You are on page 1of 22

Explaining the Effects of Transformational Leadership: An Investigation of the Effects of Higher-Order Motives in Multilevel Marketing Organizations Author(s): John

R. Sparks and Joseph A. Schenk Source: Journal of Organizational Behavior, Vol. 22, No. 8 (Dec., 2001), pp. 849-869 Published by: Wiley Stable URL: http://www.jstor.org/stable/3649575 . Accessed: 18/01/2014 06:54
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wiley and John Wiley & Sons are collaborating with JSTOR to digitize, preserve and extend access to Journal of Organizational Behavior.

http://www.jstor.org

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

Journal of OrganizationalBehavior J. Organiz.Behav. 22, 849-869 (2001) DOI: 10.1002/job.116

Explaining the effects of transformational leadership: an investigation of the effects of higher-order motives in multilevel marketing organizations
JOHN R. SPARKS* AND JOSEPH A. SCHENK
Universityof Dayton, Dayton, U.S.A.

Summary

Multilevelmarketing type enlisting (MLMs)area rapidlygrowingorganizational organizations over 20 billion dollarsin sales annually. Despitetheir nearly 10 million membersandproducing andnone of remarkable recentgrowth,few studieshave examinedthese unusualorganizations, relathe these have addressedissues of transformational In MLMs, leadership key leadership. and the members who recruited tionships are those between individualmember distributors them into the organization(i.e., their 'sponsors').Although sponsorsare expected to provide leadershipto the membersthey recruit,they possess no directsupervisory resulting--authority in an uncertain'quasi-leadership' role. Using a sampleof 736 female MLM members,the present studyempiricallytests an important leadership componentof transformational explanatory theory:thatbelief in the higherpurposeof one's work is a mechanismthroughwhich transformationalleadershipachievesits positive outcomeson cohesion, satisfaction, effort, andperformance. The results offer support to the notion that transformationalleadership indeed 'transforms' followers by encouragingthem to see the higherpurposesin theirwork.Additionally, the results show positive relationshipsbetween belief in a higher purposeof one's work andjob satisfaction,unit cohesion, and effort. Copyright? 2001 John Wiley & Sons, Ltd.

Introduction
Multi-level marketing organizations (MLMs) have long been an active part of the American business landscape. These networks of member-distributors, who earn income both from selling retail products and recruiting new members, are currently enjoying remarkable growth. According to the Direct Selling Association (1999), from 1991 to 1998, MLM participation grew from 5.1 million to 9.7 million members (75 per cent of whom are women) while annual sales grew from 13 billion to nearly 23 billion dollars. Traditionally limited to household cleaners (e.g., Amway) and cosmetics (e.g., Mary Kay), MLMs today sell such diverse products as jewellery, nutritional supplements, children's toys, life insurance, and long distance telephone services (King, 1996; Salter, 1997). Moreover, because MLMs * Correspondence OH 300College R. Sparks, to:John of Dayton, Park, Dayton, Department, University Management/Marketing U.S.A. 45469-2235, E-mail: sparks@udayton.edu Received 10 June 2000 Revised 1 September2000 Accepted 20 June 2001 Published online 25 October2001

Copyright? 2001 John Wiley & Sons, Ltd.

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

850

J. R. SPARKSAND J. A. SCHENK

can rapidly and inexpensively create large distributionnetworks, these trends seem likely to continue-strongly suggesting that an examinationof this organizationalphenomenonis long overdue. As we discuss in the following section, MLMs combine attributesof traditionalindustrialselling that pose complex with small retail entreprenuership and operateusing fluid organizationalstructures individual MLM between Much of the this in exists leadership problems. relationships complexity members and the members who recruitedthem into their respective organizations(i.e., their 'sponsors'). Although membersfrequentlycite a productivesponsor-recruitrelationshipas pivotal to operating a successful MLM distributorship,the nature of a sponsor's contributionto that success is unclear.On the one hand, sponsorsreceive significantincentives to serve as leaders to their recruits, guiding and supportingrecruits' efforts in ways similar to a sales manager.On the other hand, once therefore,sponsors lack the recruited,new membersindependentlyown their MLM distributorships; formal supervisoryauthoritythat typically accompanies leadershippositions in traditionalorganizations. Collectively, these circumstancesleave the characteristicsof effective leadershipin MLMs an open question.Therefore,we begin our study of MLMs by examining the natureand effects of transformationalleadershipin this largely unexploredorganizationalcontext. As the name implies, transformational leaders ostensibly 'transform'their followers throughspecific behaviorsthat inspire and motivatefollowers to 'transcendtheir own self-interestsfor the good of the group, organization,or society' (Bass, 1990, p. 53), resultingin followers doing 'more than they intendedand often more thanthey even thoughtpossible' (Bass 1998, p. 4; see also Bass, 1985, 1997; Bass et al., 1987; Burns, 1978; Podsakoffet al., 1996; Yukl, 1998). Althoughthe literature consistently supports transformational leadership's positive impact on attitude and performance,less is known abouthow this leadershipstyle actually achieves the transformation of followers. Leadershipscholars that followers' when transformations occur propose they begin associating work with 'higher that extend purposes' beyond simply earning money (Bass, 1985; Burns, 1978; Podsakoff et al., Shamir et 1996; al., 1993; Yukl, 1998). Because most empiricalresearchhas focused on whetherand underwhat circumstancestransformational leadershipproduces more satisfied, more committed, and higher performingindividuals,few studies in any context have investigatedthe mechanismsby which these outcomes occur. Moreover, we are aware of no studies investigating these issues in MLMs. Consequently,the purpose of this researchis to examine:(1) whethertransformational leadershipbehaviorsby an MLM sponsorinvokes their recruits'higher-order motives, and (2) the degree to which those motives serve as a mechanism which transformational through leadershipbrings about positive effects on the recruits' effort, satisand faction, performance.Specifically,we develop and test a model in which one general higher-order 'belief in a higherwork purpose,'mediatesthe causal sequence leading from transformational motive, to likely have little experiencewith leadership satisfactionandperformance.Because most researchers reasons for selecting one as a our MLMs, we first briefly describe their structureand operation, and researchsetting.

The Nature and Structure of MLMS


Simply put, MLMs consist of corporate-level producers that market branded goods or services (e.g., Mary Kay Cosmetics, Amway Products)throughnetworksof independent,member-owneddissimwhich areusually operatedfrom members'homes. Startingan MLMdistributorship tributorships, ply requires purchasingtraining and promotional materials from the corporate-levelproducer and agreeingto purchasea certainamountof inventoryduringa specifiedperiod.Membersthen resell their inventoriesprimarilyto retail customers,often beginning with family and friends. Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MARKETINGORGANIZATIONS MULTILEVEL

