Heterogeneity in Sales Districts: Beyond Individual-Level Predictors of Satisfaction and Performance

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Heterogeneity in Sales Districts:

Beyond Individual-Level Predictors


of Satisfaction and Performance
R. Venkatesh
University of Pittsburgh

Goutam Challagalla
Georgia Institute of Technology

Ajay K. Kohli
Emory University

This article examines the influence of heterogeneity within (Kohli and Jaworski 1994), rewards (Tyagi 1985), and goal
a sales unit on the unit's satisfaction and performance. orientations (Kohli, Shervani, and Challagalla 1998;
Sales unit refers to a set of salespersons working out of the Sujan, Weitz, and Kumar 1994).
same office and for the same supervisor, and heterogene- The purpose of this study is to complement this stream
ity refers to salespersons' dispersion or variance on key of research by examining the influence of heterogeneity or
dimensions. Specifically, drawing on theories in social dispersion within a sales unit on the unit's levels of job sat-
psychology, the authors study the influence of sales unit isfaction and performance. Specifically, we examine the
heterogeneity in terms of(I) demographic characteristics effects of three types of intraunit heterogeneity: (1) hetero-
(e.g., gender dispersion), (2) skills and rewards (e.g., re- geneity in demographic characteristics (e.g., gender dis-
ward dispersion), and (3) goal orientations (e.g., learning persion), (2) heterogeneity in skills and rewards (e.g.,
orientation dispersion) on a sales unit's performance and reward dispersion), and (3) heterogeneity in goal orienta-
job satisfaction levels. The hypotheses developed are tions (e.g., dispersion in learning orientation). Our basic
tested using data from a study involving 476 salespeople premise is that sales units with a wide dispersion on demo-
belonging to 105 sales units in a large organization. The graphic characteristics, skills and rewards, and goal orien-
authors find that the focal heterogeneity variables account tations influence their member salespersons very differ-
for nearly 25 percent of the total variance explained by the ently than sales units that are relatively homogeneous on
full set of independent variables included in the model. these characteristics. To the best of our knowledge, this is
the first study to examine the influence of heterogeneity
within a sales unit.
Our hypotheses are grounded in several social-psycho-
Research in sales management has identified several logical processes, two of which are particularly central--
factors that affect salespeople's satisfaction and perfor- social comparison and interpersonal competition. Social
mance (see Brown and Peterson 1994). These include fac- comparison theory, for example, suggests that salesper-
tors such as career stage (cf. Cron 1984), competence sons tend to compare their abilities and achievements with
others who are similar to themselves (cf. Festinger 1954;
Wheeler 1991). Thus, the composition of a sales unit in
Journal of the Academy of Marketing Science.
Volume 29, No. 3, pages 238-254. terms of its similarity or dissimilarity is likely to influence
Copyright 9 2001 by Academy of Marketing Science. the extent to which unit members engage in comparisons.
Venkateshet al. / INFLUENCEOF HETEROGENEITY 239

Our broad thesis is that heterogeneity in a sales unit shapes natural urge to evaluate their abilities, beliefs, and opin-
sociopsychological processes, thereby affecting the sales ions by comparing them with those of other individuals in
unit's job satisfaction and performance levels. similar positions. Such social comparisons can give rise to
The present study provides insights into several mana- affective responses such as satisfaction or dissonance (el.
gerially relevant issues such as: What are the conse- Tesser 1988). Demography theory holds that similarity in
quences of maintaining a sales unit that is diverse versus age or tenure is positively related to similarity in shared
homogeneous in its demographic composition? What are experiences and cohort effects (cf. Wiersema and Bird
the implications of having a sales unit with salespersons 1993). These, in turn, shape group processes such as com-
who differ markedly in their work motivation? How is a munication, cooperation, turnover, and performance (cf.
salesperson affected by the differentials in rewards across Milliken and Martins 1996; O'Reilly, Caldwell, and
salespersons in a sales unit? These are important questions Barnett 1989; Pfeffer 1983; Zenger and Lawrence 1989).
that have not been addressed to date and motivate the pres- Finally, the theory of interpersonal competition proposes
ent study. that competitive behavior on the part of an individual
The conceptualization and findings of this study are partly reflects striving to obtain more rewards than others
best suited to sales units (e.g., sales districts) that are com- (cf. Deutsch 1949). It also reflects a desire for perfor-
posed of independent salespeople in a branch or district mance superiority (Bond 1982; Sauers and Bass 1990).
sales office. In such sales structures, all salespeople within Whereas interpersonal competition can induce higher per-
the branch report to the same supervisor, who is responsi- formance (Weinstein and Holzbach 1973), it can also
ble for assigning quotas and allocating rewards. Salespeo- result in anxiety among salespersons and fewer interac-
ple are independent in the sense that they have their own tions among them (cf. Bettencourt, Brewer, Croak, and
sales quotas and responsibilities. By virtue of working out Miller 1992).
of the same branch office (instead of out of their homes, It is important to note that not all of the above theories
say) salespersons typically have the opportunity to interact are relevant for developing every hypothesis proposed
informally with their peers. Taken together, such sales set- later in the article. We selectively draw on the one(s) that
tings are likely to induce social-psychological processes have the ability to shed light on the particular relationships
such as social comparison. being proposed in each hypothesis.
Several Fortune 500 companies in diverse industries
(e.g., Xerox, W. W. Grainger, BellSouth, 3M, Illinois Tool Hypotheses
Works, Principal Financial Group, Prudential) have sales Development
organizations for which the findings of the present study
are relevant. ] The results of this study may also be relevant We examine the effects of three types of intraunit heter-
for sales organizations in which salespersons have consid- ogeneity on a unit's satisfaction and performance: (1) het-
erable opportunities to interact with each other (e.g., call erogeneity in demographic characteristics, (2) heteroge-
centers, retail sales). neity in skills and rewards, and (3) heterogeneity in goal
orientations. Demographic heterogeneity refers to disper-
HETEROGENEITY IN SALES sion in the inherent characteristics of salespersons. In this
UNITS: HYPOTHESES category of independent variables, we examine the role of
career stage dispersion and gender dispersion. Under
Theoretical skills and rewards, we investigate the effects of compe-
Underpinnings tence dispersion and reward dispersion within the unit.
Under goal orientation heterogeneity, we study the role of
We develop our hypotheses by relying on several theo- dispersion in the learning orientation and prove orienta-
ries of social psychology that emphasizes both the social tion of salespersons. These variables (e.g., career stage)
issues within an individual's environment such as interac- have been previously studied at the individual salesperson
tion, social comparison, and learning, as well as the associ- level; in the present study, we focus on heterogeneity
ated psychological concerns such as craving for superior- within a sales unit on the same variables (e.g., career stage
ity, attribution, and perceptions of fairness. heterogeneity within a sales district).
In arguing that heterogeneity or dispersion in a sales These six variables are argued to be related to two
unit influences satisfaction and performance levels in the unit-level dependent variables: (1) average level of job sat-
sales unit, we primarily rely on three theories of social psy- isfaction of salespeople in the unit and (2) the unit's sales
chology. The theory of social comparison (cf. Festinger performance (see Figure 1). Sales unit job satisfaction
1954; Salancik and Pfeffer 1978; Wheeler 1991)suggests refers to a unit's attitudes or affective feelings toward their
that social comparisons occur because individuals have a job, work environment, or organization (cf. Churchill,
240 J O U R N A L O F THE A C A D E M Y OF M A R K E T I N G SCIENCE S U M M E R 2001

FIGURE 1
Influence of Intraunit Heterogeneity on Sales Unit Satisfaction and Performance

Demographic Heterogeneity
9 Career Stage Dispersion (HI, -; H2, -)
9 Gender Dispersion (H3, +; H4, -)

Heterogeneity in Skills & Rewards Sales Unit Outcomes


9 Competence Dispersion 0-15, +; H6, +) ~M m Sales Unit's Job Satisfaction
9 Reward Dispersion (1-I7,+; H8, +) Sales Unit's Performance

Heterogeneity in Goal Orientations


9 Learning Orientation Dispersion (I-19,-; HI0, -)
9 Prove Orientation Dispersion (HI 1, -; H12, -)

NOTE: H1 through H12 represent the study's 12 hypotheses. An odd (even) numbered hypothesis links the corresponding antecedent variable to job satis-
faction (performance). The hypothesized sign of a relationship is shown next to the hypothesis number.

