988) hypothesized that RJP effects would be stronger for moreintelligent applicants given that they could better comprehend the informationconveyed. Breaugh and Billings (1988) hypothesized that RJP effects would bestronger if the RJP information addressed aspects of a job that were seen asparticularly important by job candidates.

    In summary, although researchers have given considerable attention to thereasons why RJPs work, more theory development is needed. In particular, webelieve that researchers need to move beyond viewing RJPs as the primarymechanism for influencing the accuracy of applicant job expectations. The realismof job expectations can be influenced by individuals having similar prior workexperience (Breaugh, 1992) or having completed internships (Taylor, 1988).

    Other message attributes. Although researchers have given the majority oftheir attention to whether a recruitment communication contains realistic (as

    417RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • opposed to only positive) information (Phillips, 1998), three other messageattributes (i.e., the breadth of topics addressed, the specificity of the informationprovided, the timing of communications) merit attention. It has been well docu-mented that during the job search process individuals frequently lack informationabout job and organizational attributes. For example, Maurer et al. (1992) foundthat engineering students who were job hunting reported they lacked informationabout such issues as starting salary, how raises are determined, benefits, and thesuccess of new hires. The students who reported they lacked information aboutpositions also reported they were less likely to accept job offers. Similarly, Barberand Roehling (1993) found that individuals who reported having more informa-tion about a job and/or an organization were more attracted to the organization.

    In attempting to explain why a lack of information may result in a negativereaction from recruits, Barber and Roehling suggested that job applicants may seethe failure of an organization to provide sufficient information as an indicator ofsloppy, disinterested recruiting practices (Barber & Roehling, 1993: 853). Inother words, an employers failure to supply information may be viewed as asignal of its lack of professionalism or of its lack of interest in the candidate.Although some researchers (Breaugh & Billings, 1988) have theorized that anemployers failure to provide information about a range of important job attributes(e.g., supervisory style, coworker relations) can hurt its ability to recruit, High-house and Hause (1995) have qualified this general prediction. These authorsagreed that decision makers seek to avoid uncertainty. Thus, in general, jobapplicants should tend to devalue a job opening for which job attribute informa-tion is lacking. However, Highhouse and Hause suggested that there is anexception to this general tendency to avoid uncertainty. They hypothesized thatdecision makers might actually prefer a job opening with missing information ona job attribute over an alternative job opening for which information was availableon the attribute but the attribute was viewed as being below average.

    Concerning information specificity, Breaugh and Billings (1988) argued that,even when an employer tried to present realistic job information, frequently theinformation presented was so general (salaries are competitive) that it did notallow for informed recruit decision-making. More recently, Barber and Roehling(1993) addressed the issue of information specificity from a different perspective.These researchers hypothesized (and confirmed) that job applicants pay moreattention to specific than to general information. Given that attracting applicantattention is a key variable in the recruitment process, it appears that by providingmore specific information an employer may be able to generate more applicantinterest.

    With regard to the timing of recruitment communications, relatively littletheory development has taken place (Barber, 1998). Based upon research indecision making, Breaugh (1992) hypothesized that a desire for uncertaintyavoidance on the part of job applicants might make it wise for an organization tomake job offers early in the recruitment cycle. Thurow (1975) suggested that theadvantage of early recruitment may only hold for more desirable employers.Rynes et al. (1991) focused upon the timing of recruitment communications interms of signaling. These authors hypothesized that delays in responding to

    418 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • applicant inquiries or in making job offers may be viewed as a sign that theorganization does not have a strong interest in the candidate.

    A Review of Recruitment ResearchHaving introduced a framework for understanding the recruitment process

    and having summarized theoretical rationales that have been offered to explain therelationships portrayed in our model, we now will provide a selective review ofempirical recruitment literature.Research on Recruitment Sources

    One of the first issues that an organization needs to address in trying to filljob openings is how to reach individuals who may be qualified for and interestedin the openings. Over the years, several researchers have investigated the use ofdifferent recruitment sources and whether certain sources are linked to beneficialoutcomes. A study by Ullman (1966) was one of the first to examine recruitmentsources. He found that new employees who were recruited by means of informalsources (i.e., employee referrals, direct applications) had a lower turnover ratethan individuals recruited via formal sources (i.e., newspaper advertisements,employment agencies).

    The publication of Ullmans study (Ullman, 1966) stimulated several addi-tional recruitment source studies (e.g., Blau, 1990; Gannon, 1971; Saks, 1994).Some of these studies attempted to test whether the two explanations (the realisticinformation hypothesis and the individual difference hypothesis) that have beenoffered to explain source effects were supported by empirical data. For example,Taylor and Schmidt (1983) examined source differences in absenteeism, turnover,and performance for employees who were recruited by a variety of sources (e.g.,referrals, rehired former employees). They also measured several individualdifference variables that might explain any source differences they found. Taylorand Schmidt reported that, compared to the other sources, rehires had longertenure and lower absenteeism. They also found some support for the individualdifference explanation for their findings. One would also presume that those in therehire group had a fairly accurate view of what the job involved.