851

MLM distributorships provide memberswith two sources of income. The firstcomes from product sales. Relying on the same basic skills as outside industrialsalespeople, successful MLM members identify and qualify sales prospects, contact them, arrangeto meet them and present the benefits of the good or service, then close the sale. To grow theirdistribution networks,corporate-levelproducers offer membersa second source of income: commissions on sales by new membersthat currentmembers recruit.To enhance this recruitingincentive, most MLMs pay commissions on multiple levels of recruits-hence, the term multi-level marketing.That is, membersreceive commissions not only on sales generatedby new membersthey recruit,but also on sales by memberstheir recruitsbringinto the organization,all recruitsthose membersbring in and so on. Thus, MLM memberscould conceivably receive commissions from sales by hundredsof other members. Although MLM members own independenthome-based businesses, they do not operate them in isolation from one another.In fact, MLMs establishelaboratesystems of supportthroughwhich members receive guidanceand leadershipfrom each other.Two componentsof these organizational support systems play crucial roles in the working relationshipsamong members. First and foremost are the relationshipsbetween 'sponsors' and 'recruits.'A sponsor is an existing member who persuades a new member,or recruit,to starta distributorship. Technically,all MLM members are recruitsin the sense that some currentmember persuadesevery new member to join. Thus, any member who has recruitedat least one other memberis at once a sponsorand a recruit.Ouruse of the terms 'sponsor' and 'recruit,'however,shouldbe thoughtof as dyadic. Thatis, we referto the relationshipbetween two members-one who recruitedanother.Based on extensive interviewswith MLM members,we regard sponsor-recruitdyads to be the key leader-follower relationshipsin MLMs. Indeed,the commissions sponsorsreceive on theirrecruits'sales serve as a powerful incentive to become active sources of leadershipto their recruits.Given the centralrole sponsor-recruitrelationshipsplay in the operationsof MLMs, they are the focal point for our study of transformational leadership. A second factor influencing the natureof members' relationshipsare informal working groups of MLM memberscreatedby sponsorsfor theirrecruits-a kind of sales team referredto here as a 'unit.' Throughunits, membersmeet regularlyto exchange informationand ideas (e.g., selling tips), create shared opportunities(e.g., sales workshops), and provide very practicalforms of assistance to each other (e.g., loaning inventoryand materials).Although membersoperateindependently,units provide a tangible organizationthroughwhich memberscreate and maintaina sense of belonging, develop a Moreover, feeling of esprit de corps, and acquireresourceshelpful to their individualdistributorships. units serve as a vehicle for socializing membersinto the norms and values of the MLM and a conduit through which members can mutually reinforce the effects of sponsors' leadership behaviors (Waldmanand Yammarino,1999; Yammarinoand Dubinsky, 1992). Studyingtransformational leadershipin this organizationalcontext is importantfor severalreasons. transformational First, leadership can be examined in relative isolation from other leadershipstyles in commonly employed traditionalorganizations,largely because sponsors may have few effective alternativesto transformational leadership.Members' independentownershipof their distributorships leaves MLMs without the hierarchical leader-subordinate relationships found in most selling organizations.The resulting leader-follower dyads provide sponsors strong incentives to cultivate high-performingmembers, but few extrinsic tools to deepen the satisfaction and loyalty of their recruits (see Dansereau et al., 1975; Graen and Cashman, 1975). In other words, sponsors are not managers or supervisors, and although corporate-level producers provide incentives encouraging sponsorsto serve as leadersto theirrecruits,sponsorshold none of the formalleverage ordinarilyheld by sales managers.Sponsorscannotpromote,discipline, or dismiss recruits;they cannot alterrecruits' compensation;they cannotdirectrecruits'daily work activities.The absenceof the normalsupervisory tools that typically accompany leadership positions therefore limits sponsors' choice of leadership styles. Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

852

J. R. SPARKSAND J. A. SCHENK

Second, although some membersjoin MLMs primarilyfor additionalincome, many connect their MLM participation or financial to a varietyof 'higher-order values' that extend beyond profit-making success. Membersexpressedthese sentimentsabouttheirwork not only duringour interviews,but also in articles from the popular business press (Bartlett, 1995; Dunkel, 1995) and academic research (Wotrubaand Tyagi, 1992). For example, some members identify so strongly with the benefits of productsthey representthat selling them assumes a special importance: e.g., educationaltoys advance the cause of children; weight managementprogrammesenhance self-confidence, and so on. Other members view MLMs as a way to strengthenfamily ties by working from home. Still others believe that MLMs are a means for women to control their own professional destinies throughentrepreneurship, opportunities many view as less availablethroughtraditional corporateavenues.As noted earlier, inasmuchas transformational followers behaviors 'transform' by appealingto these higher leadership ordervalues, MLMs offer a rich and favorableenvironmentfor exploring whethersuch appeals arise from or are enhancedby a sponsor's transformational leadership. context in which transformaFinally MLMsrepresenta growingyet underresearched organizational tional leadershipmay play a particularly and in role predictingperformance.To explaining important these this is the issues, investigate study guided by following questions:(1) Are memberswho express in beliefs the MLM of their stronger higher purpose participationmore likely to have sponsors who exhibit a transformational Are these membersmore satisfiedwith theirwork, and leadershipstyle? (2) do they work from within more cohesive and supportiveunits? (3) Do these members subsequently exert greatereffort?And (4) do they in turnachieve superiorlevels of performance? In the following a model that describes we examine these and sections, questions by developing empirically testing series of mediating a transformational effects on satisfaction and through leadership's job performance work and in introduces a a variable-belief purpose-through relationships, higher key mediating which transformational leadershipachieves its positive outcomes.

Model Development
Rationalefor model
leadership's effects on Figure 1 shows the hypothesized model, which posits that transformational satisfaction and performanceis mediated by a series of intervening variables including belief in a higher work purpose. Two reasons prompt our focus on mediation effects. First, examination of variables that may mediate leadership-outcome relationshipsresponds to a need for such research identified in the literature(see Conger, 1999; Shamiret al., 1993; Yukl, 1999). Indeed, Yukl (1999, p. 287) refers specifically to ambiguity about the underlying processes through which transformational leadershipworks and proposes more researchon 'mediating variablesrelevantto task performance' including, among other things, 'arousal of motives.' In their importanttheoretical work, Shamiret al. (1993) likewise propose thattransformational leaderships'effects occur througha series of mediating 'motivationalmechanisms' that relate primarilyto follower self-concept. Although the focus on mediation adds complexity to the model, these calls for examination of mediating relationships support the notion that transformationalleadership's effects may in reality be quite complex. Second, our model permits testing a hypothesis central to transformational leadershiptheory. As noted earlier,by definition,transformational leadershipbehaviorsare believed to motivate followers to forgo theirown needs for the sake of the organization.Leadershipscholarshave variouslydescribed the process throughwhich transformational leadershipworks as 'raising' followers 'to higherlevels of Copyright Wiley& Sons,Ltd. (? 2001 John Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

ORGANIZATIONS MARKETING 853 MULTILEVEL

AVI AV2
Articulating
a

JS2
Vision

JS2

JS

JSI

AV3

.98

Job

Sat

-AM1
8 AM2 E~M~\Model

.113
Appropriate.22

AM

.92~
.92
AM

.5

.3

,
.80 Performance

T7Trans.
Lde

.32

fligher

.18

Effort.

Pflomac

Pfl

I8

HEI

HhHPI
Expectations

HE2

T18
.83

.16
.59

HP2

,
-.02

712pa
Ef2
Ef

Unit Cohesion
s

IS1
Individual

T4

IS2

Support

~IS3

T199UC

UCU

UC3

F, F,

modelandstructural modelparameter Figure1. Hypothesized estimates*'t fit statistics *n= 736, X2(239) = 672.82(p < 0.001),GFI= 0.93, AGFI = 0.91. Additional givenin Table3. shown. tAll pathssignificant to p < 0.01 except/32,4 pathcoefficients = n.s. Standardization IFormeasurement see Table1. modelcoefficients, needs' (Yukl, 1998; also Bass 1985; Podsakoff morality' (Bums, 1978), 'activatingtheir higher-order et al., 1996), and alteringtheir 'needs, values, preferences,and aspirations'by appealingto 'ideololeadership gical values' (Shamiret al., 1993). In effect, these conceptualizationsof transformational describe its effects as being mediated by the activationof these higher order needs. The model proposed here addressesthis most basic issue using, as discussed earlier,a sample we believe to be particularly well-suited to the task. leaderFigure 1 also points to anotherimportantfeature of the model, which is transformational that transformaa as second-orderfactor.This treatment ship's operationalization explicitly recognizes tional leadershipencompasses severaldistinct sets of behaviors,while acknowledgingthatthe totality of behaviors associated with transformational leadershipreflect some larger latent 'leadershipconstruct.' The high correlationsbetween transformational leadership's various dimensions reportedin numerousstudies (see Atwaterand Yammarino,1993; Bycio et al., 1995) indirectlysupportthis view, while direct supportcomes from Carless (1998) who tested three conceptualizationsof transformational leadership and found that a second-orderfactor model fit the data better than either a single factor model or a first-order multi-factormodel.