Ford, and Walker 1976). Sales unit performance refers to a (Wiersema and Bird 1993; Zenger and Lawrence 1989).
unit's sales-related achievements relative to other units When the interaction is voluntary (as in our case), greater
within the company. amounts of interaction lead to greater learning as well as
greater social recognition, thereby leading to greater
Influence of Heterogeneity in satisfaction.
Demographic Characteristics In contrast, differences (or wide dispersion) in age and
tenure result in experiential and 'language' differences,
Career stage dispersion. A powerful construct in sales thereby leading to lower communication (Schneider 1983,
force literature that captures the manifestations of tenure 1987; Zenger and Lawrence 1989). Employees with lon-
along with its associated psychological trappings is the ca- ger tenures have been through different societal conditions
reer stage of individual salespeople. Cron (1984) identi- and job employment conditions and have different atti-
fies four career stages that salespeople go through during tudes from their colleagues with shorter tenures, which in
their professional tenure---exploration, establishment, turn lowers interaction levels among group members (see
maintenance, and disengagement. Whereas Cron and col- Elder 1975; Vroom and Pahl 1971). Jackson et al. (1991)
leagues (cf. Cron, Dubinsky, and Michaels 1988; Cron and note that dissimilarity in age and tenure results in dissatis-
Slocum 1986) focus on the influence of a salesperson's ca- faction and employee turnover. In the context of career
reer stage on his or her work attitudes and performance, we stages, the preceding arguments suggest that salespersons
examine the influence of the heterogeneity in career stages in a sales unit with narrow career stage dispersion are likely
of salespersons in a unit on unit-level job satisfaction and to have higher levels of interaction and mutual attraction
performance. leading to higher job satisfaction in the unit (cf. Colarelli
A key finding in demography theory is that demo- and Boos 1992; Peterson 1983; Pierce 1998). 5 Thus,
graphic similarity (i.e., narrow dispersion on demographic
characteristics) results in positive consequences such as Hypothesis 1: The wider the career stage dispersion in a
higher communication, whereas demographic dissimilar- sales unit, the lower the unit's job satisfaction.
ity (i.e., wide dispersion) has the opposite effect. For
example, Zenger and Lawrence (1989) found that age sim- Narrow career stage dispersion implies that individuals
ilarity and tenure similarity are positively related to fre- belonging to the unit occupy similar positions and share
quency of communication, and O'Reilly et al. (1989) note similar responsibilities. Social comparison theory sug-
that tenure similarity leads to lower turnover. Similarity gests that these similarities are likely to encourage sales-
fosters attraction and use of a common language, and persons to compare their personal skills and abilities with
results in greater levels of interpersonal communication those of their peers in the unit. Furthermore, individuals
Venkateshet al. / INFLUENCEOF HETEROGENEITY 241

with similar personal characteristics have high valences greater interpersonal competition--striving for perfor-
for similar rewards (cf. Churchill, Ford, and Walker 1983). mance superiority and externally mediated rewards (cf.
To the extent that there are real limits on the number of re- Sauers and Bass 1990). This, in turn, is likely to lead to
wards that can be given out (e.g., promotions cannot be higher performance in the sales unit as compared with
given to all people), similar individuals are likely to exert mixed-gender units.
greater effort or strive to attain those rewards and signal re-
ward superiority. In other words, greater personal similar- Hypothesis 4: The wider the gender dispersion in a sales
ity is likely to result in higher levels of interpersonal unit, the lower the unit's performance.
competition. The higher striving or effort (to attain reward
superiority) suggests that the performance levels would be Influence of Heterogeneity in Skills
higher (cf. Brown and Peterson 1994). (Competence) and Rewards
In addition to the striving for rewards that would likely
enhance performance, competition also manifests itself as Competence dispersion. Research on diversity pro-
a striving for performance superiority (cf. Sauers and Bass vides a basis for positive outcomes when there is diversity
1990). In other words, even when no rewards are at stake, in skills or knowledge (cf. Bantel and Jackson 1989;
similar individuals are likely to strive for higher levels of Milliken and Martins 1996). We define sales units in
performance if only to acquire the feeling of accomplish- which salespersons have diverse skills and abilities as hav-
ment and associated superiority within a peer group. This ing a wide competence dispersion (i.e., an assortment of
desire for performance superiority is also expected to lead different skills and abilities are represented in such a sales
to greater effort and, in turn, enhance performance (cf. unit). For example, some may have excellent product
Weinstein and Holzbach 1973). Thus, knowledge, whereas others may exhibit strong sales quali-
fication or closing skills. On the other hand, when sales-
Hypothesis 2: The wider the career stage dispersion in a
people within a unit possess similar types of selling skills,
sales unit, the lower the performance of the unit.
competence dispersion is narrow.
Gender dispersion. Several studies suggest that the
Consider a sales unit in which the competence disper-
presence of the opposite gender causes group members to
sion is high. In such a sales unit, social comparison theory
display greater social arousal. McCann, Ostrom, and
suggests that individuals are likely to see each other as
Tyner (1985) report greater facilitation in mixed-gender
possessing skills or abilities that they themselves do not
groups. Fields and Blum (1997) and Aries (1976) also sug-
possess. This is likely to spur a desire on their part to ac-
gest that members of mixed-gender groups tend to interact
quire those skills or improve themselves on those dimen-
more and are less competitive than members of homoge-
sions. In addition, the wide competence dispersion allows
neous or narrow gender dispersion groups. Furthermore,
for reciprocity in that salespersons can offer each other
as the proportion of members from the minority gender
something in exchange for acquiring their respective spe-
within a group increases, there is broader consideration of
cialized skills. The possibility of reciprocity encourages
ideas from all members (see Craig and Sherif 1986). Ely
exchanges that provide for professional growth for the
(1994) found that women behave in more supportive ways
salespeople and hence are likely to result in greater job sat-
toward their peers in gender-balanced groups than in
isfaction. Such an effect is unlikely to occur if the compe-
male-dominated groups. Greater interaction and consider-
tence dispersion is narrow, in which case all salespeople
ation of others' ideas are likely to foster greater satisfaction
are very similar to each other.
among members of mixed-gender groups.

Hypothesis 3: The wider the gender dispersion in a sales Hypothesis 5: The wider the competence dispersion in a
unit, the higher the unit's job satisfaction. sales unit, the greater the unit's job satisfaction.

There is evidence to suggest that peers in same-gender There is evidence in the diversity literature to suggest
groups tend to exhibit more individualistic behavior and a that a peer group with a diverse skill set is likely to display
greater task orientation than members of mixed-gender greater creativity. Bantel and Jackson (1989) found a posi-
groups (see Kent and McGrath 1969; Wyer and tive relationship between functional diversity and the ex-
Malinowski 1972). We expect salespersons in such units to tent of innovation. Ancona and Caldwell (1988) show that
engage in social comparisons with other members because although skill diversity can lead to short-term inefficien-
individuals tend to evaluate their abilities with similar cies, it leads to creative outcomes over time. These find-
others (of. Festinger 1954; Salancik and Pfeffer 1978). ings suggest that the exchange of ideas in a sales unit with
Furthermore, salespeople's valences for various rewards diverse skills leads to better and more creative overall sell-
are likely to be similar in single-gender groups, leading to ing and hence higher performance.
242 JOURNALOF THE ACADEMYOF MARKETINGSCIENCE SUMMER2001

Hypothesis 6: The wider the competence dispersion in a cases (cf. Lawler 1973). In turn, this is likely to decrease
sales unit, the greater the unit's performance. salespeople's performance.

