Journal of Applied Psychology Copyright 1998 by the American Psychological Association, Inc.

1998, Vol. 83, No. 6, 922-931 0021-9010/98/$3.00

The Role of Organizational Citizenship Behavior in Turnover:

Conceptualization and Preliminary Tests of Key Hypotheses

Xiao-Ping Chen Chun Hui and Douglas J. Sego

Indiana University Hong Kong University of Science and Technology

The authors conceptualized levels of organizational citizenship behavior (OCB) perfor-

mance as a behavioral predictor of employee turnover and empirically examined the
strength of this relationship. Data were collected from 205 supervisor-subordinate dyads
across 11 companies in the People's Republic of China. The results provided considerable
support for the hypothesis that supervisor-rated OCB was a predictor of subordinates'
actual turnover. In particular, subordinates who were rated as exhibiting low levels of
OCB were found to be more likely to leave an organization than those who were rated
as exhibiting high levels of OCB. The authors also found that the self-report turnover
intention was a predictor of turnover, but this relationship did not hold for 2 companies.
The explanations and implications of these findings are discussed.

Employee turnover has received much theoretical and 1985; Rusbult & Farrell, 1983; Thibaut & Kelley, 1959).
empirical attention in organizational behavior and human At the same time, much of the empirical research on
resource management studies for several decades (Dai- turnover has focused on self-report of turnover intention
ton & Todor, 1987; Horn & Griffeth, 1995; March & as a predictor of turnover (Horn & Griffeth, 1995).
Simon, 1958; McEvoy & Cascio, 1987; Mobley, 1977; Receiving much less theoretical and empirical attention
P. C. Smith, Kendall, & Hulin, 1969). Early research on is the possible relationship between behavioral anteced-
turnover focused on the identification of different anteced- ents and turnover. Behavioral antecedents are those avoid-
ents to turnover (Mobley, 1977; Muchinsky & Morrow, ance behaviors that reduce the inclusion of the individual
1980; Steers & Mowday, 1981), and many conceptual in the work role (Katz & Kahn, 1966, 1978), such as
models have been proposed to describe the employee ter- lateness, tardiness, and absenteeism. Furthermore, the re-
mination-decision process. For example, some models search that has been conducted to investigate this relation-
emphasize the central role of job affects such as job satis- ship has not proved to be very promising. Indeed, the
faction and organizational commitment in determining majority of the studies conducted have not shown lateness,
turnover intention and turnover (Mowday, Porter, & absenteeism, or tardiness to be good predictors of turnover
Steers, 1982; Price & Mueller, 1986); some models em- (Benson & Pond, 1987; Miller, 1981,1982; Rosse, 1988).
phasize the importance of cognitive processes, that is,
Perhaps the reason that these forms of avoidance-with-
comparing the value of costs and benefits of the current
drawal behavior are not good predictors of turnover is
job to one's aspiration level in determining both job affect
that they are not discretionary; that is, they are constrained
and termination decisions (Hulin, Roznowski, & Hachiya,
by the organization, are part of the organizational reward
system, or both. However, if one expands the definition
Xiao-Ping Chen, Department of Management, Indiana Univer- of avoidance and withdrawal behaviors to include behav-
sity; Chun Hui and Douglas J. Sego, Department of Management iors that are discretionary in nature and are not necessarily
of Organizations, Hong Kong University of Science and Technol- expected as part of the employee's formal organizational
ogy, Hong Kong, China. role, one may find much stronger relationships between
Binding for this project was provided by the Direct Allocation avoidance-withdrawal behaviors and turnover than pre-
Grant from the Hainan Institute at the Hong Kong University of viously demonstrated.
Science and Technology. We thank Orlando Behling, George
One form of discretionary behavior that employees
Milkovich, and William Mobley for their constructive comments
on a draft of this article.
might decide to withhold when they become disillusioned
Correspondence concerning this article should be addressed or dissatisfied with the organization is their extrarole be-
to Xiao-Ping Chen, Department of Management, Indiana Univer- haviors or organizational citizenship behaviors (OCBs).
sity, 1309 East Tenth Street, Bloomington, Indiana 47405-1701. Rosse and Miller (1984) have pointed out that dissatisfied
Electronic mail may be sent to xpchen@indiana.edu. employees would experiment with avoidance responses