Research hypotheses
Effects of transformational leadership Subordinate job satisfactionranks among the most commonly studied outcomes of transformational leadership,with significantrelationshipsbeing found among office employees (Barling et al., 1996; Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

854

J. R. SPARKS ANDJ. A. SCHENK

Niehoff et al., 1990), school teachers (Koh et al., 1995), hospital nurses (Bycio et al., 1995; Medley and Larochelle, 1995), militarypersonnel(Yammarino and Bass, 1990; YuklandVanFleet, 1982), and et Given the salespeople (Dubinsky al., 1995). strengthof this evidence, we hypothesizea similarrelawill hold in MLMs: tionship Hypothesis 1: Sponsors' transformationalleadership will be positively related to recruits' job satisfaction. According to theory,transformational leadershipincreases followers' willingness to set aside individual needs in favor of group or organizationalneeds (Bass, 1985, 1997; Bass et al., 1987; Burns, 1978; Yukl, 1998). Shamiret al. (1993) explain this effect in terms of social identification;that is, transformational leadershipworks in large measurethrougha collective process by which followers' become self-concepts strongly intertwinedwith their membershipin a particulargroup. One likely result of this process is that memberswithin a given work group will experience greatercohesiveness (Matey, 1991; Waldmanand Yammarino,1999). In MLMs, the analogous work group is the informal unit describedearlier. Hypothesis2: Sponsors' transformational leadershipwill be positively relatedto recruits'perceptions of their unit's cohesion. A central feature to our model, the mediating variablebelief in the higher purposeof one's work, remains largely unexploredin the research literaturealthough similar concepts have recently been addressedby writersin the popularbusiness press (e.g., Dumaine, 1994). As noted earlier,MLMs provide numerousavenues throughwhich membersmay find higherpurposesbeyond earningmoney. For example, strengthening family ties by workingat home or selling a productwhose benefitshold special social importancemay be specific higher-orderpurposesthat motivate MLM members. However, at this point, our goal is not to single out any one higher order purpose motivatingmembers' actions. Instead we address the more basic question of whether higher order purposes in general motivate members and whether transformational leadershipper se encourages MLM members to find higher purposesin their work. To the extent that transformational leadershipis effective in 'giving meaningfulness to work and infusing work and organizationswith moralpurpose' (Shamiret al., 1993, p. 578), MLM recruitsshould find higher moral purposein their work when their sponsors exhibit behaviors associated with a transformational style. Hypothesis3: Sponsors'transformational leadershipis positively associatedwith recruits'belief in the higher purpose of their work. Effects of belief in a higher purpose We also expect recruits' beliefs in the higher purpose of their work to positively influence theirjob satisfactionand theirperceptionsof unit cohesiveness. With respect to job satisfaction,it seems intuitive thatstrongaffirmations of the intrinsicvalue of one's work, along any of the dimensionsdiscussed earlier,shouldproducetremendouspositive affect when the tasks associatedwith thatwork areperformed. Likewise, when membersof a work groupsharesimilarlystrongidentificationwith a higherwork purpose, cooperationand cohesiveness among group members contributeto the realization of those sponsors would believe purposes.Thus, we expect that MLM membersrecruitedby transformational more strongly in the higher purposeof their work, and to subsequentlydisplay higher levels of identification and cooperationwith other members of their units. Hypothesis4: Recruits'belief in the higherpurposeof theirwork is positively associatedwith their job satisfaction. Copyrightj 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MULTILEVEL MARKETINGORGANIZATIONS

855

Hypothesis5: Recruits'belief in the higherpurposeof theirwork is positively associatedwith their perceptionsof their units' cohesiveness. The model postulatesboth indirectand direct effects between higherpurposeand individualeffort. Theoretically, we could not reasonably expect that group cohesion would completely explain the relationshipbetween higherpurposeand effort. Given that likelihood, if higherpurposedoes in reality exert directeffects on effort, or if othervariablesmediatethatrelationship,or both, then we would still expect a direct significantand positive relationshipto exist between higher purpose and effort. Thus, we hypothesize: Hypothesis6: Recruits'belief in the higherpurposeof theirwork is positively associatedwith their work effort. With respect to higherpurpose'sindirecteffects on effort, one route may be throughunit cohesion. That is, as higher purpose increases unit cohesion, the unit's greater cohesiveness produces higher levels of individualefforton the partof its members.In supportof this view, Waldmanand Yammarino (1999) propose thatgreaterwork groupcohesiveness results in greaterindividualand collective effort (see also Lowe et al., 1996). Consequently,we make the following prediction: Hypothesis 7: Recruits' perceptionof the cohesion of their unit is positively associated with their work effort. Effects of effort Despite being focal variablesin a large body of empirical research, the natureof job satisfaction's relationshipwith individualeffort remainsunsettled.Therefore,we acknowledgethat, shoulda statistical relationshipbetween effort and satisfaction exist in our data, the causal direction may be the reverse of that which we hypothesize or the relationshipmay reflect a simple correlation.That said, we draw upon theoreticalwork on the intrinsic motivationof workersby Deci (1975) and Deci and Ryan (1985) and empirical work on the effort-satisfaction relationshipfrom Brown and Peterson (1994) to posit that greatereffort leads to enhancedjob satisfaction.Finally, we predict that greater effort will also produce superiorperformance. Hypothesis 8: Recruits' work effort is positively associated with theirjob satisfaction. Hypothesis 9: Recruits' work effort is positively associated with theirjob performance.

Contextual Sidebar

OrganizationalContext
Multilevel marketing nationally Multilevel marketingorganizations(MLMs) offer an inexpensive and attractivemeans for producers or wholesalers to quickly cultivateretail channels of distributionfor theirproducts,a fact that explains the remarkablegrowth in the numberof MLMs operatingworldwide. Similarly, MLMs offer members an attractiveand inexpensive means to start their own businesses. Members earn income by selling the producers'productsand by recruitingnew members to the MLM. Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

856

J. R. SPARKSAND J. A. SCHENK

Despite theirrapidgrowth,MLMs often attractcontroversyin partbecause they are associatedwith chain letters, pyramid schemes, and other fraudulentbusiness practices. What distinguishes true MLMs from classic pyramidschemes is whethermembers' earningscome primarilyfrom product sales to ultimate consumers instead of from recruitingnew members. Importantly,in legitimate MLMs, earnings generatedby recruitingmust be limited to commissions from recruits'sales and not from recruitingitself. MLMs featurean unusualand fluidorganizational with little formalhierarchydue to memstructure bers' independentownershipof their MLM distributorships. Older memberscan advise and assist the new membersthey recruit,but cannot direct and supervise the new members' work activities. Organizational factors The firmin this studyis relativelynew-less than25 yearsold, andis experiencingvery rapidgrowth in both sales and membership.The firm serves members in all 50 states and is developing distrithe MLM faces no directcompetitivethreatfrom otherMLMs, butorshipsinternationally. Currently, channel.However, this firm the market leaderfor its productline anddistribution making recognized the firmis facing new competitionfrom traditionalretailerswho appearto have a price advantage. Because members independently own their distributorships, the firm lacks a single identifiable on developing and maintaining climate.' focuses member socialization However, 'management enthusiasmfor selling the productline throughorganizational folklore aboutthe foundingmembers and other particularlysuccessful members. Stories of the company's founding and early growth providemembersa sense of sharedhistoryas well as an emotionaltool for buildingmemberloyalty and for recruitingnew members. Worker and job factors Because each memberis essentially an independentcontractor, theirjob responsibilitiestend to be Membershold control to their the amount devote of time broad,though they distributorships. they sale. Additioneach for all to of from sales responsibility aspects product locating prospects closing these and new members members receive new to recruit financial incentives help ally, powerful members develop successful distributorships of their own. Thus, in addition to managing their own distributorships, members assume the tasks of socializing, training, and motivating the new members they recruit.