Reward dispersion. Organizational theorists have ar- Hypothesis 8: The wider the reward dispersion in a sales
gued that dispersion in salaries or wages within unit, the higher the performance of the unit.
workgroups can have a bearing on the nature of social rela-
tions within the groups and the levels of effort and commit- Influence of Heterogeneity
ment (cf. Pfeffer and Davis-Blake 1992; Pfeffer and in Goal Orientations
Langton 1993). In sales force settings, intangible perqui-
sites such as praise or awards can also influence attitude
Two types of goal orientations are discussed in sales lit-
and behavior. We use the term reward dispersion to cap- erature. Salespersons with a learning orientation have a
ture dispersion in both tangible and intangible rewards strong desire to improve and master their selling skills and
within a sales unit. abilities and view achievement situations as opportunities
Sales units with wide reward dispersion are character- to improve their competence (Dweck and Leggett 1988;
ized by some salespersons receiving big rewards and Sujan et al. 1994). Salespeople with aperformance orien-
others receiving relatively small rewards. Wide reward tation have a strong desire to prove themselves on perfor-
dispersion signals to the employees that the firm distin- mance metrics that count (Ames and Archer 1988; Sujan
guishes between good and poor performers. By doing so, it et al. 1994). Recent research further partitions perfor-
reinforces the salience of job-related performance (cf. mance orientation into two dimensions--prove and avoid
Leventhal 1976). The message to good and poor perform- (Elliot and Church 1997; VandeWalle and Cummings
ers alike is that there is an incentive for good performance 1997). Prove (also referred to as "approach") is defined as
(cf. Pfeffer and Langton 1988). Salespeople are likely to the desire to prove one's ability and gain favorable judg-
view such a system as being fair, and therefore this is likely ments about it, while avoidis defined as the desire to avoid
to lead to feelings of satisfaction on their part (cf. Mitchell, the disproving of one's ability and to avoid negative judg-
Lewin, and Lawler 1990). In contrast, when reward disper- ments about it (Elliot and Church 1997; VandeWalle and
sion is narrow in a sales unit, salespeople are likely to feel Cummings 1997). Most studies show that the prove
that regardless of their effort or performance, they are dimension of performance orientation has positive influ-
ences such as enhancing task involvement and improving
likely to receive the same level of rewards as their peers. In
performance (e.g., Elliot and Church 1997; Elliot and
such units, salespeople may resent their supervisor for tak-
Harackiewicz 1996; Harackiewicz, Barron, Carter, Lehto,
ing the easy way out by rewarding everyone more or less
and Elliot 1997; cf. VandeWalle and Cummings 1997 for
equally (as opposed to being more discriminating). Thus,
exception). The "avoid" dimension has been shown to
narrow reward dispersion is likely to reduce salespeople's
have negative consequences on individuals. Our focus in
job satisfaction.
this article is on the "prove" dimension of performance
orientation. To be succinct, we henceforth refer to prove
Hypothesis 7: The wider the reward dispersion in a sales performance orientation simply as prove orientation.
unit, the higher the unit's job satisfaction.
Learning orientation dispersion. Since individuals
As noted earlier, wide reward dispersion suggests to a with a learning orientation believe in a positive effort-mas-
salesperson that the "system" (i.e., supervisor, incentive tery relationship, they place a high value on feedback and
program) is sensitive to salespersons' differential contri- view it as diagnostic information (Dweck and Leggett
butions. In addition, wide reward dispersion indicates to 1988; Farr, Hofmann, and Ringenback 1993; VandeWalle
all salespersons except those at the very top that there are and Cummings 1997). Likewise, feedback seeking is
real opportunities for earning bigger rewards. The exis- viewed as a self-regulation strategy and as a means of con-
tence of such opportunities is likely to encourage salesper- tinued effort (VandeWalle and Cummings 1997). In sales
sons to focus on their jobs, thereby enhancing perfor- units with wide learning orientation dispersion, some
mance (Hackman and Oldham 1980). On the other hand, a salespeople are much more interested in improving their
narrow reward dispersion is likely to create the impression skills and abilities than others. Those with a learning ori-
that the link between performance and rewards is weak (cf. entation are likely to engage in feedback exchanges as a
Leonard 1990), particularly if salespersons perceive sig- means for enhancing their skill set (Butler 1993;
nificant differences in the performance across salesper- VandeWalle and Cummings 1997). However, their at-
sons within the unit. Expectancy theory suggests that tempts to seek or create an environment that is conducive
salespeople's instrumentality estimates, and consequently to these feedback exchanges are likely to be thwarted by
motivation, are likely to be low because of the perceived those who lack the motivation for mastery. Thus, salespeo-
weak linkage between performance and rewards in such ple with a learning orientation are likely to be frustrated by
Venkateshet al. / INFLUENCEOF HETEROGENEITY 243

the attitude of peers who lack similar aspirations. On the from their peers within the unit because only a subset of
other hand, salespeople in units with a narrow dispersion the salespersons in the unit (specifically, the other sales-
on learning orientation have similar attitudes toward im- persons with a high prove orientation) have an interest in
proving their skills and abilities, thereby leading to a more demonstrating their abilities and performing against their
harmonious environment. Thus, we expect narrow disper- manager's metrics (Harackiewicz et al. 1997). This being
sion on learning orientation to be positively related to a the case, the competitive arousal is likely to be relatively
unit's job satisfaction. low. In addition, salespersons with low prove orientation
are likely to withdraw further from the competitive game
Hypothesis 9: The wider the learning orientation dis- in face of aggressive impression management activities of
persion within a sales unit, the lower the unit's job salespeople with high prove orientation and their constant
satisfaction. monitoring and comparing of each other's performance
(VandeWalle and Cummings 1997). Thus, the drive to per-
Moreover, in units with a wide learning orientation dis- form for salespersons with high and low prove orientation
persion, the lack of desire on the part of some salespeople is likely to be lower, leading to lower sales performance at
to learn, improve, and exchange ideas is likely to frustrate the sales unit level.
even the salespersons with a high learning orientation, di-
minishing their urge to acquire new skills and, in turn, Hypothesis 12: The wider the prove orientation disper-
leading to lower performance. Likewise, salespersons sion in a sales unit, the lower the unit's performance.
with a low learning orientation are likely to find the queries
and suggestions of high-learning orientation salespersons
meddlesome. Overall, this is likely to diminish the perfor- RESEARCH METHOD
mance of the unit.
Data Collection
Hypothesis 10: The wider the learning orientation disper-
sion in a sales unit, the lower the unit's performance. Since the primary focus of the study is on investigating
Prove orientation dispersion. Salespeople with a prove the influence of heterogeneity in a sales unit, we selected
orientation focus on demonstrating their proficiency rela- an industrial sales company with a branch sales structure
tive to others and on being judged able (Harackiewicz et al. in which salespeople spend a significant amount of time
1997; VandeWalle and Cummings 1997). They frequently working from their office. The study was conducted in one
compare their performance to supervisory expectations company for one main reason--it offers greater control
and the performance of their peers (VandeWalle and over exogenous variables and enhances the internal valid-
Cummings 1997). Wide prove orientation dispersion im- ity of the study. Since it is the first study of its kind to the
plies that some salespeople have a strong desire to closely best of our knowledge, our intent was to keep the study
monitor and compare their performance with that of their tight by eliminating as many alternative explanations as
peers, whereas other salespersons in the sales unit are possible for the results. The single-company nature of the
much less focused on making such comparisons. Further- study suggests that prudence is warranted when generaliz-
more, a wide prove orientation dispersion implies that ing the findings of the study to other contexts. As McGrath
some salespeople work at demonstrating their abilities to and Brinberg (1983) point out, however, generalizeability
others in the sales unit, whereas other salespersons are less is not something that can be established in a single study;
interested in such activity. The heterogeneity in motiva- rather, it is established or refuted during a number of dif-
tions is likely to lead to tensions because the salespeople ferent studies, each using a somewhat different context,
who prefer to be "left alone" are likely to be irked by the sample, and measurement schema. We discuss these
frequent monitoring, comparisons, and impression- points further in the Discussion section.
management activities of salespeople with a high prove The participating company has $1.5 billion in revenues
orientation. Overall, this is likely to lower the level of job and has been in existence for more than 90 years. It is one
satisfaction in the sales unit. of the largest printers in North America with nationwide
facilities and has more than a hundred sales locations in
Hypothesis 11: The wider the prove orientation dispersion strategically located market areas. The company provides
within a unit, the lower the unit's job satisfaction. total print management products and services to Fortune
1,000 customers, producing and distributing forms, labels,
In sales units with wide prove orientation dispersion, direct response, and commercial printing. The sales orga-
individuals with low prove orientation have lower task in- nization is structured around geographic territories. The
volvement than those with a high prove orientation (Elliot average district has six salespeople, and metropolitan
and Harackiewicz 1996). Salespersons with high prove areas tend to have multiple districts. Several Fortune 500
orientation are unlikely to feel much competitive pressure companies have similar sales organizations (e.g., Xerox,
244 JOURNALOF THE ACADEMYOF MARKETINGSCIENCE SUMMER2001