    Although the results of recruitment source studies with regard to outcomevariables are not entirely consistent, it frequently has been found that employeereferrals and direct applications had lower levels of turnover and higher levels ofsatisfaction than persons recruited by more formal sources (Breaugh, 1992).However, empirical data have not consistently supported either the informationrealism or the individual difference explanations for the recruitment sourcedifferences reported (Barber, 1998). The failure of research to support theseexplanations for source differences clearly raises concern about their validity.However, methodological weaknesses of studies is an alternative explanation forthe failure to find support for the information realism and individual differencehypotheses. We will discuss two of these methodological limitations (i.e., thesamples utilized, the measurement of variables).

    In order to adequately test whether different recruitment sources result in jobapplicants who differ in terms of the realism of job expectations or with regard to

    419RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • individual differences, a researcher needs to gather data from job applicants. Yet,as noted by Barber (1998), in the great majority of recruitment source studies, datahave been gathered from new employees. Consider the study by Breaugh andMann (1984). In this study, it was reported that recruitment sources did not affectthe ability level of new employees. Although this may be the case, one canquestion how relevant the data reported by Breaugh and Mann (1984) are totesting the individual difference hypothesis. Although new employees in thisstudy may not have differed in terms of their ability, applicants recruited fromdifferent sources may have. However, such an applicant ability difference mayhave been eliminated as a result of selection procedures used by the hiringorganization. In summary, given that Breaugh and Mann tested for ability differ-ences using new employees rather than job applicants, it is not clear what one canconclude from their results.

    In contrast to the results reported by Breaugh and Mann, the two studies thathave examined source differences on applicants suggest that informal recruitmentsources were linked to more qualified job applicants. For example, Kirnan, Farley,and Geisinger (1989) found that persons recruited from informal sources (e.g.,employee referrals) were of higher quality (and were more likely to be offeredjobs). Williams, Labig, and Stone (1993) also found that informal sourcesreached differently qualified applicants in terms of nursing experience andeducation which, in turn, were valid predictors of subsequent nurse performance(p. 163). In summary, in order to advance our understanding of why recruitmentsources may influence work outcomes, more attention needs to be focused onapplicant differences on key intervening variables.

    A second methodological weakness of many recruitment source studiesconcerns the measurement of variables. In order to provide a fair test of therealism and the individual explanation for source effects, it is critical that thesevariables be measured precisely and measured at the appropriate time. Withregard to measurement precision, researchers have suggested that recruitmentsource effects may be due to individual differences in such factors as motivation,ability, or perceived ease of movement. However, instead of measuring variablessuch as ability, investigators all too often chose demographic proxies bearinglittle theoretical relevancy to criteria of source effectiveness (Griffeth, Hom,Fink, & Cohen, 1997: 20). Even when researchers have attempted to measureindividual difference variables more directly, problems emerge. For example,although Griffeth et al. assessed quality in a rigorous fashion (i.e., during inter-views, ratings were made of work experience, training, etc.), these authorsreported data on new hires rather than applicants. Measurement precision also hasbeen a problem in testing the realism explanation for source differences. Forexample, although Williams et al. (1993) gathered data from job applicants, theytested the realism explanation for source differences indirectly (i.e., they mea-sured the amount of pre-hire information the applicant possessed, rather than theaccuracy of this information).

    Conceivably, the reason that Williams et al. focused upon informationquantity is due to the time frame for measuring variables in their study. Theseauthors measured several key variables during the job application process. Given

    420 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • this time frame, it would have been difficult for individuals to assess the accuracyof their job expectations. In contrast to Williams et al., who measured variablesduring the application process, several researchers can be criticized for waitingtoo long to measure variables. For example, in attempting to measure recruitmentsource effects, many researchers (e.g, Griffeth et al., 1997) have measured jobsatisfaction, performance, and turnover at the end of one year of employment. Itis quite likely that recruitment effects would be stronger if measured closer to thetime of hiring (Wanous, 1992). Similarly, some researchers (Breaugh & Mann,1984) measured the accuracy of job expectations after a year on the job. Incontrast to relying so heavily on a persons memory, Saks (1994) measured therealism of job expectations after one month of employment.

    Having reviewed the research on recruitment sources, Barber (1998) con-cluded that past research has not made a strong case for the importance of sourcedifferences. We agree. However, based upon the findings of the two studies(Kirnan et al., 1989; Williams et al., 1993) that presented data on job applicants,we believe that it is premature to conclude that recruitment sources do not meritattention. Instead, we believe that research in this area needs to be reoriented.Future research should focus more heavily upon job applicants, rather than newhires, and should give more attention to the measurement of key variables. Withregard to future research, there are two remaining themes we would like toemphasize (i.e., expanding the range of intervening variables that are considered,giving more attention to the employers perspective).