Method
Memberinterviews
MLMs' unusualorganizationalcharacteristicsled us to begin with a qualitativeexplorationof issues related to leadershipand performanceby conducting a series of four group interviews at a mediumsized (in terms of total members)MLM's nationalconvention.The MLM corporateproducerselected focus group participantsfrom its list of convention registrantswith the goal of avoiding excessive Selected memberswere characteristics. homogeneity with respect to demographicand distributorship contacted by letter prior to the convention; most of those contacted agreed to participate.A total of that varied in length of operationand average thirty-onemembers-all women with distributorships monthly sales-attended each one- to two-hourinterview.
Copyright? 2001 John Wiley & Sons, Ltd. J. Organiz.Behav. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MARKETINGORGANIZATIONS MULTILEVEL

857

These interviewsinformedand guidedthis researchin severalrespects.First,memberson both sides of the sponsor-recruitdyad repeatedlystressedthe centralrole these relationshipsplay in developing successful distributorships. Membersdiscussed at length importantdimensions of their sponsor relamutual tionships, including expectations of sponsors and recruits, the nature of typical interaction between sponsorsand recruits,specific ways sponsorsexhibit leadership,and the similaritiesand differences between these relationships and traditionalmanager-employee relationships. Second, we asked members to comment generally on their motives for joining the MLM and, when necessary, promptedmembersto comment specifically aboutthe degree to which factors otherthan money influenced their decisions. Finally, we used these discussions as an opportunityto familiarize ourselves with typical MLM terminology,which facilitated the wording of questionnaireitems.

Surveysample
We subsequentlyenlisted the cooperationof the corporate-levelproducerto administera brief survey of its nationalmembership.The corporate-levelproducerrandomlyselected a sample of respondents from a computerizeddatabaseof currentmemberswithout regardto attendanceat the nationalconference. However,we imposed two limiting criteriafor inclusion in the sample. First,membersshould reside in the vicinity of theirsponsor-a distinctionmade on the basis of zip code. Occasionally,members move great distances from their sponsors, which generally alters the dynamics of the sponsorrecruitrelationship.These 'remote' membersoften rely less on theirsponsors(who continueto receive commissions from the members'sales) because face-to-face interactionwith them is no longer possible. Of the sponsorswe interviewedwho had remote recruits,most indicatedthat, althoughthey continue to communicatewith them, their relationshipssignificantlychange after the recruitsmove. The corporate-levelproducercould not provideestimates on the proportionof remote membersrelativeto total members,however,focus groupparticipants indicatedthat in their experiences,remote members are by far the exception. Thus, the numberof potentialresponsesaffected by this criterionis likely to be quite small. Second, because our primaryfocus is on the effects of transformational leadership,we wished to receive responsesfrom memberswho are more frequentlytargetsof leadershipbehaviorsthansources of them. Leadershipresearcherstypically make such distinctionseasily on the basis of organizational rank or title. As noted earlier,however, MLMs do not consist of traditionalhierarchicalsupervisorsubordinaterelationships.As a practicalmatter,memberswith substantialnumbersof recruitslikely turnless frequentlyto theirown sponsorsfor leadership;instead,theirprimaryrole becomes providing leadershipto their recruits. Therefore, we limited our sample to members who had fewer than five recruits and whose sponsors had more than five recruits. Five was selected as the cut-off because, the transitionfrom 'primarilyfollower' to 'primarilyleader' accordingto the focus groupparticipants, generallybecomes most pronouncedas a sponsorapproachesfive recruits.Indeed,this MLM's corporate producerrecognizes memberswith five recruitsby awardingthem with the title, 'unit manager.' After imposing these criteriaand accountingfor non-deliverables,the final sampleconsisted of 1883 members,all of whom are women. (Over 99 per cent of all membersof this MLM are women.) Each received a questionnaire,a cover letter explaining the purposeof the research,and a letter of endorsement from the corporate-levelproducer.A total of 886 respondentsreturnedtheir questionnaires;150 were excluded for not reportingperformancedata.The remaining736 usable responsesyielded a final response rate of 39 per cent. To check for non-responsebias, we comparedthe demographiccompositionof our sample to a cenNo statisticallysignificantdifsus of the membershiptakentwo years priorby the corporateproducer. ferences were found in memberage, education,maritalstatus,or outside employmentstatus.Because Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

858

J. R. SPARKSAND J. A. SCHENK

impositionof the selection criteriaproduceda stratifiedsample,its comparisonto a two-yearold membershipcensus does not constitutea direct test of non-responsebias. However,the fact that no significant differences emerged among the tested variablesprovides some evidence that non-responsebias was not problematic.Most respondentsstartedtheirdistributorships between one and threeyears ago, and operate their distributorships and 30 hours per week. Notably, 10 between part-time,working almost half do not have other employment;only 22 per cent have full-time jobs apartfrom their disMost membersare between 25 and 44 years old; 94 per cent are married;over half are tributorships. college graduates.

Measurement
Althoughmany of the conceptsused in this studyoriginatein well-establishedresearch,the scale items developed to measurethem typically refer to managerbehaviorsor work situationsthat simply do not apply in MLMs. Therefore,using published measuresas a guide, we adaptedor created items to suit the presentresearchcontext. Unless otherwise noted, each item was measuredon a seven-pointscale ranging from 'strongly disagree' (1) to 'strongly agree' (7). Transformational leadership behaviors Based on the memberinterviews, we focus on four transformational leadershipbehaviors(Podsakoff et al., 1990): articulatinga vision, providingan appropriate model, providingindividualizedsupport, and setting high performanceexpectations.After readingand discussing items from the Transformaindicatedthat fostional LeadershipInventory(TLI;Podsakoffet al., 1990), focus groupparticipants in their applicability limited tering the acceptanceof group goals and intellectual stimulationseemed to the particularsof sponsor-recruitrelationships. expressed some confuRegardingfosteringthe acceptanceof groupgoals, focus groupparticipants sion over the group to which the goals would apply. Although sponsors frequentlyassist membersin and encouragecooperationamong membersin their setting goals for their individualdistributorships units, they rarely,if ever, establish unit-wide goals or promote allegiance to the unit. With respect to intellectual stimulation,focus group participantsreportedthat only one of the three items in the TLI ('has stimulatedme to think about old problems in new ways') seemed to fit the norms of sponsormemberrelationships.As a rule, membersseek problem-solvingassistance from sponsorsabout very practicalmattersthat fit more closely with what Lord (1977) refers to as 'functional' dimensions of leadership.In light of these comments and the corporateproducer'srequestthat the questionnairebe brief, we omitted items from these two dimensions of transformational leadership. Belief in a higher work purpose Our search of the literaturedid not uncover any previous empirical researchon this construct.Thus, based on our member interviews, we developed three original items to measurethe degree to which MLM membersbelieve their work is partof a 'cause' that has purposes 'more important'than simply making money. Consistentwith our broad conceptualizationof this construct,we deliberatelywrote the items to reference no specific higher purposes, leaving that to the interpretationof individual respondents. Job satisfaction Job satisfactionis an emotional reactionthat stems from the belief that one's job fulfils certainvalues and needs (Locke, 1976). Although individualsmay be satisfiedor dissatisfiedwith an almost infinite Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MARKETINGORGANIZATIONS MULTILEVEL

859

varietyof individualfacets to a particular job (pay,coworkers,etc.), they also possess a global affective to their Lucas et al., 1990; McFarlinand Rice, 1992; Sujan, 1986). To measurethis response job (c.f., overall response, we adaptedfour items from the generaljob satisfaction subscale of Hackmanand Oldham's (1975) Job Diagnostic Survey. Unit cohesion To measurethe degree to which MLM membersbelieve their work groupsacted as cohesive units, we adapteditems from Podsakoffand MacKenzie(1994). The four items used in this study operationalize cohesiveness in terms of trust, dependability,cooperation,and respect. Effort and performance From our discussions with MLM members,we concluded that membereffort is best reflectedby two basic measures.First is the numberof hoursper week membersdevote to theirdistributorships, which vary from only a few hoursto more than40. We measuredthis using a five category scale (10 hours or less, 11 to 20 hours,21 to 30 hours,31 to 40 hours,more than40 hours). Second is the numberof sales 'workshops'held with prospectiveand currentcustomerseach month. This was measuredby asking members 'On average,how many workshopsdo you hold each month?' We used both as indicantsof effort. Memberstypically view performance both in termsof productsales andnew membersrecruited.For sales, we asked respondentsfor the 'dollar value of last month's sales;' for recruiting,we asked how many new membersthey had recruitedin total. Because the numberof new membersone recruitsis cumulative,members' total numberof recruitswas divided by the numberof years they had operated their distributorships to produce the average numberof new members recruitedper year. Both sales and recruitsper year were used to measureperformance.