922
 OCB AND TURNOVER 923

that have the fewest negative consequences. OCBs are ary. In addition, there are several other reasons why low
"the behaviors of a discretionary nature that are not part levels of OCB are better manifestations of behavioral
of employees' formal requirements" (Organ, 1988, p. 4), withdrawal than absenteeism, lateness, or tardiness. First,
so exhibiting low levels of OCB will have no direct impact OCB has a generalized value and significance that cuts
on employees' monetary payoffs or living conditions. across different jobs and work organizations (Borman &
Therefore, a person's lack of willingness to exhibit such Motowidlo, 1993), whereas absenteeism, lateness, or tar-
behaviors may indicate his or her withdrawal from the diness are more likely to be constrained by limitations of
organization. For example, although an employee who is ability or by work processes that can vary from job to
disengaging from the organization (for whatever reason) job and from organization to organization. Second, OCB
might not reduce his or her in-role performance (because represents a class of behaviors that relate to many aspects
these are not really discretionary), he or she might decide of employees' work lives (Bateman & Organ, 1983),
to reduce the amount of "citizenship" behaviors they whereas absenteeism, lateness, or tardiness stand as single
perform (because they are discretionary in nature). behaviors that may not represent a specific manifestation
Although OCB is discretionary and not directly or ex- of a more general underlying construct of withdrawal
plicitly recognized by the formal reward system, employ- (Hulin, 1991; Rosse & Hulin, 1985; Rosse & Miller,
ees sometimes do exhibit this type of behavior. High levels 1984). Finally, some aspects of OCB (e.g., conscientious-
of OCB may be a result of job satisfaction, organizational ness) seem to be related to absenteeism, tardiness, or
commitment, or both (for a review, see Organ & Ryan, lateness. Thus, it is also possible that OCB is a "better"
1995) because commitment can be viewed as "binding predictor of turnover because it includes absenteeism, tar-
to action" (Salancik, 1983); or they may be a result of diness, and lateness and other behaviors as well.
strong intention to obtain good performance evaluations, OCB has multifaceted manifestations. For example, one
because extrarole performance may influence employees' type of OCB—altruism—is the act of helping others on
overall performance evaluations (Podsakoff & MacKen- work-related matters. Another type of OCB is conscien-
zie, 1994; Podsakoff, McKenzie, & Hui, 1993) and deci- tiousness, or the willingness to perform beyond the mini-
sions about their promotion, training, and compensation mal requirement in the areas of attendance, taking breaks,
(Park & Sims, 1989). No matter what the underlying and working overtime. These behaviors reflect employees'
reason is, it is conceivable then that low levels of OCB willingness to be actively involved in, or to be part of,
indicate a low level of commitment (or involvement) in the organization from different aspects. Therefore, levels
the organization, and, moreover, for those who think of of OCB imply the distance an employee would like to
exhibiting OCB as a "prepayment" for eventual good keep between himself or herself and the organization: The
treatment, low levels of OCB will be especially useful in lower the level of OCB, the larger the distance; conse-
predicting their turnover because they will be loath to quently, the lower the level of OCB, the higher the likeli-
"write o f f ' ' that investment if they do not have the inten- hood of turnover.
tion to leave. The relationship between OCB and turnover has not
OCB has been studied extensively as a dependent vari- been well documented empirically. Although Rosse and
able in examining its relationship with transformational Hulin (1985) included a group of avoidance behaviors
leadership (Podsakoff, MacKenzie, & Bommer, 1996), (some of them appeared to be similar to OCBs) in their
job affect (Bateman & Organ, 1983), and dispositional study, they conceptualized these behaviors only as reac-
characteristics (Barrick & Mount, 1991). However, it has tions to job dissatisfaction and thus did not examine their
not been integrated as an important component in research impact on turnover. Indirect evidence is found in a study
on turnover. We attempted, in this study, to conceptualize by Wells and Muchinsky (1985), in which they found
behavioral antecedents to turnover in terms of low levels that supervisors rated employees who quit their organiza-
of OCB and then to examine its predictive power on em- tions as less reliable and less dependable than those who
ployees' actual turnover. Specifically, we examined were promoted. Although Wells and Muchinsky did not
whether employees who exhibit low levels of OCB are define the concepts of reliability and dependability, they
more likely to leave an organization than those who ex- appear to be more like extrarole than in-role characteris-
hibit high levels of OCB. tics. In fact, this concept is similar to one type of OCB—
conscientiousness. These findings suggest that OCB is
Organizational Citizenship Behavior and Turnover negatively related to actual turnover.
On the basis of the theoretical linkage and suggestive
As mentioned earlier, we expect a stronger relationship empirical relationship between OCB and turnover, we hy-
between OCB and turnover than between other with- pothesized that OCB is a direct predictor of turnover:
drawal behaviors and turnover because OCB is discretion- Employees who exhibit low levels of OCB are more likely
924 CHEN, HUI, AND SEGO