    If one reflects on the reasons why recruitment sources have been hypothe-sized to be associated with certain work outcomes, several factors that mayexplain inconsistent results become apparent. Consider the use of employeereferrals. Researchers have suggested that one reason that employee referrals maybe associated with beneficial work outcomes is due to the fact that employeesmaking referrals may pre-screen potential applicants. As was noted by Breaugh(1992), pre-screening is most likely to occur when the person making the referralis concerned about his or her reputation in the organization. Given that somerecruitment source studies have involved work situations (e.g., summer jobs) inwhich the employees making referrals may not always have been concerned abouttheir reputations, one may question how much pre-screening actually took placein certain studies. In order to better understand the recruitment process, in futureresearch, it is important that data be gathered from those who made referrals toassess whether pre-screening actually occurred. In a similar vein, it has beensuggested that individuals recruited via different sources may differ in terms oftheir motivation to find a job and/or their perceptions of job mobility. Clearly, itis time for researchers to start measuring such explanatory variables directlyrather than measuring demographic variables that may be related to them (Griffethet al., 1997).

    In the first part of this paper, we argued that more attention should be givento several process variables. With regard to recruitment sources, we think this isa critical need. For example, consider the issues of attracting applicant attention(Does an attendee at a job fair stop at a particular employers booth?) andgenerating applicant interest (Does an attendee submit an application?). To our

    421RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • knowledge, no recruitment source study has addressed the issues of attractingattention and generating interest. Yet, in a labor market in which jobs are plentiful,these are critical issues for employers.

    In addition to focusing upon process variables such as applicant attention,researchers also need to focus more heavily upon important pre-hire variables onwhich recruitment sources are likely to have a more immediate impact than theydo on post-hire variables. For example, corporate recruiters tend to be interestedin such factors as whether the quality of job applicants is higher for certainrecruitment sources, whether certain sources are more likely to yield a higherpercentage of new hires, and whether certain sources are likely to generateminority applicants (Barber, 1998). Given that only two studies (Kirnan et al.,1989; Williams et al., 1993) have presented data on applicants as well as newhires, it is difficult for current research to shed much light on these key concerns(Kirnan et al.s findings suggested that individuals recruited via referrals tendedto be of higher quality than persons recruited by means of more formal sourcesand that minorities were more likely to use formal recruitment sources).

    As noted by Barber (1998), it is curious that little research has investigatedwhy employers recruit as they do. Two studies by Rynes and her colleagues areexceptions. Rather than looking at how employers determined what recruitmentsources to use, Rynes and Boudreau (1986) examined how 145 companiesconducted college recruiting programs. Among the issues they investigated werethe factors that affected the companies selection of colleges. Rynes and Boudreaureported that the factors given the most weight were a colleges reputation incritical skill areas, the general reputation of the school, and the performance ofprevious hires from the school. As pointed out by Rynes, Orlitsky, and Bretz(1997), much of recruitment research has focused on recruiting college students.In an attempt to understand how employers make decisions about whether torecruit new college graduates as opposed to targeting more experienced individ-uals, Rynes et al. gathered data from staffing professionals at 251 companies.They found that experienced hires were evaluated more highly in terms of theirtechnical skills, their interpersonal skills, the realism of their job expectations, andtheir likelihood of success on the job. New college graduates received higherratings for being willing to learn new things. Rynes et al. also asked the staffingprofessionals to rate the overall effectiveness of several recruitment sources.Among the sources receiving above-average ratings were informal referrals,newspaper ads, private search firms, and direct applications. College placementservices, professional associations, and on-line employment exchanges receivedbelow-average ratings.

    Given the problems that have plagued recruitment source research (e.g., thefocus on new employees rather than applicants), it is difficult to draw many firmconclusions. On a more positive note, this is an area that is ripe for real progressto be made. In this paper, we have emphasized the importance of further theorydevelopment. However, we also think that more researchers should follow thelead of Rynes and her colleagues in investigating why employers make thedecisions that they do (e.g., why certain recruitment sources are selected).

    422 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • Research on RecruitersGiven the central role that recruiters play in the recruitment process, it is not

    surprising that considerable research has addressed the issue of recruiters and theirimpact on job candidates. Alderfer and McCord (1970) conducted one of the firststudies of recruiter characteristics. They found that applicant perceptions ofrecruiters behaviors (e.g., their willingness to provide information) and attitudes(e.g., their interest in the candidate) were associated with applicants expectationsof receiving a job offer and their reported probability of accepting the offer.

    Following the publication of Alderfer and McCords study, several otherresearchers (e.g., Harris & Fink, 1987; Macan & Dipboye, 1990; Taylor &Bergmann, 1987) examined the impact of recruiter informativeness and person-ableness. Although these studies differed in terms of their research designs (e.g.,Taylor and Bergmann gathered longitudinal data while most researchers gatheredcross-sectional data) and their data analysis strategies (e.g., unlike most research-ers, in order to better evaluate recruiter effects, Harris and Fink controlled forpre-interview impressions of job attributes), they generally replicated the resultsof Alderfer and McCord. That is, in several studies, job applicants who perceivedrecruiters as being informative and personable also rated the prospective job andthe prospective employer as being more attractive. These positive impressions ofthe job and the employer were, in turn, related to applicants reporting a higherexpectancy of receiving a job offer and a higher likelihood of accepting a joboffer.