W. W. Grainger, BellSouth, 3M, Illinois Tool Works, Prin- Seventy-seven percent of respondents reported having
cipal Financial Group, Prudential). work-related communications more than once a day with
The salespeople in our sample are independent, mean- their peers, while 19 percent did so "a few times per week."
ing that they have their own accounts, quotas, and territory. Thus, 96 percent of our respondents have frequent
They are completely responsible for selling to these work-related interactions with their peers, suggesting that
accounts. They, however, do assist national account man- there is a strong basis for peer-related sociopsychological
agers occasionally for selling to accounts within their ter- processes to occur within our study context. Finally, pre-
ritory. Once again, this type of sales organization is com- test interviews conducted with salespeople and managers
mon in many Fortune 500 companies. The compensation confirmed that salespeople in the sample shared selling in-
is a mix of commission and base salary. The proportion of sights, talked of their accomplishments and failures,
variable compensation to fixed compensation is roughly mentored junior salespeople, and more generally inter-
55:45, with sales contests accounting for 3 to 4 percent of acted at a professional level, thus suggesting that the sam-
their total income. The variable compensation is some- ple is appropriate for the study.
what higher than is found in most Fortune 500 companies.
Performance results are typically communicated individu- Measure Development
ally and are not posted publicly. Overall, the sales organi- and Pretesting
zation in our study is fairly typical of that in many, if not
most, major corporations. While existing scales were used for measuring several
of the constructs, for a few constructs, scales were unavail-
Data collection from salespeople. A questionnaire was able or were otherwise limited in their applicability to the
developed for gathering responses from salespeople. This context of the study. It was therefore necessary to adapt
questionnaire captured the four career stages, learning ori- some existing measures or develop new ones. The partici-
entation, prove orientation, rewards, job satisfaction, and pating company assisted the researchers in adapting exist-
demographic and job context information. We forwarded ing and new measures to suit their context. Once the mea-
617 survey packets, one for each salesperson, to the partic- sures were finalized, the instrument was pretested at a
ipating company, which mailed the packets to the sales dis- sales office of the company. Six salespeople from this
tricts. Each packet consisted of a survey, postage-paid office were provided a questionnaire and were asked to
return envelope, and a letter from the sales vice president identify scale items that were ambiguous and/or repetitive.
of the company. The letter explained the purpose of the Based on the inputs received, some items were eliminated,
study and assured salespeople complete anonymity. Sales- others modified, and new items added. Overall, few con-
people were asked to mall the surveys directly to the re- cerns were reported by the respondents, suggesting that
searchers. Approximately 10 days after the first mailing, a the questionnaire was ready for final administration.
reminder letter from the sales vice president was sent to
all salespeople and sales managers. Another 2 weeks
Measures
later, second survey packets, identical to the first, were
sent to all salespeople and sales managers. These efforts
When using established scales, we made an effort to
yielded 476 responses from salespeople for a response
preserve the wording of the original scales. Only minor
rate of 77 percent.
modifications were made to suit the context. The scales are
Data collection from sales managers. A packet similar presented in the appendix. Unless indicated otherwise,
to the one sent to salespeople, but with a different ques- salespeople are the respondents. The dependent variables
tionnaire, was mailed to sales managers at each of the 116 and the "root" independent variables (e.g., career
districts. The manager questionnaire captured competence stagemas opposed to heterogenity in career stage) have a
dispersion, competence mean, performance, and unit de- reliability coefficient higher than the .70 level recom-
mographics. The data collection process was identical to mended by Nunnally (1978). The coefficient alpha for
that of salespeople. Responses were received from 105 competence dispersion and performance are not reported
sales managers, for a response rate of 91 percent. Man- because they are formative scales.
agers indicated that the average unit size is six.
Career stage dispersion. The four career stages were
Sample characteristics. Ninety-three respondents are measured using items drawn from Cron and Slocum
female. On average, respondents have about 5 years of (1986). The original scale of Cron and Slocum (which is
selling experience at the company, 6 years of total selling adapted from Super, Zelkowitz, and Thompson 1981) has
experience, and 9 years of total work experience. Respon- 56 items. We elected to use a subset of the original scale to
dents spend about 46 percent of their time working from reduce respondent fatigue by keeping the questionnaire
their office, about 47 percent of their time working in the length manageable. We used a two-step process for select-
field, and the remainder (i.e., 7%) working from home. ing the items. In the first step, we selected items that cap-
Venkateshet al. / INFLUENCEOF HETEROGENEITY 245

ture the breadth of the construct. For example, during the "knowledgeable about selling"). The computed mean
exploration stage, salespeople are concerned about finding score for the four items represents the unit's competence
the "right" occupation. The items we selected (e.g., De- dispersion (because the scale anchors capture the extent of
ciding what I really want to do for a living; Making sure of dissimilarity or assortment in competency among the
my current occupational choice) capture this notion. In the members).
second step, when items in the original scale were quite
Reward dispersion. Four items were developed by the
similar to each other, we selected a subset of them to save
researchers to assess the rewards received by an individual
space. This two-step process yielded 16 items to measure
salesperson relative to others in his or her unit. The items
the four career stages--exploration, establishment, main-
were measured using a 7-point scoring format ranging
tenance, and disengagement.
from much worse to much better. Three items capture three
Fifteen of the 16 items used in the study are identical to
tangible rewards (e.g., salary + commissions), while one
the original scale. In the course of the pretest, 1 item was
item captures intangible rewards (i.e., recognition from
modified from "advancing to a more responsible position"
the manager). The reward dispersion within a unit was
to "advancing to a higher position in my company" for
measured by the standard deviation of reward scores of
better clarity. The career stage items were scored by sales-
salespersons within the unit.
people on a 5-point scale ranging from not at all concerned
to very concerned. Learning orientation and prove orientation disper-
The average score on each of the four career stage sions. As noted earlier, three goal orientations--learning,
scales was computed for each salesperson. The highest prove, and avoid--are identified in the literature. The lat-
score among these was used to assign a salesperson to a ter two are dimensions of performance orientation. Our fo-
career stage (cf. Cron and Slocum 1986). For example, ifa cus in this study is on learning and prove orientations.
salesperson's average score on the establishment scale was Sujan et al.'s (1994) performance orientation measures
the highest among the four, then the salesperson was cate- map closely on to the prove orientation.
gorized as being in the establishment stage.3After catego- Five items were drawn from Sujan et al. (1994) to mea-
rizing the salespeople, the career stages were assigned the sure learning orientation. The five-item scale has been
values 1 through 4, with 1 and 4 representing exploration used in prior research by Kohli et al. (1998) and has been
and disengagement stages, respectively. Next, the standard shown to have good psychometric properties. For clarifi-
deviation of career stages of all respondents from each unit cation, a prefix "In my line of work" was added to one
was computed and used as a measure of the career stage learning orientation item, and a suffix "at (company)" was
dispersion of a unit. added to another item. Items from both constructs were
Gender dispersion. We first computed the proportion of measured using a 5-point scale anchored by strongly dis-
majority gender relative to group size for each sales unit. agree to strongly agree. Prove orientation was measured
(1 - proportion of majority gender) was used as a measure using the scale developed by Sujan et al. (1994). One addi-
of gender dispersion. The computed measure of gender tional item was added to the six-item Sujan et al. scale.
dispersion ranges from 0 to 0.5, with 0.5 indicating the Minor modifications were made to these items to suit the
widest possible gender dispersion. study context (e.g., "district manager" instead of "supervi-
sor," "district office" instead of "company" "product sales
Competence dispersion. Managers were respondents to rep" instead of "salesperson"). The dispersion within a
the competence dispersion items. Specifically, unit man- unit on each of these variables was measured by the stan-
agers were asked to report on a 5-point scale the extent to dard deviation of the scores on these variables.
which salespersons within their unit were similar or dis-
similar on four competence items. The four-item compe- Unit job satisfaction. Four of five items from Brown
tence scale was adapted from Kohli and Jaworski's (1994) and Peterson (1994) were used to measure a salesperson's
five-item scale on individual salesperson competence. The job satisfaction. One item (i.e., "How worthwhile the
items had to be adapted because our study context is salesperson considered the investment of time spent sell-
broader than that of Kohli and Jaworski's (i.e., industrial ing") was dropped because it was narrow for the context of
sales vs. automobile sales), and our unit of analysis is also our study. While the content of the items is identical to
different from Kohli and Jaworski's, thereby requiring mi- Brown and Peterson's, the anchors of the scale were
nor modifications to items. For example, the item "I trust adapted for the study. Thus, whereas Brown and Peterson
my coworkers' judgment of relevant business matters" used item-specific anchors, for example, extremely satis-
cannot be used to measure dispersion of competence in a fied-not satisfied at all, extremely excitingmnot at all ex-
sales unit. As such, it was truncated to "judgment about citing, we used a standard set of response categories
relevant business matters" Of Kohli and Jaworski's five anchored by strongly disagree to strongly agree to mini-
items, one item was dropped during the pretest (i.e., the mize the burden on the respondents. A unit's job satisfac-
item "know a lot about selling" was deemed too similar to tion score was determined by taking the mean of the job
246 JOURNALOF THE ACADEMYOF MARKETINGSCIENCE SUMMER2001