to leave an organization than those who exhibit high levels The sample was 51% male and had a mean age of approximately
of OCB. 28.7 years. Sixty-one percent of the sample were unmarried.
The average job tenure for the sample was 2 years. Short job
tenure is typical of the region because most of these firms had
Turnover Intention, Organizational Citizenship
just recently been established. In this sample, 7% had not com-
Behavior, and Turnover pleted high school, 43% had a high school degree, 7% had
attended some college, and the remaining 43% had received a
To examine the predictive power of OCB on turnover,
4-year undergraduate or more advanced degree. This sample
it is important to examine the relationship between OCB
was deemed by the managers to be highly qualified educationally
and turnover intention, because turnover intention has
for the region.
been assumed and found to be the only antecedent having
a direct effect on actual turnover (Coverdale & Terborg,
1980; Fishbein & Ajzen, 1974; Michaels & Spector, 1982; Procedure
Miller, Katerberg, & Hulin, 1979; Mobley, 1977; Mobley, Data collection involved one or two personal trips to each
Homer, & Hollingsworth, 1978; Mowday, Koberg, & data collection site. During the first trip, we interviewed vice
McArthur, 1984). After reviewing the literature on em- presidents, managers, and personnel officers of 7 of the 11 com-
ployee turnover processes, Mobley, Griffeth, Hand, and panies in the sample to gain a better understanding of the compa-
Meglino (1979) concluded that "behavioral intentions to nies in the study. This was to ensure that a questionnaire relevant
stay or leave are consistently related to turnover behav- to the sample would be created. During the second trip, we
were there to collect the research data personally because we
ior" (p. 505). Indeed, turnover intention has consistently
had learned from previous experience that in collecting quality
been demonstrated to have a significant and positive rela-
research data in China, the presence of research professionals
tionship with turnover, with the average coefficient being is often important.
+.38 (Carsten & Spector, 1987). On the other hand, such Questionnaires were distributed personally to each respon-
a relationship has not been found between different types dent. To assure all respondents that their individual responses
of withdrawal behavior (such as lateness and voluntary would not be revealed to members of their working unit, we
absence) and actual turnover (Miller, 1981, 1982). We collected questionnaires immediately after completion. Respon-
argue that the lack of such a relationship is due to the dents were asked to identify, to the researchers, who their imme-
fact that (a) these behaviors are constrained by the organi- diate supervisors were. Researchers then approached these su-
zational reward system or by the work process and (b) pervisors for the OCB part of the assessment. The supervisors
they affect employees' monetary payoff. However, when were also informed that their evaluations would be kept confi-
dential. Three (and 10) months after the survey, each company
we use OCB as an index of measuring behavioral with-
was contacted to identify the workers who had left the
drawal, these limitations will be removed. Moreover, lev-
organization.
els of OCB reflect employees' true willingness regarding
how much they want to be involved in, or how much they
like to be distant from, the organization. The lower the Measures
levels of OCB, the stronger the indication of the unwilling- Responses regarding turnover intention and other perceptions
ness of the employee to be part of the organization, and such as organizational commitment, job satisfaction, job re-
thus the higher possibility that the employee would leave wards, costs, and job alternatives were obtained from partici-
the organization. If turnover intention is the attitudinal pants. Immediate supervisors of these participants were asked
indication of staying or leaving, OCB is the behavioral to evaluate their subordinates' levels of OCB.
indication. Therefore, we hypothesized that in addition to All data collection instruments used in the present study were
written in Chinese. To ensure equivalence of the measures in the
turnover intention, OCB is a direct predictor of turnover.
Chinese and the English versions, back-translation from Chinese
into English was first performed (Brislin, 1980). The two trans-
Method lations revealed no substantive differences in the meanings of the
items. Two Chinese judges then personally reviewed all Chinese
Sample
translated items to ensure that they would be meaningful to
The sample for this study was composed of 205 leader- Chinese participants. Finally, a version of the questionnaire was
subordinate dyads. Data were collected from middle-level man- sent to the personnel office of the development corporation to
agers and technical workers in 11 companies in Hainan Prov- be examined by a local expert. All items were modified to fit
ince, a southern province in the People's Republic of China that into the 5-point Likert-scale format (1 = extremely disagree, 2
has witnessed considerable economic growth in recent years. = slightly disagree, 3 = neither agree nor disagree, 4 = slightly
These companies manufacture diversified products from con- agree, and 5 = extremely agree).
struction materials to computer diskettes to bottled mineral wa- Turnover intention. The 3-item scale from Camman, Fich-
ter. Among them, 4 were joint ventures and 7 were locally owned man, Jenkins, and Klesh (1979) was used to measure turnover
enterprises (none of them were state or government owned). intention: (a) "1 often think of leaving the organization," (b)
 OCB AND TURNOVER 925