    At the same time that researchers investigated the relationship betweenapplicant perceptions of recruiter behavior and their reactions to the recruitmentprocess, they frequently also examined whether recruiter demographic character-istics or other recruiter attributes (e.g., training) were related to applicant reac-tions. In terms of recruiter demographic characteristics, a few studies have linkedthem to applicant reactions to the recruitment process. However, in general,studies have either not found a relationship or the relationships reported weresmall in magnitude (Barber, 1998). As an example, consider recruiter gender.Barber and Roehling (1993) and Taylor and Bergmann (1987) found that inter-acting with male recruiters had a small positive association with applicantsreactions to the recruitment process. In contrast, Connerley (1997), Connerley andRynes (1997), Harris and Fink (1987), and Turban and Dougherty (1992) foundno evidence that recruiter gender was related to applicant reactions.

    Instead of looking directly at recruiter demographic variables, a few studieshave examined whether being similar to the recruiter on certain dimensions mightbe important to job applicants. For example, Turban and Dougherty (1992)studied whether applicants and recruiters being of the same gender or from thesame university might have a beneficial effect on applicants. They found that bothtypes of similarity were related to job attractiveness. In contrast, Cable and Judge(1996) did not find demographic similarity (e.g., age, race, gender) betweenapplicants and recruiters to be associated with such variables as job choiceintention.

    423RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • In trying to understand why certain recruiters may have a more positiveeffect on applicants than other recruiters, researchers have looked at several otherpossible explanatory variables. With regard to recruiter experience and training,results have been mixed. For example, Connerley and Rynes (1997) reported thatmore experienced recruiters were viewed more positively by job applicants. Incontrast, Connerley (1997) found no effect for experience. In general, recruitertraining has not been linked to applicant reactions (e.g., Connerley, 1997; Stevens,1998). Researchers also have investigated whether a recruiters functional area(e.g., personnel department vs. a line function) made a difference. Connerley(1997) and Taylor and Bergmann (1987) reported that the use of line managers asrecruiters had a positive influence on applicants. In contrast, Harris and Fink(1987) and Connerley and Rynes (1997) did not find a recruiters functional areato make a difference.

    In trying to make sense out of research on recruiters, it is useful to considerthe reasons that have been offered by researchers for why recruiters might havean effect on job candidates. Recruiters have been hypothesized to be importantbecause they serve a key role in directly communicating information about aposition (e.g., types of duties) and an organization (e.g., its values) to applicants.In order to carry out this communication role effectively, a recruiter needs to haveboth the ability to communicate effectively (e.g., sufficient time) and personalcredibility. In addition to intentionally conveying information to a job candidate,a recruiter may also have an impact on applicants for unintentional reasons. It hasbeen argued (Rynes, 1991) that applicants often view a recruiter as a signal ofunknown aspects of the organization. Concerning recruiters being used as asignal, Rynes, Bretz, and Gerhart (1991) reported evidence of signaling. Inparticular, they reported that applicants were likely to view the recruiter as beingsymbolic of broader organizational characteristics (p. 487) when applicantslacked detailed knowledge of the hiring organization and when the recruiter wasfrom their functional area. This signaling effect occurred both during the initialcampus interview and during site visits.

    Although there is considerable evidence that recruiters who are perceived asbeing more personable and more informative are associated with beneficialoutcomes for the organization (e.g., applicants reporting a higher likelihood ofaccepting a job offer), the results of some studies have not found such associa-tions. Furthermore, in many studies these associations have been modest inmagnitude. There are a number of possible explanations for inconsistent findingsand for the modest magnitude of some of the results reported (e.g., recruitersfrequently do not have sufficient time to present all of the information they wouldlike to). Although space limitations do not allow for a discussion of potentiallimiting factors, the topic of recruiter credibility does merit some attention.

    Research on recruiter credibility has shown that job incumbents are per-ceived as a more credible source of information than are full-time corporaterecruiters (Coleman & Irving, 1997; Fisher et al., 1979). If one considers the twofactors (expertise and trustworthiness) that are seen as resulting in perceptions ofcredibility, this finding is not surprising. During the recruitment process, appli-cants want specific information about several job-related issues. Recruitment

    424 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • research (e.g., Maurer et al., 1992) regularly has shown that applicants report nothaving received as much job-related information as they would like. If oneconsiders the issue of expertise, the reason for this lack of information is apparent.Frequently, recruiters are not very familiar with the job and the job environment.One can imagine the difficulty that a full-time recruiter would have trying toanswer questions about a particular supervisors style or a work groups climate.A few researchers (e.g., Harris & Fink, 1987) have investigated whether inter-acting with a recruiter from the candidates functional area as opposed to one fromthe human resource department makes a difference. Although such a distinctionmay be worth examining, a better line of inquiry may be whether the person withwhom a job candidate interacts is the candidates potential supervisor or coworker(e.g., a person in ones functional area may know little about important aspects ofa particular job). One would expect that such a supervisor or coworker would beseen as having much more expertise than other types of recruiters.