satisfaction scores obtained from the respondents within (Dillon and Goldstein 1984). The highest pairwise corre-
that unit. lation was between learning and prove orientations and
was .49.
Unit performance. Performance information was
sought from managers who assessed their unit's perfor-
mance on three items--percentage of quota achieved, HYPOTHESIS TESTING
sales volume, and overall performance. These items are AND RESULTS
the same measures used by the company's managers for
evaluating the performance of their salespeople and sales Recall that the antecedent variables of the study are the
units. A 7-point scale ranging from much worse to much six types of dispersion within each sales unit, namely,
better (compared to an "average" sales unit of the com- career stage dispersion, gender dispersion, competence
pany) was used for measuring unit performance. dispersion, reward dispersion, learning orientation disper-
sion, and prove orientation dispersion. As we are inter-
Control measures. Past research suggests that the mean
ested in the impact of these unit-level antecedents on sales-
levels of the independent variables included in the study
persons in those units, we perform our hypothesis testing
also affect job satisfaction and performance. To ensure a
at the unit level.
conservative test of the hypotheses, we control for the ef-
fects of mean career stage of salespersons in each unit, unit While we received responses from 77 percent of the
mean competence level, unit mean reward level, unit mean 617 salespersons in the 105 sales units that responded, the
learning orientation, and unit mean prove orientation. In response rate varied across units. As our antecedent vari-
addition, since one may argue that the larger the unit, cet- ables are unit-level dispersions, four of which are com-
eris paribus, the higher the performance, we also control puted based on responses from member salespersons, we
for unit size when performance is the dependent variable. had to ensure that we had an adequate number of responses
from each sales unit to accurately measure the dispersion
Unidimensionality, Convergent constructs. For this reason, we dropped units with a
and Discriminant Validity response rate of less than 50 percent, resulting in 83 sales
units, from which data were used for analysis at the unit
The independent variables are all computed dispersion level. These units have an average group size of 5.82. We
scores, and hence confirmatory factor analysis (CFA) is compared the distributions of the antecedent variables in the
not relevant for these variables. However, the items from full sample with the study sample. The means and standard
the 'root' scales (e.g., learning orientation) were subjected deviations are virtually identical, and only two correlations
to CFA for scale refinement purposes. Two CFAs were are different between the two samples. Thus, the study sam-
performed, each involving constructs that were most simi- ple is representative of the full sample. (Interested readers
lar to each other. The first was on the career stage con- may obtain the correlation matrix based on the full sample
structs--exploration, establishment, maintenance, and from the first author.)
disengagement--and the second was on the two goal ori- The correlation matrix for the antecedent, control, and
entations-learning orientation and prove orientation. dependent variables in the studied sample is reported in
The career stage CFA involved estimating a four-factor Table 1.4
model. Two items, one from the maintenance career stage The correlation matrix reveals that 12 of the 15 possible
and one from the disengagement career stage, were elimi- correlations among the antecedent variables are statisti-
nated. The resulting model had a good fit, even though the cally nonsignificant (p > 0.10). We see a weak yet statisti-
chi-square index was significant 0C271= 160.72, p < .10). cally significant correlation between career stage disper-
The Comparative Fit Index (CFI), Goodness-of-Fit Index sion and prove orientation dispersion (iS = .21, p < . 10),
(GFI), and root mean square error of approximation between gender dispersion and competence dispersion
(RMSEA) were 0.97, 0.95, and 0.054, respectively. In the (~ = .22, p < .05), and between learning orientation disper-
goal orientations' CFA, one item belonging to the learning sion and prove orientation dispersion (15 = .33, p < .01).
orientation scale was discarded because of its low loading. These results are reassuring in at least two ways. First, they
The resulting two-factor model had an acceptable fit (~z42= offer empirical support for our premise that our antecedent
161.84, CFI = 0.90, GFI = 0.93, RMSEA = 0.08). variables are largely distinct constructs. Second, the
Convergent validity is indicated if the path coefficients results strongly suggest that multicollinearity among the
from latent constructs to their corresponding manifest antecedent variables is unlikely to be an issue.
indicators are statistically significant (i.e., t > 2.0). All We hypothesized relationships between the above ante-
items loaded significantly on their corresponding latent cedent variables and two unit-level dependent vari-
constructs. A pairwise comparison of the constructs indi- ables-job satisfaction and performance. As noted earlier,
cated that all latent-trait (i.e., r correlations were signifi- we included the following control variables to ensure a
cantly different from 1, suggesting discriminant validity stringent test of the hypotheses: mean career stage of
,.,,,, I

4(.

| r'-
,-., t"-
I

I'

I' t'l"

~ . o ~ ~
I' I I I I" I

.x- *

I" I" I" 1 I' I I

q= I I"
@
t,,,,1
I" I'1'' I I

$
@ I I
@
@
I I I I I

0,,

L_
i ~l ~t

,x,

q
V

] ,x,
d
V

247
248 JOURNALOF THE ACADEMYOF MARKETINGSCIENCE SUMMER 2001

salespersons in each unit, unit mean competence level, dents per unit as the weight variable to estimate the param-
unit mean reward level, unit mean learning orientation, eters of the satisfaction model. Such an issue does not arise
unit mean prove orientation, and unit size (for the perfor- in the estimation of parameters of the performance model.
mance model). Two multiple linear regression models This is because the unit performance information was ob-
were set up as follows: tained from the unit managers (one manager per unit).
We find that the two dependent variables have a weak
JOBSAT=o~I + t~t*CS_D+O~*GEN_D (1) but statistically significant correlation coefficient (15 = .30,
+ ct3*COMP_D + O~4*REW_D p < .05). The correlation coefficient between residuals
+ cq*LO_D + O~6*PO_D+ o.7*CS_M1 from the two models based on separate regressions is weak
+ as*CS_M2 + o.9*COMP_M and statistically nonsignificant (15 = .20, p > .05). Yet, to be
+ ato*REW_M + o~H*LO_M conservative in our estimation of the model parameters,
+ ~t12*PO_M+ error we used seemingly unrelated regression (SUR) adjusting
for heteroscedasticity in the satisfaction model.
PERFOR = ~o + ~I*CS_D + ~I2*GEN_D (2) The regression coefficients of the estimated regression
+ ~3*COMP_D+ ~4*REW_D models are reported in Table 2. The table also contains
+ 135*LO_D+ [36*PO_D + [iT*CS_M1 summary statistics for the two models.
+ ~s*CS_M2 + [39*COMP_M
As indicated in Table 2, the SUR system of models
+ ~Io*REW_M + I~H*LO_M
with sales unit's job satisfaction (JOBSAT) and perfor-
+ 1312*PO_M+ ~13*UNIT_SIZE+ error,
mance (PERFOR) as the dependent variables is statisti-
cally significant (F value = 12.452, p < .001, adjusted R 2 =
where the dependent variables are
65.2%, sample size = 77 units). Four of the six antecedent
JOBSAT = a unit's mean job satisfaction level and variables---career stage dispersion, gender dispersion, com-
PERFOR = the unit's performance level.
petence dispersion, and learning orientation dispersion--
have a statistically significant influence on job satisfaction
The antecedent variables are
in the hypothesized direction, supporting Hypotheses 1, 3,
CS_D = the unit's career stage dispersion, 5, and 9 (see Table 2). All of the related regression coeffi-
GEND_D = the unit's gender dispersion, cients are significant with p < .05 except that for gender
COMP_D = the unit's competence dispersion,
dispersion, which has a p value of .06. In the model with a
REW_D = the unit's reward dispersion,
unit's performance (PERFOR) as the dependent variable,
LO_D = the unit's learning orientation dispersion,
and three of the six predictors--gender dispersion, reward dis-
PO_D = the unit's prove orientation dispersion. persion, and prove orientation dispersion--are significant
with p < .05 and in the posited direction. Thus, we find sup-
The control variables are port for Hypotheses 4, 8, and 12.5
CS_M 1 = the unit's mean career stage (a dummy The results suggest the following. Only gender disper-
variable that equals 1 if the mean is sion is significantly related to both dependent variables
establishment and 0 otherwise), investigated in the study. It is positively related to unit job
CS_M 2 = the unit's mean career stage (another satisfaction and has a negative relationship with unit per-
dummy variable that equals 1 if the formance. The other antecedents have a significant rela-
mean is maintenance and 0 otherwise), tionship either with unit satisfaction or performance but
COMP_M = the unit's mean competence level, not both. For example, while career stage dispersion and
REW_M = the mean reward level within the unit, learning orientation dispersion have a significant negative
LO_M = the mean level of learning orientation
relationship with unit job satisfaction, neither is a signifi-
in the unit,
cant predictor of unit performance. Competence disper-
PO_M = the mean level of prove orientation,
and sion has a significant, positive relationship with unit job
UNIT_SIZE = the number of salespersons in a unit. satisfaction but is unrelated to unit performance. On the
other hand, while reward dispersion has a positive rela-
Recall also that a unit's job satisfaction measure tionship and prove orientation dispersion has a negative
(Model 1) is the mean of the job satisfaction scores ob- relationship with unit performance, neither is related to
tained from the respondents within that unit. As the num- unit job satisfaction.
ber of respondents varies across units, the error term could Overall, these findings suggest that the heterogeneity
violate the assumption of homoscedasticity (Johnston and within a sales unit has a reasonably strong influence on a
DiNardo 1997:171). Therefore, akin to weighted least unit's satisfaction and performance levels. In all, 7 of our
squares (WLS) procedure, we use the number of respon- 12 hypotheses are supported by the data. The control
Venkateshet al, / INFLUENCEOF HETEROGENEITY 249

TABLE 2
Influence of Within-Unit Dispersions on Unit Salespersons' Job Satisfaction and Performance
Unstandardized Regression Coefficients(N = 77)
Model 1: Dependent Variable Model 2: Dependent Variable
Variable Description Unit's Job Satisfaction (JOBSAT) Unit's Performance (PERFOR)
Antecedent variables
CS_D Unit's career stage dispersion -.334*** .063
GEN_D Unit's gender dispersion .283* -.870***
COMP_D Unit's competence dispersion .123"** -.056
REWD Unit's reward dispersion -.026 .495***
LO_D Unit's learning orientation dispersion -.402** .492
POD Unit's prove orientation dispersion .106 -1.532"**
Control Variables
CS_M1 Unit's mean careerstage (establishmentor otherwise) .621"** .440"
CS_M2 Unit's mean careerstage (maintenanceor otherwise) .671'** .335
COMP_.M Unit's mean competencelevel .040" .293"**
REWM Unit's mean rewardlevel .155"* .097
LO_M Unit's mean levelof learning orientation .241" .058
PO_M Unit's mean levelof proveorientation .150 .038
UNIT_SIZE Unit's size NA .084**

Model Summary (based on seemingly unrelated regressionadjustedfor heteroscedasticity)


F value 12.452 (p < .001)
SystemwideadjustedR2 65.2%
Adjusted R2 attributableto control variablesalone 49.2%
Adjusted R2 attributableto heterogeneityvariables 16.0%
*p < .10. **p < .05. ***p< .01.

