Escolar Documentos
Profissional Documentos
Cultura Documentos
.
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.
American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to
Journal of Marketing.
http://www.jstor.org
The
This article reviews pertinent research related to
the antecedents of turnover and discusses a research study involving a national industrial firm's
salesforce. The study examined the moderating
effect of salespeople's performance on the relationship between their job satisfaction and propensity to leave their firm. The paper concludes
with implications for researchers and sales managers.
of
Relationship
and
Satisfaction
to
Performance
Salesforce
Turnover
Previous Research
of
ITHthecost, time, andenergyrequirements
of MarProfessors
areAssociate
M.Futrell
andA.Parasuraman
Charles
The
TexasA&MUniversity.
Administration,
keting,
Collegeof Business
H.Mobleyfor his helpfulsuggestions,
authors
wishto thankWilliam
ofthesectioneditorandthree
comments
theconstructive
andappreciate
reviewers.
anonymous
Journal of Marketing
Vol. 48 (Fall 1984), 33-40.
ior.
andPerformance
of Satisfaction
/ 33
TheRelationship
Research Hypothesis
The purpose of this study was to test the following
hypothesis in a salesforce context:
Performancewill have a significant moderating effect on the relationship between job
satisfaction and propensity to leave. Specifically, the sensitivity of propensityto leave to
job satisfactionwill be significantlyhigher for
low performersthan for high performers.
This hypothesis is consistent with the discussion presented earlier, especially the empiricalfindings of the
Spencer and Steers study (1981). Furthermore,such
a hypothesisis quite plausible in a salesforce context.
When a salesperson performs poorly on the job,
we would expect the sales manager to make few attemptsto keep the employee. In fact, pressuresto increase performancemay result in continuousnegative
feedback and even disciplinaryaction, such as being
given a probationaryperiod to improve performance.
The salespersonwould have little reasonto stay unless
he/she was satisfied with the job, and liked the work
itself and his/her co-workers. Should this be the case,
we could expect an inverse relationshipbetween job
satisfactionand turnoverof low performers.
Inabilityto find anotherjob is anotherreason why
a low performermight not quit a job. Nevertheless,
pressuresto improve work performanceshould force
the employee to continue searchingfor a new job, and
hence propensityto leave could still be high, unless
job satisfactionis also high. For the salesperson performing above average, the individual's sales manager is likely to provide positive feedback and strive
to acknowledgejob performancethrough pay raises,
praise, recognition in front of peers, etc. Hence, for
the high performingsalesperson,job satisfactionmay
be relatively less importantin a decision to stay or
leave the job. The employee would experiencenumerous motivationsto stay on the job, with job satisfaction being only one of them. The association of job
satisfactionwith propensityto leave can be expected
to be much strongerfor low performersthan for high
performers.
Research Method
Study Sample
The total salesforce and the salespeople's immediate
supervisors in a national pharmaceuticalfirm were
surveyedby the use of two mail questionnaires.The
salespeople were sent one questionnaireto measure
theirjob satisfaction and propensity to leave. Supervisors were sent a different questionnairethat asked
them to evaluate their subordinates' job performances. The national sales managerof the cooperat-
Fall1984
of Marketing,
34 / Journal
ing firm provided a letter encouraging the sales supervisorsto cooperatewith the study. The salespeople
and supervisorswere told that the survey was independent of their employer and were promised anonymity. The questionnaireswere numbercoded (necessary to match salespeople's replies with the
supervisors'ratings), and a follow-up postcard was
used to increasethe returnsfrom those not responding
initially. Two follow-up letters were sent to supervisors who did not respond initially.