    It is not sufficient for a person to have expertise to be seen as a crediblesource; the person also needs to be seen as trustworthy. Research (e.g., Coleman& Irving, 1997) has shown that recruiters from a human resource department areseen as less trustworthy than job incumbents. Given that full-time recruiters maybe evaluated in terms of the number of individuals they bring into the selectionprocess (Breaugh, 1992), it is not surprising that applicants may sometimes seethem as presenting an overly positive view of a job opening. In contrast, potentialcoworkers may be seen as being more trustworthy because they will have tointeract with the job candidate if the person is hired. From this brief discussion ofrecruiter credibility, the need for future research that examines the influence ofwhether the recruiter is ones potential supervisor or coworker should be apparent.It is likely that recruiter effects would be stronger if distinctions were madebetween potential supervisors and coworkers versus full-time corporate recruiters.

    Over time, recruiter-oriented research has become more complex and moremethodologically sound. For example, early studies of recruiters (e.g., Alderfer &McCord, 1970) often involved simply gathering survey data from college studentsimmediately following a campus interview. More recently, research on recruitershas involved: (a) gathering data both before and after interviews (e.g., Harris &Fink, 1987), (b) gathering longitudinal data over various stages of the recruitmentprocess (e.g., Taylor & Bergmann, 1987), (c) gathering data on recruiter behaviorfrom both recruiters and job applicants (e.g., Connerley & Rynes, 1997), and (d)more intensively studying recruitment interactions by means of in-depth inter-views (e.g., Rynes et al., 1991) and the audiotaping of applicant-recruiter inter-actions (e.g., Stevens, 1998). This increasing sophistication of research hasresulted in a better understanding of recruiters and their potential effects. Forexample, in the 1980s, some researchers questioned whether recruiters had animportant effect on job applicants (Rynes, 1991). Today, this view is lesscommonly held (Barber, 1998). For instance, in their intensive study of therecruitment process, Rynes et al. (1991) found that recruiters had a powerfulimpact. In a number of cases, poor recruiters resulted in individuals losinginterest in jobs.

    425RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • In terms of future research directions, clearly the trend toward more complexresearch should continue. As others have suggested (e.g., Barber, 1998), we toobelieve that relatively less attention should be given to recruitment on collegecampuses and to the initial stage of the recruitment process. In the future,researchers need to explore the influence of recruiters at job fairs or during sitevisits (Turban, Campion, & Eyring, 1995 have presented evidence that the host ofa site visit can influence job applicant perceptions of job attractiveness and theirlikelihood of accepting a job offer).Research on Realistic Job Previews

    Over the last thirty years, realistic job previews (RJPs) have attractedsubstantial attention. As described earlier, several different RJP models have beenoffered. These models suggest that providing realistic information to job candi-dates should have beneficial effects for a number of reasons (e.g., RJPs allowcandidates to self-select out of consideration for jobs that would not meet theirexpectations, RJPs cause individuals to be more committed to their job acceptancedecision). To provide a sense of the research on RJPs, we will review a few typicalstudies.

    In one of the early RJP studies, Wanous (1973) examined the effects ofapplicants for the job of telephone operator viewing either a traditional recruit-ment film or an experimental film that provided both positive and negativeinformation about this position. He found that viewing the RJP film had no effecton job offer acceptance rate (78 of the 80 individuals in his study accepted offers),but did result in lower job expectations. Wanous reported there was no RJP effectfor turnover, but that receiving an RJP did positively affect job satisfaction.

    RJPs generally have been provided by means of a booklet or a video(Phillips, 1998). Colarelli (1984) felt that an interview with a job incumbent mightbe a more powerful RJP medium given that such a two-way conversation may (a)increase an applicants attention and comprehension, (b) enhance the amount ofpersonally relevant information conveyed, and (c) increase the importance at-tached to the information conveyed as a result of job incumbents being a highlycredible source of information. Colarellis study involved applicants for theposition of bank teller. His research design involved comparing a control groupagainst individuals who received an RJP via a brochure or through an interviewwith a teller or a teller trainer. Given that everyone offered a job accepted it,Colarelli found no evidence of self-selection. He did find that the incumbent grouphad a lower turnover rate than the other two groups. With regard to the realism ofjob expectations, Colarelli reported that both of the RJP groups felt they had morerealistic expectations than those in the control group. In terms of the personalrelevance of the information received, the incumbent group had higher scores thanthe brochure group, which had higher scores than the control group. Colarelli didnot find an RJP effect for job satisfaction. Unfortunately, although Colarellihypothesized that receiving an RJP from a job incumbent would increase appli-cant attention and comprehension, he did not measure these variables.

    In 1985, two RJP meta-analyses were published. McEvoy and Cascio (1985)compared the effects on turnover of providing an RJP versus using a job enrich-

    426 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • ment strategy. They reported the average phi coefficient between turnover and theuse of an RJP was .09. They also reported that RJPs had a stronger effect onturnover for more complex jobs. Premack and Wanous (1985) reported a similarphi coefficent (.06) between the use of an RJP and turnover to that reported byMcEvoy and Cascio. Premack and Wanous also examined RJP effects on severalother variables (e.g., job satisfaction, performance, initial expectations). For themost part, the RJP effects they reported were in the expected direction but weresmall in magnitude.