Assessment of construct validity


To assess the performanceof individual scale items and evaluate the degree to which the data fit the hypothesized factor structure,we estimated a measurementmodel using the 26 items as indicantsof nine total constructs.Following recommendations by Breckler(1990), the data were split into derivation and cross-validationsamples.The derivationsample is used to fit a favorablemeasurementmodel to the data, which is then validatedusing the cross-validationsample. Using the derivationsample to estimate a model containing all 26 items, we found two items-one each from the higher work purpose and unit cohesiveness scales-did not load as anticipated.After deleting these items, the model was re-estimatedusing the cross-validationsample, which yielded X2(216)= 340.80 (p < 0.001). The significantX2 statisticindicates thatthe outputcorrelationsproducedby the model do not fit the input sample correlations within sampling error. However, as recommended by Anderson and Gerbing (1988) and Bagozzi and Yi (1988), we examined alternativefit indices, which suggest satisfactory model fit: GFI= 0.93, AGFI= 0.90, NFI = 0.95, RMSEA = 0.04 (p= 0.98). Finally,using the entiresample, model estimationyielded X2(216)= 463.53 (p <0.001), GFI= 0.95, AGFI= 0.93, NFI = 0.96, RMSEA = 0.039 (p= 1.0), absoluteGFI= 0.85 (see McDonaldandMarsh, 1990). All factorloadings are significant(p <0.01) and are greaterthantwice theirrespectivestandard errors, collectively supporting the items' convergent validity (see Anderson and Gerbing, 1988). Table 1 gives the scale items, descriptivestatistics, standardized path coefficients, and internalconsistency measures.Generallyspeaking,the items performedwell. One point of concern, however,arises from the low amount of common variance extracted by the job satisfaction items. Although both
Copyright@ 2001 John Wiley & Sons, Ltd. J. Organiz.Behav. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

860

J. R. SPARKS ANDJ. A. SCHENK

Table 1. Assessment of measurement Constructand items* 1. Articulatinga vision AV1: My sponsor is able to get me committed to a positive vision of the futurewith this MLM AV2: My sponsorpaints a positive picture of my possible futurewith this MLM AV3: My sponsorhelps me see a positive future for myself as a member of this MLM 2. Providingan appropriate model AMI: My sponsor leads 'doing' ratherthan 'telling' AM2: My sponsor leads by example AM3: I look to my sponsor as an example of how to be an effective MLM member 3. Setting high performanceexpectations HEI: My sponsor actively encouragesall members in her unit not to settle for less than out best performance HE2: My sponsor shows the members in her unit that she expects a lot from us as members of this MLM 4. Providingindividualizedsupport ISI: My sponsor shows respect for my personal feelings IS2: My sponsor behaves in a mannerthat is thoughtfulto my personal needs IS3: My sponsor cares about my feelings 5. Belief in higher purposeof one's work HPI: In my work with this MLM, I feel I'm part of a 'cause' that's about more than earning money HP2: While earningmoney is nice, there are other more importantreasons for my work with this MLM 5. Job satisfaction JS1: In general, I get a great deal of personal satisfactionfrom being a member of this MLM JS2: My work as a member of this MLM is personally very fulfilling JS3: I consider my work as a MLM member to be very rewarding JS4: I am generally satisfiedwith the work I do as a member of this MLM 6. Unit cohesion UCI: There is a great deal of trustamong fellow members in my unit UC2: The membersof my unit are very cooperativewith each other UC3: The members of my unit believe we can depend on each other for advice and support concerning our distributorships 7. Effort Efl: Hours per week devoted to workO distributorship Ef2: Numberof sales workshopsheld per month M 4.71 4.75 4.98 4.49 4.80 4.65 4.39 3.46 SD 1.61 1.67 1.65 2.43 1.67 1.77 1.75 1.62 Standardized LISRELestimatet 0.82 0.81 0.80 0.83 0.90 0.91 0.77 0.95 0.91 0.74 0.78 0.93 0.65 0.96 0.96 5.41 5.29 5.32 5.35 5.45 1.60 1.68 1.62 1.23 1.24 0.95 0.97 0.93 0.80 0.81 0.73 0.91 0.78 0.74 5.29 5.62 4.97 5.11 4.80 5.14 5.09 1.27 1.65 1.22 1.36 1.63 1.39 1.48 0.79 0.59 0.62 0.57 0.92 0.92 0.87 0.90 0.89 0.68 0.69 1.26 2.36 0.78 1.73 0.67 0.75
(Continues)

P, Pc(7)11 0.85 0.66

0.78

0.65

0.90

0.68

0.42

0.78

0.50

Copyright( 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MULTILEVEL MARKETINGORGANIZATIONS Table 1. (Continued)

861

Constructand items* 8. Performance Pfl: Productsales Pf2: Numberof recruitsper year of membership

M $668 0.51

SD 576 0.75

Standardized LISRELestimatet 0.85 0.63

Pc

Pvc(7)i

0.77 0.71

0.56

*Wordingof some items are altered slightly from the questionnaireto conceal the MLM's identity. Except for Effort and Performance,responses are measuredon 7-point scales rangingfrom 'stronglydisagree' (1) to 'stronglyagree' (7). tAll paths significant at p <0.01. (X2(216)=463.53; p<0.001); GFI=0.95, AGFI=0.93, NFI=0.96, RMSEA=0.039, (p = 1.0). All paths load only to their assigned constructs;no cross-loadingswere included. tMeasuresinternalconsistency (see Fornell and Larcker,1981). percentof common varianceextracted(see Fornell and Larcker,1981). lIMeasures ?Measuredon a five category scale (10 hours or less, 11 to 20 hours, 21 to 30 hours, 31 to 40 hours, more than 48 hours).

measuresof internalconsistency indicatedsatisfactoryitem reliability,and althoughthe job satisfaction items originate in well-established scales, they share only 42 per cent of their total variance. Importantlymodificationindices gave no indication of significantlycorrelatederrorvariances, sugof unexplainedvariancewas indeed random.Nonetheless, the perforgesting that the preponderance mance of the job satisfactionscale should be regardedas a researchlimitation,and is so noted in the paper's conclusions. Also following Andersonand Gerbing(1988), we tested the discriminantvalidity of the constructs correlation confidence interval aroundan interconstruct by calculating whether a two standard-error includes 1.0. If so, the constructsare suspected of not adequatelydiscriminatingbetween each other. Consideringfirst all but the transformational leadershipitems, the largest interconstructcorrelation occurs between effort and performance (r= 0.79), and has a standard error of 0.04; a two confidence interval around this correlation would not include 1.0. Apart from the standard-error correlationeven approachesthe confidenceintervalcriterion. leadershipitems, no otherinterconstruct We discuss the discriminantvalidity of the transformational leadership items separatelybecause are modelled a to correlatestrongly.The correlation as second-order factor we them and they expect between articulatinga vision and providing an appropriate model (r= 0.92) has a standarderrorof 0.04. A two standarderrorinterval about this correlationwould just include 1.0. However, this and the other generally high correlationsamong the transformational leadership factors are not problematic. In fact, they lend support to modelling the transformationalleadership items to reflect a second-orderleadershipfactor. The interconstruct correlationsand their standarderrorsare given in Table 2.

Hypotheses tests
Figure 1 shows the full model estimatedto conduct the hypothesestests, which were carriedout using LISREL8 (Joreskogand Sirbom, 1993).