A total of 508 salespeople and 52 sales supervisors
were polled. The final response rates were 52% and
98% respectively, and performance ratings were
availablefor all 263 salespeople who responded.Thus
the study was based on 263 sets of matched, usable
questionnaires.The respondents'ages, experience,and
educationallevels were also measured. According to
officials in the firm, these measured demographic
characteristicsof the respondentswere similar to the
characteristicsof all salespeople in their firm. It was
not possible to get measures of job satisfaction and
propensityto leave for the nonrespondents.However,
since respondents and nonrespondentswere similar
on demographiccharacteristics(age, experience, and
education), and since research has shown that such
demographiccharacteristicsare associated with job
satisfaction and propensity to leave of salespeople
and Futrell 1983), any nonresponsebias
(Parasuraman
in the predictorandcriterionvariablesused in this study
was assumed to be negligible.
Job performancewas measuredusing the 10 items
on the firm's "districtsales manager'sappraisalform"
that the company used to measure its salespeople's
performance.Sales figures were not incorporatedinto
the salespeople's performanceappraisalbecause the
organization distributed its products through drug
wholesalersand did not sell directly to its customers.
The sales supervisorsbelieved that it was difficult to
appraisea salesperson'sperformanceusing actualsales
as a primaryindicatorof job performance.
The questionnairesentto the supervisorsaskedthem
to rate each of their salespeople on a 7-point poor to
excellent scale on the following 10 factors: willingness to work hard, current general attitude, current
sales ability, planning ability, activity reporting,current overall job performance,territorycoverage, improvementin totaljob performanceover last year, human relations ability, and product knowledge. In
makingtheseperformanceevaluations,supervisorswere
asked to consider the length of time each salesperson
had been with the company, so that each salesperson
would be evaluated on an equal basis.
Propensity to Leave
The immediateprecursorof actually quitting appears
to be intentionto leave. This is consistent with Fishbein's (1967) model of attitudes, intentions, and behavior,and with Locke's task motivationmodel, which
contendsthatthe individual'sintentionis the most immediate motivational determinantof choice (Locke
1968; Locke, Cartledge, and Knerr 1970). Numerous
studies have supportedthis contention by presenting
empirical evidence of a strong relationship between
intentions and withdrawal behavior (Kraut 1975;
Newman 1974; Porteret al. 1974; Porter, Crampon,
and Smith 1976; Steers 1977; Waters, Roach, and
Waters 1976).
Bluedor's (1982a) recent literaturereview cited
23 studies which reportedfinding significant positive
relationshipsbetween leaving intentions and actual
leaving behavior. Fifteen of these 23 studies allowed
the predictivepower of the intent to leave variable to
be comparedwith the predictive power of other variables. In 19 of 20 comparisonsmade in these 15 studies, intentto leave was the most accuratepredictorof
turnoverbehavior. These types of findings are even
promptingsome researchersto recommend using an
intent to leave measure, ratherthan actual staying or
leaving behavior, as a criterion variable (Coverdale
and Terborg 1980).
Propensityto leave the job was measured in this
study using the single question, "To what extent are
you presently seeking to change jobs?" Salespeople
were asked to check one of the following five responses: to no extent, to a small extent, to some extent, to a considerableextent, and to a very great extent. A single-item scale was used due to lack of any
standardmulti-item instrument,although it is recognized that a single-item scale may lead to some measurementerror. Moreover, past studies have invariably used single-item scales to measure propensityto
leave. For instance, a recent study by Busch (1980)
simply used a three-pointsingle-item scale to measure
salespeople'spropensityto leave. Thus the propensity
to leave measureused in the presentstudy was at least
as refinedas, if not betterthan, those used in previous
studies.