    Despite the somewhat pessimistic conclusions of some authors (e.g., McE-voy & Cascio, 1985) concerning RJP effects, RJPs have continued to attractattention. During the 1990s, researchers examined such issues as (a) whether theRJP medium (written vs. oral) makes a difference (Saks & Cronshaw, 1990), (b)whether having prior exposure to a job moderates the impact of receiving an RJP(Meglino, Ravlin, & DeNisi, 1997), and (c) whether RJPs make higher qualityapplicants less likely to pursue job openings (Bretz & Judge, 1998). Although itis not possible for us to provide a review of all of the recent RJP studies, a recentpaper by Phillips (1998), which includes a meta-analysis of empirical research,provides an excellent summary of the current state of RJP research. Overall,Phillips found that RJPs were related to higher job performance and to lowerlevels of initial job expectation, attrition from the recruitment process, andvoluntary turnover. Phillips also reported that RJP effects were stronger when theRJP was provided verbally rather than via a videotape or a booklet. Phillipsemphasized that many of the RJP effects were quite modest in magnitude.

    A review of Phillips paper (Phillips, 1998) may help explain some of theconfusion in the RJP literature. For example, in almost one-half of the studiesincluded in her meta-analysis, the RJP was provided after an individual had beenhired. Given that RJPs have been hypothesized to work because they increaseself-selection, provide a sense the employer is trustworthy, and increase commit-ment to the job choice decision, providing an RJP after hiring should lessen itseffect (e.g., having already accepted a job offer, a person may not see theorganization as being particularly trustworthy and may not feel committed to theuninformed decision he or she made). RJPs also are thought to be most effectivewhen an applicant has unrealistic expectations and feels capable of turning downa job opportunity that is not seen as desirable. Given the visibility of several of thejobs (e.g., bank teller, grocery store checker, gas station cashier) that have beenaddressed in RJP studies (see Phillips, 1998: Table 1), one has to consider thedegree to which the job applicants in a number of these studies had unrealistic jobexpectations. In a similar vein, it appears that economic conditions may havemade self-selection unlikely in some RJP studies. For example, 78 of the 80individuals offered jobs in Wanous study (Wanous, 1973) accepted them. Wa-nous noted that a difficult labor market probably reduced an individuals freedomto reject a job offer on the basis of a preview (p. 331).

    Another possible reason for some of the modest RJP effects that have beenreported is the nature of the job previews that have been provided. Research (e.g.,Barber & Roehling, 1993) suggests that applicants want specific informationabout job and organizational attributes. Dean and Wanous (1984) pointed out that

    427RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • the RJPs provided in many studies did not involve the communication of specificRJP information.

    In summary, there are several potential explanations for the modest RJPeffects that have been reported. One can only imagine the potential effects RJPsmight have if future studies involve situations in which RJPs (a) provide specificinformation about a wide range of important job attributes, (b) are conveyed bya credible source such as a future coworker, and (c) are targeted at applicants (notnew employees) who lack realistic job information and have other viable jobopportunities.

    Recently, Meglino et al. (1997) and Bretz and Judge (1998) have raised theissue of whether providing realistic job preview information may hurt an employ-ers chances of hiring desirable job applicants. For example, based upon their ownresearch and a meta-analysis of other relevant research, Meglino et al. (1997)concluded that providing realistic information to applicants who had prior expo-sure to that type of position resulted in their overestimating the negative aspectsof the job. This overestimation could result in individuals with prior job experi-ence, individuals who are frequently viewed as desirable job candidates, rejectingjob offers. Bretz and Judge (1998) also suggested that higher quality job candi-dates may be less willing to pursue jobs for which negative information has beenprovided. They provided some evidence that supports this position. Althoughneither set of authors recommended that RJP information should not be provided,others may come to this conclusion from their research. In a recent paper,Buckley, Fedor, Carraher, Frink, and Marvin (1997) took a very different position.These authors argued that an employer has an ethical imperative to providerecruits with realistic job previews (p. 468).

    One of the reasons that RJPs are thought to have positive effects on turnoverand satisfaction is because they lower job expectations. In a recent paper,Buckley, Fedor, Veres, Wiese, and Carraher (1998) advocated the wider use ofRJPs but also noted that creating an RJP can be an expensive process. Given thisfact, they experimented with an alternative expectation lowering procedure (ELP).In their study, an ELP involved an individual participating in a workshop that (a)stressed the fact that individuals tend to have unrealistically high job expectations,(b) addressed the likelihood that such expectations are likely to be unfulfilled, and(c) discussed the common outcomes attached to unfulfilled expectations. To testthe impact of receiving an ELP, Buckley et al. conducted an experiment withnewly hired employees who had yet to start work. These individuals were dividedinto four groups. One group went through the traditional orientation program. Asecond group received a written RJP that was based upon a detailed job analysis.A third group was composed of those who participated in the ELP workshop. Thefourth group was a control group. Buckley et al. found that the RJP and the ELPwere equally effective in lowering job expectations. They reported a strongassociation between expectation level and subsequent employee turnover. In ouropinion, ELPs merit future research as a potential low-cost alternative to RJPs.