"It is very possible that I will look for a new job next year," job investments, and job alternatives) that are related to turnover
and (c) "If I may choose again, I will choose to work for the (cf. Horn & Griffeth, 1995). Moreover, we treated all of these
current organization" (reverse coded). The internal consistency variables as control variables because we attempted to test the
coefficient (a) for this scale was .78. effectiveness of OCB in predicting actual turnover, in addition
OCB measures. Low levels of OCB were used as a proxy to the effects these variables may have on employees' turnover
for behavioral withdrawal in the present study. We adopted three behaviors.
of the five OCBs: (a) altruism—discretionary behaviors that Job satisfaction. In general, there are two types of job satis-
have the effect of helping a specific other person with an organi- faction measures: facet and overall. To simplify the question-
zationally relevant task or problem; (b) conscientiousness— naire, we used an overall measure in the form of a single item:
discretionary behaviors on the part of the employee in the areas "All things considered, I am satisfied with my current job."
of attendance, obeying rules and regulations, taking breaks, and Prior research has shown that the validity of this 1-item general
so forth; and (c) sportsmanship—willingness of the employee job satisfaction measure is comparable to that of facet measures
to tolerate less than ideal circumstances without complaining. involving multiple items such as the Minnesota Satisfaction
We chose these three types of OCB because, first, altruism Questionnaire (MSQ; e.g., Scarpello & Campbell, 1983).
and conscientiousness have been central categories of OCB, as Organizational commitment. Organizational commitment
evidenced by their inclusion in early OCB studies (C. A. Smith, was measured by adapting the 18-item scale developed by Porter,
Organ, & Near, 1983) and by the later demonstration of their Steers, Mowday, and Boulian (1974). The three items on inten-
validity in other cross-cultural settings (Farh, Barley, & Lin, tions to leave were excluded because we had a separate measure
1997), and second, from the discussion we had with some of for turnover intention. This scale was designed to assess an
the managers and scholars who were familiar with Chinese man- employee's identification with, and involvement in, an organiza-
agement, we found that altruism and conscientiousness, together tion. Previous research (e.g., Mowday, Steers, & Porter, 1979)
has demonstrated that this scale possesses adequate psychomet-
with sportsmanship, were deemed as most relevant to the sample
ric properties, and the data pertaining to its reliability and valid-
under study.
ity are generally positive. The internal consistency coefficient
The items for these three scales were developed by Podsa-
(a) for this scale was .87.
koff, MacKenzie, Moorman, and Fetter (1990). We used their
Job rewards. Job rewards refer to the positives and negatives
scale with minor modifications. Specifically, altruism was
associated with a job. hi the present study, this variable was mea-
measured with three items: (a) helps orient new employees
sured with three items adopted from the Rusbult and Farrell (1983)
even though it is not required as part of his or her job, (b)
study. These three items were (a) "AH things considered, there
always is ready to help or to lend a helping hand to those