Testfor commonmethodvariance
The use of self-reportsby questionnairemay carrythe prospectof inflatedcorrelationsdue to common method variance (Podsakoffand Organ, 1986). Following proceduresdescribedby MacKenzie et al. model with a (1993), we tested for common methodvarianceby estimatingthe hypothesizedstructural 0 2001 JohnWiley& Sons,Ltd. Copyright Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

o 0"

?
0

C)
Ct

Table 2. Interconstruct correlations*'t 1 1. Performance 2. Effort 3. Job satisfaction 4. Unit cohesion 5. Higher purpose 6. Articulatinga vision 1.00 0.79 0.04 0.36 0.03 0.00 0.03 0.11 0.03 0.11 0.03 0.03 0.03 0.03 0.02 0.01 0.03 2 1.00 0.43 0.03 0.05 0.03 0.20 0.02 0.10 0.03 0.01 0.02 0.03 0.03 0.00 0.03 1.00 0.42 0.03 0.67 0.04 0.45 0.03 0.35 0.03 0.38 0.03 0.30 0.03 1.00 0.34 0.03 0.62 0.04 0.57 0.03 0.45 0.04 0.68 0.04 1.00 0.33 0.03 0.28 0.03 0.25 0.03 0.26 0.03 1.00 0.92 0.04 0.86 0.04 0.81 0.04 3 4 5 6

CO
=0 CD2

t-

oo

4 I

7. Providing appropriate model 8. Setting high expectations 9. Providing individualsupport

to o

errors *Standard belowcorrelations. givendirectly shownare interconstruct estimated fromconfirmatory factoranalysis. tCorrelations correlations n = 736. X2(216)=463.53 (p < 0.001), G = 0.039(p= 1.0),absolute RMSEA GFI= 0.85 (see McDonald andMarsh, 1990).
t.)

00 O-

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MARKETINGORGANIZATIONS MULTILEVEL

863

common method factor added to the items of the multi-item latent constructs(job satisfaction, unit cohesion, belief in a higher work purpose, and the transformational leadershipbehaviors). Because most constructsare measuredusing two- or three-itemscales, the additionof a method factor to all 24 items producedan underidentified model. To overcome this problem,we excluded the measuresfor effort and performancefrom the method factor on the grounds that they measure quantifiableand objective concepts, and are therefore less susceptible to method variance. No paths were rendered non-significantby the inclusion of the method factor. Based on this evidence, we concluded that the data likely do not suffer from common method variance.

Results
In this section, we describethe resultsof fittingthe datato the model shown in Figure 1. In additionto reportingthe resultsof individualhypothesestests, we also comparethe hypothesizedmodel to several alternativemodels in orderto provide evidence that no reasonablerival models provide superiorfit to the data. Estimation of the hypothesized model produced a significant X2 statistic: X2(239)= 672.82 (p <0.001). However, several fit indices collectively suggest that the model fits the data well: GFI= 0.93, AGFI= 0.91, NFI = 0.95, CFI= 0.97, RMSEA= 0.049 (p = 0.66), RMSR = 0.40, Absolute GFI= 0.75 (see McDonald and Marsh, 1990). Figure 1 shows the standardized path coefficients for the structural effects. Table 3 and indirect direct a of model; gives summary

Hypotheses tests
The firsttwo hypothesespredictsignificantpositive relationshipsbetween transformational leadership and bothjob satisfactionand perceptionsof unit cohesion. As shown in Figure 1, both paths are significant (73,1 = 0.22, _4,1 = 0.59, bothp < 0.01). Thus, supportinghypotheses 1 and 2, transformational Table3. Summary of standardized directandindirect effects* Relationship
Transformational leadership-+ unit cohesion

Indirect effectst
0.05

Directeffectst
0.59

Totaleffects
0.64

Transformational - higher leadership purpose - effort Transformational leadership Transformational leadership --+jobsatisfaction Transformational leadership -- performance Higher purpose -- unitcohesion
Higher purpose-* effort

n.h. 0.04 0.19 0.04 n.h.


0.00

0.32 n.h. 0.23 n.h. 0.16


0.18

0.32 0.04 0.42 0.04 0.16


0.18

-- job satisfaction Higher purpose


Higher purpose performance Unit cohesion - --effort Unit cohesion -+ satisfaction Unit cohesion -+ performance Effort -+ performance

0.06
0.15 n.h. - 0.01 - 0.02

0.55
n.h. - 0.02 n.h. n.h.

0.61
0.15 - 0.02 - 0.01 - 0.02

- job satisfaction Effort

n.h.
n.h.

0.33
0.80

0.33
0.80

*n=736, X2(239)=672.82 (p<0.001), RMSEA=0.049 (p =0.66), RMSR=0.40, GFI=0.93, AGFI=0.91, NFI= 0.95, CFI = 0.97, absolute GFI= 0.75 (see McDonald and Marsh, 1990). tn.h. = effects not hypothesized. IDirect effects correspondto standardized path coefficients shown in Figure 1. All direct effects significantat p < 0.01 except unit cohesion on effort not significant.

Copyright(0 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

864

J. R. SPARKSAND J. A. SCHENK

leadershipbehaviorsrelate positively to members'job satisfactionand unit cohesion. In hypothesis 3, we predicta positive relationshipbetween transformational leadershipand belief in a higherworkpurthis the from transformational pose. Supporting prediction path leadershipto higherpurposeis significant (Ys5,= 0.32, p < 0.01). The next three hypotheses make predictionsabout the effects of higher purpose;positive relationships are expected between higherpurposeandjob satisfaction(hypothesis4), unit cohesion (hypothesis 5), and effort (hypothesis 6). Results supportedall three hypotheses: 03,5 = 0.55, /4,5 = 0.16, /2,5 = 0.18 (all p < 0.01). Hypothesis 7 predicts a positive relationshipbetween unit cohesion and effort; however, the results did not support this hypothesis (02,4=- 0.02). Finally, hypotheses 8 and 9 respectivelypredictpositive relationshipsbetween effort andjob satisfaction,and effort and performance. Both hypotheses were supported(03,2 = 0.33, 01,2 = 0.80, both p < 0.01).

Testof alternative models


The fact that data fit one model, of course, does not preclude the possibility that the data might fit alternativemodels as well. To permitdirectcomparisonof ourhypothesizedmodel to alternativemodels, we estimateda series of less restrictivenested models by freeing one theoreticallyplausiblepathat a time. We chose this particularapproachin order to assess the extent of mediation relative to our hypotheses and to better frame the results. Does transformational leadership directly relate to effort and performance? Given the central role that mediation by higher purpose plays in the model's underlyingthesis, two criticalquestionsthatshouldbe addressedare whethertransformational leadershipexhibits directrelawith effort and to after With tionships freeing the associatedpath, model performance. respect effort, estimationproduceda non-significantdifferencein X2 (X2diff= 0.27); transformational leadershipdid not directly associate with effort. Similarly,freeing the path from transformational leadershipto performance resulted in an almost equally small change in X2 (X2diff= 0.30). Collectively, these results provide additionalsupportfor the hypothesized mediating role of belief in a higher work purpose. Do higher purpose, job satisfaction, and unit cohesion directly relate to performance? The model also hypothesizes that effort mediates the effects on performanceby higher purpose and unit cohesion. To test these mediating effects, we estimated alternative models that included the respectiverelationships.First, we allowed the direct path from higher purposeto performanceto estimate. Relaxing this restrictiondid not yield a significantchange in X2(X2diff = 1.48). Likewise, freeing the path from unit cohesion to performanceproducedlittle change in X2 (X2diff=1.58). Finally, we freed the path fromjob satisfactionto performance; estimationresultedin a near negligible difference in X (X2diff= 0.02). These results furthersupportthe mediatingeffects of effort hypothesized in the model.