    Given that many new employees appear to have incorrect expectationsconcerning what their new jobs entail (Wanous, 1992), we believe that a realisticjob preview represents an important recruitment mechanism. However, we think

    428 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • there may be value in focusing less on an RJP as a specific event (e.g., showinga 10-minute video) and instead focusing more on the process of providingaccurate job information at several points during the recruitment process. Forexample, an employer could begin to provide realistic information with its jobadvertisement. Following this, it could increase the amount of information con-veyed during the initial interview by selecting recruiters who are likely to reallyunderstand the job situation. Additional realistic job information could be com-municated by a visit to the work site and/or through applicants taking part in awork sample (Wanous, 1992). By viewing realistic recruitment as a process, anemployer may more thoughtfully plan out recruitment events. Trying to providerealistic job information at a series of points via different mechanisms may alsolessen the likelihood of a recruits becoming overloaded with information (Barber,1998).

    In summary, we believe research on RJPs has resulted in a better under-standing of the consequences of hiring individuals who have realistic expecta-tions. Our hope is that in future years researchers will focus more attention on theconditions in which RJPs are utilized (e.g., Did job candidates lack accurate jobinformation?) and on the psychological variables (e.g., the credibility of themessage) that are hypothesized to underlie RJP effectiveness.Other Recruitment Topics

    Recruitment advertising. Although firms spend considerable money on jobadvertising, little research has addressed this topic. Previously, we reviewedresearch (e.g., Barber & Roehling, 1993) that documented the positive effects ofproviding a recruitment message that conveyed both more information and morespecific information. Other research dealing with recruitment ads has also shownthe benefits of their including specific information (Mason & Belt, 1986) andmore information (Yuce & Highhouse, 1998). Recently, Highhouse, Beadle,Gallo, and Miller (1998) examined whether the wording of an advertisement fora restaurant position affected college students perceptions of the estimated hourlypay, the companys image, and their job pursuit intentions. The only differenceHighhouse et al. reported was a position scarcity effect on pay (e.g., when only afew jobs were advertised, pay was estimated to be over $1.70 more than if manyopenings were advertised). The results of Highhouse et al.s study suggest that jobapplicants may infer certain information (i.e., pay rate) from the wording of an ad.It is important for future research to investigate what other inferences may bedrawn from recruitment ads.

    The site visit. To date, little research has addressed the importance of thesite visit. The findings of the studies that have are mixed. For example, Taylor andBergmann (1987) reported that, after entering expected job attributes as predictorvariables in regression analysis, the site visit (e.g., how well recruits were treated)had no effect on applicants ratings of company attractiveness or of their likeli-hood of accepting job offers. In contrast, Rynes, Bretz, and Gerhart (1991) foundthat for their sample the site visit played a significant role in the recruitmentprocess. In particular, job seekers noted the importance they attached to the statusof the people they met, whether they felt specially treated, and the organiza-

    429RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • tions professionalism during their visit. Turban, Campion, and Eyring (1995)reported that various ratings of the site visit (e.g., host likeableness) werepredictive of the recruits ultimate decision of whether to accept a job offer. Aninteresting feature of this study was the fact that the site visit host would be acoworker of the recruit if he or she were hired. As noted by Barber (1998), incomparison to the issue of generating job applicants, the topic of maintainingapplicant interest has attracted relatively little research attention. Given that a sitevisit may have an important influence on whether job applicants remain interestedin a position, we believe the site visit deserves more research attention.

    Timing issues. Although being timely in terms of recruitment actions isgenerally viewed as being important (Barber, 1998), only a few studies haveaddressed timing issues. With regard to the timing of recruitment actions, Rynes,Bretz, and Gerhart (1991) reported that for their sample: (a) long delays betweenvarious phases of the recruitment process were common, (b) job applicantsfrequently drew negative inferences about the delays (e.g., I am not their favoritecandidate), (c) delays affected the willingness of individuals to accept job offers,and (d) the most marketable job seekers frequently saw delays as reflectingsomething was wrong with the organization. In contrast to the results reported byRynes et al. (1991), Taylor and Bergmann (1987) reported no timing effect fortheir study (conceivably, large sample attrition in this study could have affectedresults). Given the limited amount of research that has been conducted, it ispremature to attach too much significance to the timing of recruitment actions.However, this appears to be yet another area that is ripe for research.

    Concluding Remarks

    Over the last three decades, the amount of research on recruitment topics hasincreased substantially. In spite of this increase, recent reviews of the recruitmentliterature (e.g., Barber, 1998; Rynes, 1991) have had a somewhat pessimistic tone.For example, reviewers frequently have concluded that we still do not know agreat deal about why recruitment activities (e.g., using certain recruitmentsources) have the effects they do. In particular, reviewers have criticized manyrecruitment studies for being poorly designed, narrow in focus, and not groundedin theory. Based upon our review of recruitment research, we think that thesecriticisms are valid. We also believe that, in order for future research to lead to abetter understanding of recruitment issues, studies need to be designed with anappreciation of the complexity of the recruitment process.