are many good thing associated with my job," (b) "hi general,
around him or her, and (c) willingly gives of his or her time
my job is very rewarding," and (c) "This job has more positive
to help others who have work-related problems. Conscien-
aspects than most people have at their places of employment."
tiousness was also measured with three items: (a) is one of
The internal consistency coefficient (a) for this scale was .60.
my most conscientious employees, (b) believes in giving an
Job investment. Job investment refers to an employee's in-
honest day's work for an honest day's pay, and (c) takes
put into the job and was operationalized using very general
fewer breaks at work than other employees. Sportsmanship
terms. It was measured with two items adopted from Rusbult
was measured with five items (all reverse coded): (a) con-
and Farrell (1983): (a) "In general, I put a lot into my job"
sumes a lot of time complaining about trivial matters, (b)
and (b) "In general, I have invested a lot into my job." The
always finds fault with what the company is doing, (c) is the internal consistency coefficient (a) for this scale was .69.
classic "squeaky wheel" that always needs greasing, (d) Job alternative. Job alternative refers to the assessment of
tends to make "mountains out of molehills," and (e) always the ease of movement and the favorability of job alternatives.
focuses on what's wrong with his or her situation, rather than Although many researchers treat ease of movement (job mobil-
the opposite. The internal consistency coefficients (a) for ity) and perceived job alternative as two separate concepts, our
these three scales were .87, .75, and .69, respectively. factor analysis revealed that both loaded on the same factor.
Actual turnover. Actual turnover data (voluntary turnover) Thus, job alternative was measured with six items adapted from
were obtained from company records 3 and 10 months after Rusbult and Farrell (1983) to fit the local context: (a) "If I
the questionnaire survey was completed. The average overall want, it will be easy for me to find another job near my present
turnover rate for the 11 organizations included in this study was employer"; (b) "If I want, it will be easy for me to find a job
7.5% after 3 months and 18% after 10 months. anywhere on Hainan Island"; (c) "If I want, it will be easy
Control variables. We included age, sex, and job tenure as for me to find a job anywhere in mainland China"; (d) "In
biographical control variables in the present study because stud- general, my alternatives are much better compared to my current
ies have shown that these variables are related to turnover (Ar- job"; (e) "All things considered, my alternatives to this job
nold & Feldman, 1982; Marsh & Mannari, 1977; Rhodes, 1983; are excellent"; and ( f ) "My alternatives are much better com-
Werbel & Bedeian, 1989). We also included job satisfaction, pared to my ideal way of occupying my time." The internal
organizational commitment, job rewards, job investments, and consistency coefficient (a) for this scale was .70.
job alternatives as another set of control variables because they
Results
have been identified as two types of antecedents (attitudinal
antecedents that include job satisfaction and organizational com- The first step in the analyses was to examine the psy-
mitment and calculative antecedents that include job rewards, chometric properties of the scales used in the present
926 CHEN, HUI, AND SEGO