andPerformance
of Satisfaction
/ 35
TheRelationship
Data Analysis
Althoughthe general meaning of the term moderator
variable is perhapsstraightforward,there are several
conceptual and statistical problems associated with
formal analysis of such variables (Zedeck 1971). A
numberof researchersin the organizationalbehavior
and industrialpsychology areas have discussed appropriateways for analyzing and interpretingmoderatorvariables(Arnold 1982, Guion 1976, Petersand
Champoux1979, Saunders1956, Schmidtand Hunter
1978, Zedeck 1971). More recently, a useful frameworkfor identifyingmoderatorvariablesand isolating
different types of moderating effects has also appeared in the marketingliterature(Sharma, Durand,
and Gur-Arie 1981). Since this study deals with the
moderatingeffect of performanceon the job satisfaction-propensity to leave relationship, it is important
to emphasizeat the outset that only one type of moderatingeffect was the focus of this study. Specifically,
the study examined the moderatingeffect of performanceon the form (slope) of the job satisfaction-propensity to leave relationship because the study hypothesis was concerned with the sensitivity of propensityto leave to job satisfactionfor high performers
vis-a-vis low performers.
Table 1 contains the productmoment correlations
among the study variables, as well as the reliability
coefficientsfor the variousmultipleitem scales-i.e.,
the alpha coefficients proposed by Cronbach(1951).
The high alphacoefficients demonstrategood internal
consistencyfor the various measurementscales used.
Propensityto leave has a significant, negative correlation with each of the five JDI satisfaction dimensions. The size of these correlations, although moderate, are of the same order as, or higher than, the
correlationsobtained by other researchersinvestigating the association between satisfaction and turnover
(e.g., Spencerand Steers 1981). Moreover, these correlationsare perhapsin line with what one might expect, since many factors other than satisfaction can
also affectpropensityto leave (Porterand Steers 1973).
As suggested by Sharma, Durand, and Gur-Arie
(1981), the moderatingeffect of performanceon the
form (slope) of the satisfaction-propensity to leave
relationship was examined by conducting multiple
regressionanalyses of the criterionand predictorvariables for differentsubgroupsof the sample formed on
the basis of performance.Specifically, two subgroups
were formed by an approximatemedian split of the
sample, based on the total performance scores that
ranged between 22 and 69 for the individual salespeople. The two subgroups were (a) a "low performance" group of 130 salespeople, with performance
scores ranging from 22 to 53, and (b) a "high performance" group of 133 salespeople, with performancescoresrangingfrom 54 to 69. It mustbe pointed
out that the literatureon moderatorvariablesdoes not
offer any specific guidelines regardingthe numberof
subgroupsto be formed. For instance, it is not clear
whether two subgroups would be better than three
subgroups,or vice versa. The analysis proceduresand
the resultsdiscussed here pertainto the two subgroups
identifiedearlier.However, some of the analyses were
repeatedusing a three-way split of the sample to verify the stability of the results. This will be discussed
later.
Propensityto leave (Y) was regressed on the five
satisfactiondimensions (XI throughX5) using the following model:
Yi = ai + bliX1 + b2iX2 + b3iX3 + b4iX4 + b5iX5
TABLE1
Correlation Matrix and Alpha Reliability Coefficients for Study Variables
Correlation Matrix
Variable Names
JDI Satisfaction Measures:
Pay (X1)
X2
X3
X5
.48
.38
.36
.32
-.27
.44
.46
.38
.34
.39
.45
-.36
-.24
-.46
-.38
Promotion (X2)
Co-workers (X3)
Work (X4)
Supervision (X5)
Propensity to leave (Y)
Performance (Z)
No. of
Scale Items
Coefficient
Alpha
.09a
9b
.75
.03a
9b
.90
.85
.83
.82
.22
.09"
.03a
-.12a
18
18
18
1
10
aThese correlation coefficients are not significant at p = .05; all other correlation coefficients are significant at p < .001.
bThetotal scores on these measures were multiplied by two in accordance with the instructions of JDl's developers.