    In order to stimulate the type of research that has been called for in previousreviews, we presented an organizing framework of the recruitment process. Inintroducing this framework, we gave particular attention to the process variables(e.g., applicant attention, self-insight) that theory suggests should mediate therelationships between recruitment activities and recruitment outcomes. In review-ing the empirical recruitment literature, we highlighted how often these processvariables have been ignored.

    Given Barbers recent review of recruitment research (Barber, 1998), wewere able to be selective in reviewing research. In selecting studies to review, we

    430 J.A. BREAUGH AND M. STARKE

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • tried to choose studies that (a) gave the reader a sense of the typical study that hasbeen conducted (e.g., Wanous paper [Wanous, 1973] on realistic job previews),(b) called into question conventional thinking (e.g., Greenhaus, Seidel, & Mari-niss paper [Greenhaus et al., 1983] on the issue of value attainment), (c)highlighted a better research design for investigating an issue (e.g., the focus byWilliams et al. [1993] upon recruiting source effects on job applicants), or (d)suggested new streams of research.

    Although one can view the existing recruitment literature as a hodgepodge ofinconsistent results, we feel there are certain themes that emerge. For example,research on recruiters makes apparent the critical role these individuals can playboth in terms of being informative and in terms of treating applicants in apersonable fashion. These themes of informativeness and personable treatmentalso permeate other areas of recruitment research, such as research on the site visitand the timing of recruitment actions. The importance of providing realistic jobinformation (e.g., via employee referrals or realistic job previews) has also beenan important research theme in the recruitment literature. Another theme thatemerged in our review of the literature is the importance of signals thatemployers may unintentionally be sending to job applicants. For example, Ryneset al. (1991) showed that applicants used information with regard to recruiterfriendliness and informativeness as an indicator of how a firm treated an em-ployee.

    Although our review frequently criticized past research, our objective indoing this was not to chastise previous researchers. Rather, our intent was to usethe deficiencies of past research to stimulate future research that is not subject tothese criticisms. Having critiqued past research, we should acknowledge thedifficulty of doing research on recruitment topics. To begin with, it is difficult todefine the term recruitment and to distinguish what falls within the definition. Forexample, we began this paper by citing Barbers definition of recruitment (re-cruitment includes those practices and activities carried on by the organizationwith the primary purpose of identifying and attracting potential employees;Barber, 1998: 5). This definition says nothing about recruiting individuals who arelikely to be successful on the job. Some would argue that the notion of successshould be part of this definition. In a similar vein, consider an employer that usesa work sample as a selection device. If going through a work sample provides ajob candidate with realistic information about a position, should the work samplebe considered a selection device, or recruitment device, or both? The topic oforganizational image also blurs the boundary between recruitment and othertopics. For example, Turban, Forret, and Hendrickson (1998) found that anorganizations reputation influenced job applicants. We chose not to review theorganizational image literature given that an employers reputation frequently isnot directly a function of recruitment activities (e.g., media accounts of businesssuccesses or failures are more likely to affect a companys image than are jobadvertisements). However, an organizations image can clearly influence organi-zational recruitment.

    In closing, we choose to end on an optimistic note. In our opinion, this is anexciting time for recruitment researchers. The difficulty in filling jobs has focused

    431RESEARCH ON EMPLOYEE RECRUITMENT

    JOURNAL OF MANAGEMENT, VOL. 26, NO. 3, 2000

  • organizational attention on the importance of well-designed recruitment activities.Recent research has been more theory-based and more methodologically sophis-ticated (e.g., Stevens [1997] used audiotapes to better understand recruiter/applicant interactions; Barber & Roehling [1993] used verbal protocol analysis).The continued use of such methods should prove fruitful in investigating the manyunanswered questions that surround the recruitment process.

    ReferencesAlderfer, C. P., & McCord, C. G. 1970. Personal and situational factors in the recruitment interview. Journal of

    Applied Psychology, 54: 377385.Barber, A. E. 1998. Recruiting employees. Thousand Oaks, CA: Sage Publications.Barber, A. E., & Roehling, M. V. 1993. Job posting and the decision to interview: A verbal protocol analysis.

    Journal of Applied Psychology, 78: 845856.Blau, G. 1990. Exploring the mediating mechanisms affecting the relationship of recruitment source to employee

    performance. Journal of Vocational Behavior, 37: 303320.Breaugh, J. A. 1992. Recruitment: Science and practice. Boston: PWS-Kent.Breaugh, J. A., & Billings, R. S. 1988. The realistic job preview: Five key elements and their importance for

    research and practice. Journal of Business and Psychology, 2: 291305.Breaugh, J. A., & Mann, R. B. 1984. Recruiting source effects: A test of two alternative explanations. Journal

    of Occupational Psychology, 57: 261267.Bretz, R. D., & Judge,