study. Second, we created an aggregate variable of OCB homogeneity of variance for the three types of OCB across
because we wanted to examine the effects of overall OCB turnover and no turnover conditions were all insignifi-
on actual turnover. This summed scale was developed by cant, indicating that the variance of these OCBs did not
aggregating the three types of OCB into one single index, differ significantly across turnover versus no turnover
which was accomplished by averaging the scores of altru- conditions.
ism, conscientiousness, and sportsmanship. In addition, Results of the analysis showed that the multivariate test
the three types of OCB were treated discretely in separate of the effects was significant across all indices (i.e., the
analyses to tease out their differential impact. Pillais test, Hotelling's T, Wilks's A, and Roy's test).
Table 1 presents the mean, standard deviation, reliabil- For example, Hotelling's T, a commonly used multivariate
ity coefficient, and correlations of all the variables used statistic, had the value of F(3, 177) = 8.25, p < .001.
in the present study. An examination of this table indicates Separate univariate tests indicated that the levels of all
that the reliability for the scales ranged from a low of three types of OCB differed significantly between the
around .60 to a high, for most scales, above .70. Because turnover and no turnover conditions. Specifically, the level
the results of the turnover data after 3 and 10 months of altruism was higher in the no turnover (M = 3.69)
were almost identical in our analysis, we used the turnover than in the turnover condition (M = 2.94), F(l, 179) =
data after 10 months as the actual turnover index. An 10.93, p < .001; the level of conscientiousness was higher
examination of the zero-order correlations indicated that among those who stayed (M = 3.80) than among those
(a) overall OCB was negatively related to actual turnover who left (M = 3.36), F(l, 179) - 4.84, p < .001; and
(r - — .28, p < .01), (b) turnover intention was positively the level of sportsmanship was higher for the nonquitters
related to turnover ( r = . 1 5 , p < . 0 5 ) , and (c) turnover (M = 3.82) than for the quitters (M = 3.40), F(1, 179)
intention was negatively related to OCB (r = -.17, p = 7.47, p < .01. These results further support our hypoth-
< .05). In sum, zero-order correlations provide general esis that employees who exhibit low levels of OCB are
support for our hypotheses that OCB and turnover inten- more likely to leave an organization than those who ex-
tion are both related to actual turnover. hibit high levels of OCB.
To examine the relationship between overall OCB and Because the 205 subordinate-supervisor dyads were
actual turnover more rigorously, we conducted a logistic from 11 companies, we created 10 dummy variables to
regression analysis, with actual turnover as the dependent retest the effects to control for unmeasured differences
variable, and turnover intention, overall OCB, and all of among organizations that may impact the dependent vari-
the control variables as the independent variables. Results ables. First, a logistic regression analysis was conducted on
of the analysis are reported in Table 2. A closer examina- actual turnover using the 10 dummy variables, all of the
tion of the beta coefficients indicates that the three vari- control variables, and the two variables—turnover intention
ables—job tenure, turnover intention, and OCB—had and OCB—as the independent variables. The results indi-
significant effects on actual turnover. Specifically, em- cate that OCB was a significant predictor of actual turnover
ployees who worked longer in the organizations were (ft = —2.64, p < .001), but turnover intention was no
more likely to leave1 (ft - .36, p < .05), employees who longer a significant predictor (ft = 0.67, ns). A further
had stronger intentions to leave were more likely to quit examination of the results indicates that the lack of signifi-
(ft = 1.07, p < .05), and employees who exhibited low cance of turnover intention on actual turnover was due to
levels of OCB were more likely to leave the organization two dummy variables (two companies).
(/? = -2.36, p < .001). These results provide support Then we conducted a multivariate analysis of covari-
for our hypothesis that turnover intention, as well as OCB, ance using the 10 dummy variables, turnover intention,
are predictors of actual employee turnover. and all of the other control variables as the covariates,
To observe the patterns of the three types of OCB exhib- turnover versus no turnover as the independent variable,
ited by employees who stayed and by those who left, we and the three types of OCB as the dependent variables.
examined supervisors' ratings of their subordinates by The results were identical to what we obtained without
dividing them into two categories: turnover versus no turn-
over. We performed a multivariate analysis of covariance,
1
This result may be explained by the fact that the sample we
with turnover versus no turnover as the independent vari-
surveyed was relatively young (the average age was 28.66, with
able, all of the control variables and turnover intention as
an average job tenure of 1.85 years). To these people, a few
covariates, and the three types of OCB as the dependent
years of working experience may have given them more confi-
variables. This analysis was not to show a causal relation- dence about their own knowledge and skills so that they would
ship between turnover and OCB; rather, it was to reveal like to take the risk of quitting their current job (in the hope of
the patterns of the three types of OCB in the turnover finding a better one) as the Chinese economy makes the transi-
and no turnover conditions. The Bartlett-Box tests for the tion from government control to being market oriented.
 OCB AND TURNOVER 927