variables in our models seem to have a stronger influence DISCUSSION


on job satisfaction than on unit performance. In the model
involving job satisfaction, we find positive and significant The purpose of the present study is to introduce and
regression coefficients for five control variables, namely, explore the usefulness of the sales unit perspective in pro-
the two dummy variables that capture whether the mean viding fresh insights into determinants of job satisfaction
career stage is establishment or maintenance, the unit's and sales performance. Specifically, we argue that hetero-
mean competence level, the mean reward level, and the geneity within a sales unit along several dimensions influ-
mean level of learning orientation. In the performance ences the sales unit's job satisfaction and performance. We
model, mean career stage (establishment or otherwise), use data from a number of sales units in a single company
mean competence, and unit size are significantly (and pos- to test our thesis.
itively) related to unit performance. The results suggest The sales unit perspective has implications for sales
that the mean and dispersion levels of a variable (e.g., managers at two distinct levels. First, the findings help a
career stage) can have distinct and independent effects on a manager better understand why salespersons feel and
dependent variable. These effects cannot be clubbed behave a certain way. This is because the sales unit per-
together and must be interpreted on a ceteris paribus spective urges the manager to consider the nature of the
basis3 unit to which the salesperson belongs rather than focus on
To apportion the explained R 2 between the covariates the salesperson only. Thus, for example, a salesperson
and the focal heterogeneity variables, we also ran SUR for may be making good money but may still be unhappy
models involving the covariates and dependent variables because he or she feels that the reward system does not dis-
only. We find that the covariates alone yield an adjusted R 2 criminate based on relative contributions. Second, the
of 49.2% for the model system. It is very encouraging that sales unit perspective helps a sales manager think about
over and above the explanatory power of the covariates, structuring the sales unit in a manner that is likely to lead to
the antecedent variables explain 16 percent of the variance desired outcomes. In particular, the study suggests impor-
in the dependent variables or, equivalently, 24.7 percent of tant considerations when making recruiting, transfer, and
the variance explained by the full set of independent vari- compensation decisions as discussed below.
ables included in the model.
250 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2001

Managerial Implications unit's performance. It seems that this might be best han-
dled by transfers of salespeople to other units such that
The study's findings have straightforward implications prove orientation dispersion within a sales unit is not very
for managers. For example, our finding that salespeople in high.
heterogeneous or mixed-gender units have higher satisfac-
tion levels than those in single-gender units probably sug- Contributions to Theory
gests creating a gender-balanced sales unit if the objective
is to increase employee job satisfaction. In contrast, career While our article is rooted in extant theories of social
stage dispersion appears to reduce salespeople's satisfac- psychology and sales force management, it extends cur-
tion. Presumably, career stage dispersion within a sales rent thinking in these two areas in a significant manner. To
unit leads to divergent attitudes and tends to diminish the best of our knowledge, ours is the first study that exam-
intraunit communication. This finding suggests that sales ines the effects of unit-level variables (dispersion) on the
managers should institute specific mechanisms that satisfaction and performance of sales units. Furthermore,
encourage salespeople belonging to units with divergent the set of antecedent variables in this study (e.g., career
career stages to interact frequently with fellow salespeople stage dispersion, reward dispersion, learning orientation
within the same unit. Independent of their career stage, dispersion) has not been the focus of studies examining
senior and longtime salespeople are typically storehouses group processes. In addition, while the impact of peers is
of market knowledge and selling skills and can serve recognized in social psychology, an overwhelming num-
important roles as mentors and coaches of junior salespeo- ber of these studies are in experimental settings involving
ple. Fostering interaction between such veterans and "artificial" groups and individuals as varied as small chil-
younger recruits may help reduce the negative effects of dren, undergraduates, and residents in hostel dormitories.
career stage dispersion on satisfaction. In contrast, ours is a field study involving salespeople who
It also appears that narrow dispersion in learning orien- have sustained associations with their peers.
tation enhances salespeople's satisfaction. This is perhaps Although researchers have recognized the importance
because salespeople in such units have similar attitudes of organization climate, empirical research on the critical
toward improving their skills and abilities, thereby leading role of peers is scarce. Moreover, the present study sug-
to more interactions and harmony. The implication for gests that well-known notions, such as career stages of
managers is that they should institute mechanisms that salespeople, gender, and rewards, perhaps may be of even
ramp up the learning orientation of individuals who are greater managerial import when viewed from the perspec-
low on this dimension lest the others become demoralized. tive of their dispersion within a group.
We find that sales unit performance is negatively
related to dispersions in gender and prove orientation but Research Limitations
positively related to reward dispersion. The finding on and Directions
gender dispersion makes gender diversity a tricky issue to
manage, especially because it is positively related to satis- Our study has several limitations. These represent
faction. We suggest that managers promote gender diver- opportunities for both theoretical and empirical work in
sity because of its positive impact on satisfaction. This can the area. Our focus has been on salespeople who work
be done through judicious recruitment, although such poli- independently in the field while maintaining social contact
cies tend to take time to implement. To mitigate the nega- with their colleagues. The emergence of sales teams in
tive effect of gender diversity on performance, it is impor- which cooperation is a major element warrants close atten-
tant that managers take additional actions to ensure that tion. In particular, it would be very useful to examine how
performance is not hurt. Based on past research, one sug- the group dynamics change if the selling organization is a
gestion is that managers rely on motivational mechanisms multifunctional collaborative team rather than composed
(e.g., goal setting, providing feedback) to alleviate the of independent salespersons who "compete" with other
downside of gender diversity. independent salespersons.
The observed positive relationship between reward dis- Our goal in this article is to model and demonstrate that
persion and performance suggests that firms may gain by the dispersion levels of particular variables are of concep-
having a compensation scheme that is not only more dis- tual and empirical significance beyond those of the mean
criminating across high, average, and low performers but levels of these variables. Our empirical findings support
is also more transparent. The negative relationship our premise in a number of instances. It is possible that in
between prove orientation and performance suggests that addition to the main effects, the mean and dispersion levels
the frequent monitoring and impression management of a focal variable may interact with each other on the
activities by only a subset of salespersons in a unit may be dependent variables. This approach is likely to be interest-
bothersome to their colleagues and bring down the entire ing and worthy of future research.
Venkatesh et al. / INFLUENCE OF HETEROGENEITY 251

The relationships between the antecedent and depend- Maintenance (ix = .72)
ent variables in our study are probably mediated by other 9. Holding my own against the competition of new sales
variables. For example, the positive relationship between reps entering the fieldb
gender diversity and satisfaction may be due to forces of 10. Attending meetings and seminars on new methods
attraction and camaraderie in mixed-gender units. These 11. Finding out about new opportunities as my field
factors might be responsible for a lack of competitive changes
spirit, thereby leading to lower performance. Future 12. Getting refresher training
research into such potential mediating mechanisms is
Disengagement (cx= .73)
likely to yield rich insights.
Our study is based on data from a single company. 13. Avoiding excess occupational pressures
While our unit-level analysis required us to focus on multi- 14. Developing more hobbies to replace work interests
ple units within the same company to enhance internal 15. Doing the things I've always wanted to do but never
validity, it would be valuable to replicate (or refute) the had the time because of work
findings in other contexts (e.g., salary vs. commission, 16. Having a good life in retirementb
Competence Dispersion (Adapted from Kohli and Jaworski 1994)
prevalence of sales contests, closeness of supervision, and
Managers reported the extent to which their sales unit was similar
so on), thus moving toward developing generalizable
or dissimilar on the following items:
findings. 1. Knowledge about selling
As noted earlier, we used a subset of items for measur-
2. Product knowledge
ing career stages of salespeople. If practical, it would be
3. Overall ability to perform key tasks
instructive to use the complete scale in future studies to see
4. Judgment of relevant business matters
if similar finds are obtained. Finally, although the reward
scale used in the study has an alpha of .76, it does not fully Rewards b (new scale) (a = .76)
capture the domain of intangible rewards. Future studies 1. Prizes and awards from my companyr
should use a more complete scale to measure tangible and 2. Recognition from my sales manager
intangible rewards (see, e.g., the scale developed by Tyagi 3. Salary + commissions
1985). 4. Overall compensation (including perks)
In conclusion, the present study provides some evi- Learning Orientation a (Drawn from Sujan, Weitz, and Kumar
dence of the effects of heterogeneity in sales units on the 1994) (or = .76)
satisfaction and performance of salespeople in those units. 1. There really are not a lot of new things to learn about
We believe the unit-level perspective has the potential to selling at (my company),b (R)
provide deeper insights than might be obtained by focus- 2. In my line of work, it is worth spending a great deal of
ing on the individual level alone. We hope some of the con- time learning new approaches for dealing with cus-
ceptualization and arguments introduced in this study pro- tomers.
vide some bases for future investigations employing the 3. It is important for me to learn from each selling expe-
unit-level perspective. rience I have.
4. An important part ofbeing a good sales rep is continu-
Appendix ously improving your selling skills.
Scale Items 5. Learning how to be a better sales rep is of fundamentai
importance to me.
Career Stagesa (Drawn from Cron and Slocum 1986) Prove Orientation a (Drawn from Sujan et al. 1994) (or= .74)
Exploration (a = .94) 1. I feel very good when I know I have outperformed
1. Deciding what I really want to do for a living other sales reps in my district office.
2. Finding a line of work that really appeals to me 2. It is very important to me that my district manager
3. Finding opportunity to do the kind of work I really sees me as a productive sales rep.
like to do 3. I very much want fellow sales reps in my district of-
4. Making sure of my current occupational choice fice to consider me to be good at selling.
4. I evaluate myself using my district manager's criteria.
Establishment (a = .73)
5. It is important for me to perform better than other reps
5. Developing a reputation in my company in my district office.
6. Becoming especially skillful in my work 6. I always try to communicate my sales accomplish-
7. Finding ways of making my competence known ments to my district manager.
8. Advancing to a higher position in my company 7. I spend a lot of time thinking about how my perfor-
mance compares with that of other sales reps in my
district office.
252 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2001