Discussion
This study presents an initial examination of leadership in multilevel marketing organizations (MLMs), a unique, growing, and understudiedorganizationalcontext. The results of this study, we to the organizationalliterature.First,they provide believe, make two distinctbut relatedcontributions Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MARKETINGORGANIZATIONS MULTILEVEL

865

tentative insights into the nature of MLMs, and in particular,the importance of sponsor-recruit Second, they supportan importantcomponent relationshipsto the success of MLM distributorships. of transformational belief in a leadershiptheory, higher work purpose,as a mechanismthroughwhich transformational achieve and performance. satisfaction leadershipmay With respect to our understandingof MLMs, one noteworthy insight is the apparentimportance of sponsor-recruit relationships to the performance of recruits' distributorships and recruits' satisfaction with their work. The positive direct and indirect relationshipsbetween transformational leadership and effort, job satisfaction, and performance suggest that sponsors may exert positive effects on their recruitsand their recruits'distributorships despite the fact that sponsors cannot exert direct supervisory control over their recruits' work activities. Beyond the individual relationships between sponsors and recruits,sponsors' transformational leadershipbehaviors may produce effects at the group level as well. In supportof this inference, the positive direct and indirect associations between leadership behaviors and unit cohesion suggest that members' transformationsmay lead to cooperative and cohesive behaviors among members within the sponsor-formed units. The subsequent results of this cohesiveness, however, remain unclear in light of the non-significant relationship between unit cohesion and effort. To the extent that the true relationship is indeed non-significant, it could be due to mitigating factors such as social loafing (Karau and Williams, 1993). The independent nature of MLM distributorshipsmakes this possibility plausible and perhaps worthy of furtherinvestigation. An additionalinsight into the characterof MLMs comes from members' ascriptionof higher purposes to their work, which appearsto be associatedwith their performance.This result may hold particular interest to MLM corporate producers, who frequently refer to higher purposes in communicationswith their membership(King, 1996). To the extent that membersdo achieve greater performanceand satisfactionthroughtheir beliefs in the higher good of their work, corporateproducers would be well served to learn more about which higher purposesbest motivate and satisfy their members. Apart from what may be learned about MLMs as an organizationalcontext, this study tests an importantexplanatorycomponentof transformational leadershiptheory:that the ascriptionof higher moralpurposeto one's work is at least one mechanismthroughwhich transformational leadershipproduces higherperformingand more satisfiedfollowers. Ourresultsare suggestive of thatbasic premise. Significant relationshipsbetween higher purpose and job satisfaction, unit cohesion, and effort and subsequentrelationshipswith satisfaction and performancesupportthe model's basic hypothesized mediatedsequence.Thus,one might reasonablyinferthattransformational leadershipappearsto transIn particular, what form followers. However,much remainsto be discoveredaboutthattransformation. the which best and does transformational produce specific higher purposes leadership produce, in effort and satisfaction? responses The member interviews may prove instructivein this regard.A post hoc examinationof interviewees' comments seem to indicate that a useful way of conceptualizinghigher purposesis in terms of benefits accruedto particular individualsor groups.For example, some membersspoke of exceptional the productbenefits; productsthey represented improvedthe lives of theircustomersin some uniqueor even Some this of individualcustomer benefit to society at large. extended notion exceptional way. Othermembersnoted the enjoymentthey derived from meeting new people and making new friends either throughcustomersor MLM membersin theirunits. Thus, they realize social benefitsfrom their MLM participation. Futureresearchcould develop a taxonomy of higherpurposesand providea basis for explicating this constructand furtherexamining its nomological validity. Our results must be consideredin light of the study's limitations, each of which suggest directions for futureresearch.As just discussed, one is our broadconceptualizationof belief in a higherpurpose, which may be composed of many types or facets. Anotherlimitation arises from our focus on four of Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

866

J. R. SPARKSAND J. A. SCHENK

the six transformational by the unique qualities leadershipbehaviors.Although we believe warranted of sponsor-recruit conclusions about constrains in from this their exclusion study relationships MLMs, from the marginalperthe universeof transformational comes limitation A third behaviors. leadership formanceof the job satisfactionitems, which did not sharemore than half theirtotal variance.Fourth, to the extent that non-responsebias influencedour results, the response rate of 39 per cent should be viewed as a study limitation. A series of potentiallimitationsalso arises from our use of two- and three-itemmeasuresand selfreportsas our data source. Althoughour use of two- and three-itemmeasuresresultedin partfrom the MLM corporateproducer'srequestthat the questionnaire be as brief as possible, the use of such measuresmay nonethelessgive some pause. Certainlytheiruse decreasesthe likelihood of adequatelycapturingthe theoreticaldomain of a given construct.To the extent that such measuresare theoretically problematic,we acknowledgethem as a limitationof our study.The use of two- and three-itemmeasures also occasionally presents methodological challenges. In our case, this challenge arose in the form of model underidentification when using a methodfactorto test for same source bias. As a result, we limited the applicationof the methodfactorto the latentconstructs,and thus could not conductour preferredtest for method bias. On this point, however, we offer two comments. First, Cramptonand Wagner's (1994), metaanalysis concludes that 'percept-perceptinflation may be more the exception than the rule' (p. 72). Second, Singh (2000, p. 31) notes that to the extent that self-reportsproduce bias at all, it is 'more likely to bias the [construct]mean values (upward)but less likely to bias their correlationswith other constructs.'Thus, while we cannot ignore the possibility that self-reportsbiased the data used in this study, evidence suggests that this likelihood is small. Additionally,the unique context and sample used in this study, while offering certain advantages, may inhibit the generalizabilityof the results, as does the use of a single MLM as a sample frame. However, we see good reason to expect that these results will generalize beyond the presentcontext. To the extent thatmany organizationscontinuetheirtrendstowardflatterstructures, higherrelianceon less and of behavior, clearly defined lines of multidisciplinaryteams, encouragement entrepreneurial then to characteristics of MLMs pertinent leadershipmay apply in a variety of orgaauthority, many nizational settings. By sampling organizationsthat permit a variety of leadershipbehaviors, future researchcould examine the interrelationships between transformational leadershipand other leaderfor in their abilities to motives evoke order followers' working. ship styles higher Finally, we note that our model, like all theoreticalmodels, does not include possible explanatory variables.In termsof the futureresearch,we suggest threebasic domainsof variablesthatmay fit well within the generalmediationalframeworkfollowed here. First,researchersmight examine othermediators of higherpurposeand effort. Beyond unit cohesion, the organizationalliteratureis a rich source of possibilities including organizationalcommitment,motivationto work, or work confidence. Second, as discussed earlier,we believe that explication of the higher purposeconstructholds greatpromise for deepening our understanding of the mechanismsthroughwhich transformational leadership works. Third, beyond higher purpose, other variablesmay serve to explain the mechanism through which transformational leadership transforms.As noted earlier, Shamir et al. (1993) suggest selfesteem as just one possibility. In conclusion, this researchprovides a glimpse into a seldom-studiedand complex organizational the study examines and finds supportfor the context, multilevel marketingorganizations.In particular, importanceof a key dyadic MLM relationship,that which exists between sponsorsand recruits.The evidence suggests that sponsors' transformationalleadership may be effective in bringing about greater effort, performance,and satisfaction from her recruits. Moreover, this research investigates a fundamentalexplanatorycomponent of transformational leadership theory; that one mechanism throughwhich leadersobtainsuperioroutcomes from theirfollowers is by appealingto theirfollowers' Copyright ? 2001 JohnWiley& Sons,Ltd. Behav. J. Organiz. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MARKETINGORGANIZATIONS MULTILEVEL

867

higher ordermotives. Ourresults offer supportfor that explanatorycomponentand provide a starting point for its understanding.

Author biographies
John R. Sparks is an Associate Professorof Marketingat the Universityof Dayton. He received his Ph.D. from Texas Tech University in 1994. His researchinterestsinclude leadership,ethics, and persuasion. His research has appeared in the Journal of Marketing, Communication Monographs, and the Journal of the Academy of Marketing Science among others.

Joseph A. Schenk is an Associate Professorof Managementat the Universityof Dayton. He received


his D.B.A. from Kent State University in 1976. His research interests include business strategy, leadership, and entrepreneurship.