Fall1984
36 / Journalof Marketing,
.90
Findings
The results of the multiple regression analyses for the
total sample as well as for the two subgroups are summarized in Table 2. For the total sample, the five satisfaction dimensions together account for 27% of the
variance in propensity to leave. However, there is a
sharp difference between the high and low perforthe satmance groups in terms of R2 values-while
isfaction measures only account for about 19% of the
criterion variance in the high performance group, they
account for almost 38% of the criterion variance in
the low performance group. Furthermore, as recommended by Sharma, Durand, and Gur-Arie (1981), a
Chow test (Chow 1960) was conducted to see if the
set of coefficients in the regression for high performers was significantly different from the set of coefficients in the regression for low performers. Based on
the Chow test, the two sets of coefficients were significantly different beyond the .05 level of significance [F(5,257) = 2.40], suggesting thereby that
Independent Variable
Total Sample (n = 263)
Satisfaction with:
Pay (X,)
Promotion (X2)
Co-workers (X3)
Work (X4)
Supervision (X5)
High Performers (n = 133)
Satisfaction with:
Pay (X,)
Promotion (X2)
Co-workers (X3)
Work (X4)
Supervision (X5)
Low Performers (n = 130)
Satisfaction with:
Pay (X,)
Promotion (X2)
Co-workers (X3)
Work (X4)
Supervision (X5)
TABLE 2
Results of Multiple Regressions
Standard Error
Significance Level
Regression Coefficient
of "b" value
value
of "b"
("b" value)
I~~~~~~~~~~~~~~~
R2
-.010
.003
-.034
-.021
.005
.004
.007
.007
007
.530
.020
.705
.000
.003
.186 (p < .001)
.003
-.012
-.009
-.017
-.013
.007
.006
.011
.010
.009
.719
.035
.414
.091
.164
.379 (p < .001)
-.006
-.009
.016
-.050
-.027
.007
.006
.010
.011
.010
.429
.150
.095
.000
.011
andPerformance
of Satisfaction
TheRelationship
/ 37
- -
TABLE3
Mean Scores for High and Low Performers
High Performers
SD
Mean
Construct
Satisfaction with:
28.34
Pay
34.57
Promotion
46.74
Co-workers
38.77
Work
45.04
Supervision
1.55
Propensity to leave
aAllsignificancelevels are for two-tailedt-tests.
11.14
15.80
8.01
8.60
9.02
.90
Low Performers
SD
Mean
27.99
34.86
44.16
37.68
44.95
1.76
T-value
Sig. Level'
.23
.15
2.38
.99
.08
1.70
.82
.88
.02
.32
.94
13.55
16.45
9.48
9.25
9.15
1.11
Fall1984
of Marketing,
38 / Journal
.09
ple deliberatelywent to work for MerrillLynch to receive their trainingand gain experience, and then left
to work with competitors. The firm developed a programaimed at identifying their most valuable brokers
insteadof attemptingto lower overall turnover.They
classified all brokers according to length of service
and performance.Once grouped, the high performing
brokers were targeted for special retention efforts.
Turnoverdroppedfrom 8% above the industry average to 11%below industryaverage.Furthermore,most
of the brokerswho quit were below average performers.
Too many sales managers may be needlessly devoting resources to their entire salesforce personnel
instead of determiningthe salespeople they want to
retainand finding out what needs to be done to retain
them. Sales organizations should begin to manage
turnoverstrategicallyandselectively.Ourfindingsshow
thatimprovingthe work satisfactionof salespeoplemay
reduce turnoverof low performersbut may do little
to reduceturnoveramonghigh performers.On the other
hand, visual forms of recognition, such as promotions, awards, etc., may increase the chances of retaining high performers. More research needs to be
done to identify other factors that may be critical in
lowering turnoveramong high performing salespeople.
Replication of this study in different salesforce
settings is essential before making conclusive statementsaboutthe role of performancein turnovermodels,
especiallysince situation-specificfactors,differentfrom
those present in our study, may well influence turnover. Furthermore,it is also worthexploringthe causal
linkage, if any, between salespeople's satisfactionand
their performance.For instance, will an increase in
work satisfactionalso improvethe performanceof low
performers,in addition to significantly reducing their
propensityto leave? Although such questions could
not be examined in our study, focusing on them in
future studies can be valuable in formulating strategies for effectively managing salesforce turnover.