Job Satisfaction (based on Brown and Peterson 1994) ((x = .86) with his or her total sales experience. There was no difficulty breaking the
tie in this manner.
1. I find my work exciting. 4. The studied sample consists of data from 77 sales units for which
2. I would advise a friend to sell for my company. we had at least a 50 percent response rate and for which we also had re-
3. I would recommend the company as a place to work. sponses from the unit managers. The statistical information for the stud-
4. Overall, I am satisfied with my job. ied sample is similar to that for the full sample.
5. We adopt the conventional approach of using significance levels of
Performance (new scale) 1 percent, 5 percent, and 10 percent to test our hypotheses as opposed to
Managers reported the extent to which their unit is much worse or setting the significance level based on a desired power of the test because
much better relative to other units in the organization on the fol- we do not know what the alternative is (except that it is not equal to the
null value).
lowing items:
6. To see why, consider the analogy ofthe"wealth of nations." Of two
1. Percentage of quota achieved nations, the one with a greater wealth per capita is arguably happier, all
2. Sales volume else the same. Yet, between two nations with the same wealth per capita,
3. Overall performance the nation with greater disparity (i.e., dispersion) in wealth distribution
may be expected to be less happy. In other words, the influences of the
NOTE: Unless indicated otherwise, salespeople are the respondents. Ca- mean and dispersion should be interpreted separately.
reer stages were measured on scale ranging from 1 (not at all concerned)
to 5 (very concerned). Competence dispersion items were measured at
the unit level using a 5-point scale ranging from very dissimilar to very
similar. Rewards (at the salesperson level) and performance (at the unit REFERENCES
level) were measured using a scale ranging from 1 (much worse) to 7
(much better). Learning and prove orientations and job satisfaction were Ames, Carole and Jennifer Archer. 1988. "Achievement Goals in the
measured using scales ranging from I (strongly disagree) to 5 (strongly Classroom: Students' Learning Strategies and Motivation Pro-
agree). (R) indicates reverse-scored items. cesses." Journal of Educational Psychology 80 (September):
a. Dispersions were computed for these measures using the procedure de- 260-267.
scribed in the Measures section. Ancona, Deborah G. and David F. Caldwcll. 1988. "Beyond Task and
b. This item was eliminated during the scale purification process. Maintenance: Defining External Functions in Groups." Group and
c. This item was deleted because of low item-to-total correlation. Organization Studies 13:468-494.
Aries, Elizabeth. 1976. "Interaction Patterns and Themes of Male, Fe-
male and Mixed Groups." Small GroupBehavior 7 (Fcbrnary): 7-18.
ACKNOWLEDGMENTS Bantel, Karen A. and Susan E. Jackson. 1989. "Top Management and In-
novations in Banking: Does the Composition of the Top Team Make a
Difference?"StrategicManagementJourna110 (Summer): 107-124.
The authors thank Steve Brown, Pradeep Chintagunta, Bcttencourt, B. Ann, Marilynn B. Brewer, Marian Rogers Croak, and
J a m e s Craft, A1 Silk, P a n d u T a d i k a m a l l a , L o i s Vargas, a n d Norman Miller. 1992. "Cooperation and the Reduction of Intergroup
Bias: The Role of Reward Structure and Social Orientation." Journal
p a r t i c i p a n t s at t h e s y m p o s i a at t h e H a r v a r d B u s i n e s s
of Experimental Social Psychology 28:301-319.
S c h o o l a n d t h e U n i v e r s i t y o f S o u t h e r n C a l i f o r n i a for t h e i r Bond, Charles F., Jr. 1982. "Social Facilitation: A Self-Presentational
h e l p f u l c o m m e n t s . T h e a u t h o r s are e s p e c i a l l y t h a n k f u l to View." Journal of Personality and Social Psychology 42 (6):
1042-1050.
the J A M S e d i t o r a n d t h r e e a n o n y m o u s r e v i e w e r s for t h e i r
Brown, Steven P. and Robert A. Peterson. 1994. "The Effect of Effort on
c o n s t r u c t i v e c o m m e n t s . R e s e a r c h s u p p o r t p r o v i d e d b y the Sales Performance and Job Satisfaction." Journal of Marketing 58
H a r v a r d B u s i n e s s S c h o o l is g r a t e f u l l y a c k n o w l e d g e d . A l l (April): 70-80.
t h r e e a u t h o r s c o n t r i b u t e d e q u a l l y to t h e article. Butler, R. 1993. "Effects of Task and Ego-Achievement Goals on Infor-
mation-Seeking during Task Engagement." Journal of Personality
and Social Psychology 65:18-31.
Churchill, Gilbert A., Jr., Ncil M. Ford, and Orville C. Walker, Jr. 1983.
NOTES "Organizational Climate and Job Satisfaction in the Salesforce:'
Journal of Marketing Research 13 (November): 323-332.
1. Fortune 500 companies, including those mentioned above, tend to Colarelli, Stephen M. and Amy L. Boos. 1992. "Sociometric and Abil-
ity-Based Assignment to Work Groups: Some Implications for Per-
have two types of sales forces---those that call on national or key ac-
sonnel Selection." Journal of Organizational Behavior 13 (2):
counts and those that work with relatively smaller accounts. National- 187-196.
account salespeople typically engage in team selling, while others do not Craig, Jane M. and Carolyn W. Sherif. 1986. "The Effectiveness of Men
work in teams for the most part. The focus of this study is on the latter, and Women in Problem-Solving Groups as a Function of Group Gen-
that is, salespeople who are engaged in individual selling from a branch der Composition." Sex Roles 14 (April): 453-466.
office rather than in team selling. Cron, William L. 1984. "Industrial Salesperson Development: A Career
2. More important, we recognize that additional factors such as com- Stages Perspective." Journal of Marketing 48 (Fall): 41-52.
petence and mean career stage in the sales unit are likely to affect sales - - and John W. Slocum, Jr. 1986. "The Influence of Career Stages
unit satisfaction and performance; we control for these factors in our on Salespeople's Job Attitudes, Work Perceptions and Performance."
analysis discussed later. We advance Hypotheses 1 through 12 focusing
Journal of Marketing Research 23 (May): 119-129.
, Alan J. Dubinsky, and Ronald E. Michaels. 1988. "The Influ-
on heterogeneity on a ceteris paribus basis. ence of Career Stages on Components of Salesperson Motivation"
3. There were very few cases in which the highest score on a particu- Journal of Marketing 52 (January): 78-92.
lar career stage scale was followed by a very close second score that was Deutsch, Morton. 1949. "A Theory of Competition and Cooperation."
two or more categories off. Specifically, this situation arose in 13 cases Human Relations 2:129-152.
out of 476 responses (2.73% of the total responses). In each of these in- Dillon, William R. and Matthew Goldstein. 1984. MultivariateAnalysis:
stances, we assigned the respondent to the career stage commensurate Methods and Applications. New York: John Wiley.
Venkatesh et al. / INFLUENCE OF HETEROGENEITY 253