References
AndersonJC, GerbingDW. 1988. Structural equationmodeling in practice:a review and recommendedtwo-step approach.Psychological Bulletin 103: 411-423. AtwaterLE, Yammarino FJ. 1993. Personalattributesas predictorsof superiors'and subordinates'perceptionsof military academy leadership.HumanRelations 46: 645-668. Bagozzi RP, Yi Y. 1988. On the evaluationof structural equationmodels. Journal of the Academyof Marketing Science 16: 74-94. andfinancial BarlingJ, WeberT, Kelloway KE. 1996. Effects of transformational leadershiptrainingon attitudinal outcomes: a field experiment.Journal of Applied Psychology 81: 827-832. BartlettRC. 1995. Mary Kay's foundation.Journal of Business Strategy 16: 16-19. Bass BM. 1985. Leadershipand PerformanceBeyond Expectations.Free Press: New York. Bass BM. 1990. Bass and Stodgill's Handbookof Leadership:Theory, Researchand ManagerialApplications,3rd edn. Free Press: New York. Bass BM. 1997. Personalselling and transactional/transformational leadership.Journal of Personal Selling and Sales Management17: 19-28. Bass BM. 1998. Transformational Leadership:Industrial,Military,and EducationalImpact.LawrenceErlbaum Associates: Mahwah,New Jersey. Bass BM, Avolio BJ, Goodheim L. 1987. Biography and the assessment of transformational leadershipat the world-class level. Journal of Management13: 7-19. Breckler SJ. 1990. Applications of covariance structural modeling in psychology: cause for concern? Psychological Bulletin 107: 260-273. Brown SP,PetersonRA. 1994. The effect of efforton sales performanceandjob satisfaction.Journalof Marketing 58: 70-80. Burns JM. 1978. Leadership.Harperand Row: New York. and Bycio P, HackettRD, Allen JS. 1995. Furtherassessmentsof Bass's 1985 conceptualizationof transactional transformational leadership.Journal of Applied Psychology 80: 468-478. leader behaviouras measuredby the Carless SA. 1998. Assessing the discriminantvalidity of transformational MLQ. Journal of Occupationaland OrganizationalPsychology 71: 353-358. leadershipin organizations:an insider'sperspectiveon these Conger JA. 1999. Charismaticand transformational developing streamsof research.LeadershipQuarterly10: 145-179. research:an investigationof CramptonSM, WagnerJA. 1994. Percept-perceptinflationin microorganizational prevalenceand effect. Journal of Applied Psychology 79: 67-76. Dansereau F, Graen G, Haga WJ. 1975. A vertical dyad linkage approach to leadership within formal organizations:a longitudinal investigation of the role making process. OrganizationalBehavior & Human Performance13: 46-78. Copyright ? 2001 John Wiley & Sons, Ltd. J. Organiz.Behav. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

868

J. R. SPARKS ANDJ. A. SCHENK

Deci EL. 1975. IntrinsicMotivation.Plenum Press: New York. Deci EL, Ryan RM. 1985. IntrinsicMotivationand Self Determinationin HumanBehavior.Plenum Press: New York. Direct Selling Association. 1999. 1999 Direct Selling Growthand OutlookSurvey.The Direct Selling Association: Washington,DC. Dubinsky AJ, Yammarino FJ, Jolson MA, Spangler WD. 1995. Transformationalleadership: an initial investigationin sales management.Journal of Personal Selling and Sales Management15: 17-31. Dumaine B. 1994. Why do we work? Fortune130: 197-204. Dunkel T. 1995. The Tupperware lady is an MBA. WorkingWoman20: 45-49. variablesand measurement FornellC, LarckerDE 1981. Evaluatingstructural equationmodels with unobservable error.Journal of MarketingResearch 18: 39-50. Graen G, Cashman JF. 1975. A role-making model of leadership in formal organizations:a developmental Sciences 6: 143-165. approach.Organizationand Administrative HackmanJR, OldhamGR. 1975. Development of the job diagnostic survey.Journal of Applied Psychology 60: 159-170. Chicago. JdreskogKG, Sorbom D. 1993. LISREL8User's Reference Guide. Scientific SoftwareInternational: Karau SJ, Williams KD. 1993. Social loafing: a meta-analyticreview and theoretical integration.Journal of Personality & Social Psychology 65: 681-706. King CW. 1996. Services: the New Frontin the MLM Revolution. Success 43: 72-75. Koh WL, SteersRM, Terborg JR. 1995. The effects of transformational leadershipon teacherattitudesandstudent performancein Singapore.Journal of OrganizationalBehavior 16: 319-333. Locke EA. 1976. The nature and causes of job satisfaction. In Handbook of Industrial and Organizational Psychology, DunnetteMD (ed.). Rand McNally: Chicago; 1297-1350. Lord RG. 1977. Functionalleadershipbehavior:measurementand relationshipto social power and leadership Science Quarterly22: 114-133. perceptions.Administrative and transactional Lowe KB, Kroeck KG, Sivasubramaniam N. 1996. Effectiveness correlatesof transformation leadership:a meta-analyticreview of the MLQ literature.LeadershipQuarterly7: 385-425. Journalof Lucas GH, BabakusE, IngramTN. 1990. An empiricaltest of the job satisfaction-turnover relationship. the Academy of MarketingScience 18: 199-209. MacKenzie SB, Podsakoff PM, Fetter R. 1993. The impact of organizational citizenship behavior on the evaluationsof salespersonperformance.Journal of Marketing57: 70-80. leadershiptheory on the hospital manager. Matey DB, 1991. Significance of transactionaland transformational 36: 600-606. Hospital & Health Services Administration and goodness of fit. Psychological model: noncentrality McDonaldRP,MarshHW. 1990. Choosing a multivariate Bulletin 107: 247-255. in job satisfactionprocesses.Journalof McFarlinDB, Rice RW. 1992. The role of facet importanceas a moderator OrganizationalBehavior 13: 41-54. Medley F, Larochelle DR. 1995. Transformational leadership and job satisfaction. Nursing Management26: 64JJ-64NN. actions on employee attitudes.Group& Niehoff BP, Enz CA, GroverRA. 1990. The impact of top-management OrganizationManagement15: 337-352. Podsakoff PM, MacKenzie SB. 1994. An examination of the psychometric properties and nomological validity of some revised and reduced substitutes for leadership scales. Journal of Applied Psychology 79: 702-713. Podsakoff PM, Organ DW. 1986. Self-reports in organizationalresearch:problems and prospects. Journal of Management12: 531-544. leaderbehaviorsand theireffects PodsakoffPM, MacKenzieSB, MoormanRH, FetterR. 1990. Transformational on followers' trust in leader, satisfaction, and organizationalcitizenship behaviors. Leadership Quarterly1: 107-142. Podsakoff PM, MacKenzie SB, Bommer WH. 1996. Transformationalleader behaviors and substitutes for leadership as determinants of employee satisfaction, commitment, trust, and organizational citizenship behaviors.Journal of Management22: 259-298. Salter J. 1997. Multilevel MarketingGoes Mainstream. MarketingNews 31: 1-5. ShamirB, House RJ, ArthurMB. 1993. The motivationaleffects of charismaticleadership:a self-concept based theory. OrganizationScience 4: 577-594. Singh J. 2000. Performanceproductivityand quality of frontlineemployees in service organizations.Journal of Marketing64: 15-34.

Copyright 0 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

MULTILEVEL MARKETING ORGANIZATIONS 869 SujanH. 1986. Smarterversusharder:an exploratoryattributional analysis of salespeople'smotivation.Journalof MarketingResearch 23: 41-49. WaldmanDA, YammarinoFJ. 1999. CEO charismaticleadership:levels-of-managementand levels-of-analysis effects. Academy of ManagementReview 24: 266-285. WotrubaTR, Tyagi PK. 1992. Motivationto become a direct salespersonandits relationshipwith work outcomes. Journal of MarketingChannels 1: 41-56. Yammarino FJ, Bass BM. 1990. Long-termforecastingof transformational leadershipandits effects among naval officers. In MeasurescaLeadership, ClarkKE, ClarkMB (eds). LeadershipLibraryof America:West Orange, NJ; 151-170. YammarinoFJ, Dubinsky AJ. 1992. Superior-subordinate relationships:a multiple levels of analysis approach. HumanRelations 45: 575-600. Yukl G. 1998. Leadershipin Organizations,6th edn. Prentice-Hall:Upper Saddle River, NJ. and charismaticleadershiptheories. Yukl G. 1999. An evaluationof conceptual weaknesses in transformational Leadership Quarterly10: 285-305. Yukl GA, Van Fleet DD. 1982. Cross-situational, multimethod research on military leader effectiveness. OrganizationalBehavior and HumanPerformance30: 87-108.

Copyright 9 2001 John Wiley & Sons, Ltd.

J. Organiz.Behav. 22, 849-869 (2001)

This content downloaded from 109.100.54.2 on Sat, 18 Jan 2014 06:54:47 AM All use subject to JSTOR Terms and Conditions

You might also like