REFERENCES
Allison, P. D. (1974), "Inter-organizationalMobility of Academic Scientists," paper presented at the 69th annual
meeting of the American Sociological Association, Montreal.
Arnold, Hugh J. (1982), "Moderator Variables: A Clarification of Conceptual, Analytic, and Psychometric Issues,"
andPerformance
of Satisfaction
TheRelationship
/ 39
and Influence Upon the Sales Force," Journal of Marketing, 44 (Summer), 91-101.
Chow, G. C. (1960), "Test of Equality between Sets of Coefficients in Two Linear Regressions," Econometrika, 28
(July), 591-605.
Churchill, Gilbert A., Jr. and Anthony Pecotich (1982), "A
Structural Equation Investigation of the Pay Satisfaction
Valence Relationship among Salespeople," Journal of
Marketing, 46 (Fall), 114-124.
Coverdale, S. and J. R. Terborg (1980), "A Re-examination
of the Mobley, Homer, and Hollingsworth Model of Turnover: A Useful Replication," paper presented at the 40th
annual meeting of the Academy of Management, Detroit.
Cronbach, Lee J. (1951), "Coefficient Alpha and the Internal
Structureof Tests," Psychometrika, 16 (September), 297334.
Dreher, George F. (1982), "The Role of Performance in the
Turnover Process," Academy of Management Journal, 25
(March), 137-147.
Fishbein, M. (1967), "Attitude and the Prediction of Behavior," in Readings in Attitude Theory and Measurement, M.
Fishbein, ed., New York: Wiley.
Forrest, C. R., L. L. Cummings, and A. C. Johnson (1977),
"Organizational Participation: A Critique and Model,"
Academy of Management Review, 2 (October), 586-601.
Futrell, Charles M. (1979), "Measurement of Salespeople's
Job Satisfaction: Convergent and Discriminant Validity of
Corresponding INDSALES and Job Descriptive Index
Scales," Journal of Marketing Research, 16 (November),
594-597.
Guion, R. M. (1976), "Recruiting, Selection, and Job Replacement," in Handbook of Industrial and Organizational
Psychology, M. D. Dunnette, ed., Chicago: Rand McNally, 777-828.
Kraut, A. I. (1975), "Predicting Turnover of Employees from
Measured Job Attitudes," Organizational Behavior and
Human Performance, 13 (April), 233-243.
Locke, E. A. (1968), "Toward a Theory of Task Motivation
and Incentives," Organizational Behavior and Human Performance, 3 (May), 157-189.
(1976), "The Nature and Consequences of Job Satisfaction," in Handbook of Industrial and Organizational Psychology, M. D. Dunnette, ed., Chicago: Rand
McNally, 1297-1349.
, N. Cartledge, and C. S. Knerr (1970), "Studies of
the Relationship between Satisfaction, Goal-Setting, and
Performance," Organizational Behavior and Human Performance, 5 (March), 135-158.
Martin, T. N., J. L. Price, and C. W. Mueller (1981), "Research Note on Job Performance and Turnover," Journal
of Applied Psychology, 66 (February), 116-119.
Mobley, W. H. (1977), "Intermediate Linkages in the Relationship between Job Satisfaction and Employee Turnover," Journal of Applied Psychology, 62 (April), 237-240.
(1982a), "Some Unanswered Questions in Turnover
and Withdrawal Research," The Academy of Management
Review, 7 (January), 114.
(1982b), Employee Turnover: Causes, Consequences, and Control, Reading, MA: Addison-Wesley.
, R. W. Griffeth, H. H. Hand, and B. M. Meglino
(1979), "Review and Conceptual Analysis of the Employee
TurnoverProcess," Psychological Bulletin, 86 (May), 493522.
Fall1984
of Marketing,
40 / Journal