Dweck, Carol S. and Ellen L. Leggett. 1988. "A Social-Cognitive Ap- O'Reilly, Charles A., III, David F. Caldwell, and William P. Barnett.
proach to Motivation and Personality." PsychologicalReview 95 (2): 1989. "Work Group Demography, Social Integration, and Turnover"
256-273. Administrative Science Quarterly 34 (1): 21-37.
Elder, Glen H. 1975. "Age Differentiation and the Life Course." Annual Peterson, Mark F. 1983. "Co-Workers and Hospital Staff's Work Atti-
Review of Sociology 1 (1): 65-190. tudes: Individual Difference Moderators." Nursing Research 32 (2):
Elliot, Andrew J. and Marcy A. Church. 1997. "A Hierarchical Model of 115-121.
Approach and Avoidance Achievement Motivation?' Journal of Per- Pfeffer, Jeffrey. 1983. "Organizational Demography." In Research in Or-
sonality and Social Psychology 72 (1): 218-232. ganizational Behavior, Vol. 5. Eds. L. L. Cummings and Barry M.
- - and Judith M. Harackiewicz. 1996. "Approach and Avoidance Staw. Greenwich, CT: JAI, 299-357.
Achievement Goals and Intrinsic Motivation: A Mediational Analy- - - and Allison Davis-Blake. 1992. "Salary Dispersion, Location in
sis?' Journal of Personality and Social Psychology 70 (3): 461-462. the Salary Distribution, and Turnover Among College Administra-
tors." Industrial and Labor Relations Review 45 (4): 753-763.
Ely, Robin J. 1994. "The Effects of Organizational Demographics and
and Nancy Langton. 1988. "Wage Inequality and the Organiza-
Social Identity on Relationships Among Profession Women." Admin-
- -

tion of Work: The Case of Academic Departments?' Administrative


istrative Science Quarterly 39 (2): 203-238.
Science Quarterly 33:588-606.
Farr, James L., David A. Hofmann, and Kathleen L. Ringenbach. 19939 - - and .1993. "The Effect of Wage Dispersion on Satisfac-
"Goal Orientation and Action Control Theory: Implications for In- tion, Productivity, and Working CoUaboratively: Evidence From
dustrial and Organizational Psychology." In International Review of College and University Faculty:' Administrative Science Quarterly
Industrial and Organizational Psychology. Eds. C. L. Cooper and I. 38 (3): 382-407.
T. Robertson. New York: John Wiley, 139-232. Pierce, Charles A. 1998. "Factors Associated With Participating in a Ro-
Festinger, Leon. 1954. "A Theory of Social Comparison Processes." Hu- mantic Relationship in a Work Environment?' Journal of Applied So-
man Relations 7 (May): 117-140. cial Psychology 28 (18): 1712-1730.
Fields, Dail L. and Terry C. Blum. 1997. "Employee Satisfaction in Work Salancik, Gerald R. and Jeffrey Pfeffer. 1978. "A Social Information Pro-
Groups With Different Gender Composition9 Journal of Organiza- cessing Approach to Job Attitudes and Task Design?'Administrative
tional Behavior 18 (2): 181-196. Science Quarterly 23 (2): 224-253.
Hackman, Richard J. and Greg R. Oldham. 1980. WorkRedesign. Read- Sauers, Daniel A. and Ken Bass. 1990. "Sustaining the Positive Effects of
ing, MA: Addison-Wesley. Goal Setting: The Positive Influence of Peer Competition." Akron
Harackiewicz, Judith M., Kenneth E. Barton, Suzanne M. Carter, Alan T. Business and Economic Review 21 (4): 30-40.
Lehto, and Andrew J. Elliot. 1997. "Predictors and Consequences of Schneider, Benjamin. 1983. "Interactional Psychology and Organiza-
Achievement Goals in the College Classroom: Maintaining Interest tional Behavior" In Research in Organizational Behavior. Eds. L. L.
and Making the Grade." Journal of Personality and Social Psychol- Cummings and B. M. Staw. Greenwich, CT: JAI, 1-131.
ogy 73 (6): 1284-1295. 9 1987. "The People Make the Place?' Personnel Psychology
Jackson, Susan E., Joan E Brett, Valerie I. Sessa, Dawn M. Cooper, Johan 40:437-454.
A. Julin, and Karl Peyronnin. 1991. "Some Differences Make a Dif- Sujan, Harish, Barton Weitz, and Nirmalya Kumar. 1994. "Learning Ori-
ference: Individual Dissimilarity and Group Heterogeneity as Corre- entation, Working Smart, and Effective Selling?' Journal of Mar-
lates of Recruitment, Promotions, and Turnover." Journal of Applied keting 58 (July): 39-52.
Psychology 76 (5): 675-689. Super, Donald, Robin S. Zclkowitz, and Albert S. Thompson. 1981.
Johnston, Jack and John DiNardo. 1997. Econometric Methods. New CcareerDevelopmentInventory:Adult Form I. New York: Teacher's
York: McGraw-Hill. College, Columbia University.
Kent, R. N. and J. E. McGrath. 1969. "Task and Group Characteristics as Tesser, A. 1988. "Toward a Self Evaluation Maintenance Model of Social
Factors Influencing Group Performance." Journal of Experimental Behavior." In Advances in Experimental Social Psychology, Vol. 21.
Social Psychology 5 (4): 429-440. Ed. L. Berkowitz. San Diego, CA: Academic Press, 181-227.
Kohli, Ajay K. and Bernard J. Jaworski. 1994. "The Influence of Tyagi, Pradeep K. 1985. "Organizational Climate, Inequities, and Attrac-
Coworker Feedback on Salespeople." Journal of Marketing 58 (Oc- tiveness of Salesperson Rewards?' Journal of Personal Selling &
tober): 82-94. Sales Management 5 (2): 31-37.
, Tasadduq A. Shervani, and Goutam N. Challagalla. 1998. VandeWalle, Don and Larry L. Cummings. 1997. "A Test oftbe Influence
"Learning and Performance Orientation of Salespeople: The Role of of Goal Orientation on the Feedback Seeking Process?' Journal of
Supervisors?' Journal of Marketing Research 35:263-274. Applied Psychology 82 (3): 390-400.
Lawler, Edward E. 1973. Motivation in WorkOrganizations. Monterrey, Vroom, Victor H. and Bernd Pahl. 1971. "Relationship Between Age and
Risk Taking Among Managers." Journal of Applied Psychology 55
CA: Brooks/Cole.
(5): 399-405.
Leonard, Richard. 1990. "ESPOs Bring Participation Home?' Manage-
Weinstein, Alan G. and Robert L. Holzbach. 1973. "Impact of Individual
ment Review 79 (11): 45-49. Differences, Reward Distribution, and Task Structure on Productiv-
Leventhal, G. S. 1976. "The Distribution of Rewards and Resources in Or- ity in a Simulated Work Environment." Journal of Applied Psychol-
ganizations" In Advances in ExperimentalSocial Psychology, Vol. 9. ogy 58 (3): 296-301.
Ed. Leonard Berkowitz. New York: Academic Press, 91-131. Wheeler, Ladd. 1991. "A Brief History of Social Comparison Theory." In
McCann, Douglas C., Thomas M. Ostrom, and Linda K. Tyner. 1985. Social Comparison: Contemporary Theory and Research. Eds. Jerry
"Person Perception in Heterogeneous Groups?' Journal of Personal- Suls and Thomas Ashby Wills. Hillsdale, NJ: Lawrence Erlbaum,
ity and Social Psychology 49 (December): 1449-1459. 3-21.
McGrath, Joseph E. and David Brinberg. 1983. "External Validity and the Wiersema, Margarethe F. and Allan Bird. 1993. "Organizational Demog-
Research Process: A Comment on the Calder/Lynch Dialogue?' Jour- raphy in Japanese Firms: Group Heterogeneity, Individual Dissimi-
nal of Consumer Research 10 (June): 115-124. larity, and Top Management Team Turnover." Academy of
Milliken, Frances and Luis L. Martins. 1996. "Searching for Common Management Journal 36 (October): 996-1025.
Threads: Understanding the Multiple Effects of Diversity in Organi- Wyer, Robert S. and Christine Malinowski. 1972. "Effects of Sex and
zational Groups?'Academy of Management Journal 21 (2): 402-433. Achievement Level Upon Individualism and Competitiveness in So-
Mitchell, Daniel J. B., David Lewin, and Edward E. Lawler III. 1990. cial Interaction." Journal of Experimental Social Psychology 8 (4):
"Alternative Pay Systems, Firm Performance, and Productivity?' In 303-314.
Payingfor Productivity:A Look at the Evidence. Ed. Alan S. Blinder. Zenger, Todd R. and Barbara S. Lawrence. 1989. "Organizational De-
Washington, DC: Brookings Institution, 15-88. mography: The Differential Effects of Age." Academy of Manage-
Nunnally, Jura C. 1978. PsychometricTheory. New York: McGraw-Hill. ment Journal 32 (2): 353-376.
254 JOURNALOF THE ACADEMYOF MARKETINGSCIENCE SUMMER 2001

ABOUT THE AUTHORS Goutam Challagalla is an associate marketing professor at


Georgia Tech. He received his Ph.D. in marketing from the Uni-
R. Venkatesh is an assistant professor of business administration versity of Texas at Austin, where he won the Outstanding Disser-
at the University of Pittsburgh's Katz Graduate School of Busi- tation Award. He has published articles on sales management and
ness. His articles on product bundling, cobranding and sales marketing theory in marketing and psychology journals.
force management have appeared or are forthcoming in the Inter-
national Journal of Research in Marketing, Journal of the Acad- Ajay K. Kohli is Isaac Stiles Hopkins professor of marketing at
emy of Marketing Science, Journal of Business, Journal of Mar- Emory University's Goizueta Business School. During 2000-
keting, Journal of Marketing Research, and Marketing Science. 2001, he is on leave from Emory and working at the Monitor
Company. He has published in several journals on market orien-
tation, sales management, and organizational buying behavior.

You might also like