the 10 dummy variables: The levels of all three types of

OCB—altruism, conscientiousness, and sportsman-
ship—were significantly lower in the turnover than in the
no turnover situation, F(l, 169) = 5.42, p < .05; F(l,
169) = 15.81,p< .001;andF(l, 169) = 9.67, p < .01,
respectively.
These results again provide support for our hypothesis
that OCB is a valid predictor of employee turnover. More-
over, the results suggest that OCB is an even better pre-
dictor than turnover intention.

Discussion and Conclusions

00 OO OO \D M

The results of the present study provide support for

our hypothesis that OCB has predictive utility regarding
employee turnover. Our results clearly show that employ-
ees who exhibit low levels of OCB are more likely to
leave an organization than those who exhibit high levels
of OCB. The results also show that this relationship held
across the 11 companies investigated, suggesting that
OCB is a valid predictor of employee turnover.
Turnover intention was found to be a predictor of turn-
®_ U-> — O -H 3 00 Q
„ p I-H -H q p 25
over in general, that is, employees who had stronger inten-
tions to leave were more likely to leave than those who
had weaker intentions, which replicated the findings from
most of the other turnover studies (Horn & Hulin, 1981;
Mobley et al., 1978; Mowday et al., 1979). However,
this relationship did not hold for two companies. Further
investigation of the two companies revealed that there
was tension between the management and the workforce
in one company (a joint venture) and in the other com-
pany, some data were collected by the company's repre-
sentative because we could not stay long enough to see
all respondents finish. We suspect that employees in both
SS8S2SSSSSS companies might have been afraid to reveal their true
- - -
intentions either because they chose to give false informa-
)*C'-'^
tion in case this information would be available to the
top management or because they were trying to avoid
possible negative outcomes that would be associated with
people who had such intentions. Another plausible expla-
nation for the inconsistency between turnover intention
and turnover is that the social norm in these two compa-
nies may be contradictory to the "express yourself freely
and honestly" rule assumed by researchers (for a review,
1 see C. F. Yang & Chiu, 1987) and that Chinese behavior

I is more often influenced by situational and relational fac-

tors than internal factors (K. S. Yang, 1981).
The strengths of the present study are threefold. First,
the conceptualization of OCB as a form of employee be-
havioral withdrawal provides a new perspective from
which to study the turnover problem. As discussed earliei;
previous models of turnover have not paid much attention
to the role of behavioral withdrawal, and those few that
928 CHEN, HUI, AND SEGO

Table 2 MacKenzie (1997), Podsakoff and MacKenzie (1997),

Results of the Logistic Regression Analyses and Walz and Niehoff (1996) that demonstrate that OCBs
Actual turnover
are related to organizational effectiveness, especially be-
cause turnover is related to organizational performance
Independent variable SE and effectiveness (cf. Horn & Griffeth, 1995). In addition,
Age -0.04 .05 .00
these findings suggest that future research should examine
Sex -0.08 .47 .00 the effects of OCB on other important organizational out-
Job tenure 0.36* .17 .13 come variables such as improved quality and efficiency,
Job satisfaction 0.42 .32 .00
Org. commitment -0.75
reduced costs, and increased productivity.
.55 .00
Job reward 0.73 .45 .07 Finally, the finding that OCB is a critical determinant
Job investment 0.41 .36 .00 of turnover has practical value for managers in organiza-
Job alternative -0.23 .45 .00
Turnover intention 1.07* .51 .12
tions. On the one hand, managers can observe employees'
OCB -2.36*** .53 -.34 levels of OCB to predict the likelihood of their turnover,
which is easier than knowing their turnover intention; on
Note, n = 190. Org. = organizational; OCB = organizational citizen-
ship behavior.
the other hand, it will be worthwhile for managers to focus
* p < . 0 5 . ***p<.001. on helping to increase subordinates' levels of OCB to get
them involved in organizational activities and subse-
quently to decrease turnover. This practice will be effec-
did pay attention to such withdrawal treated it mainly as tive because according to the cognitive dissonance theory
a reaction to job affect (Hulin, 1991; Rosse & Hulin, (Festinger, 1957), once one is engaged in OCB, one has
1985) or focused on a narrow range of behaviors such as to admit to having positive attitudes toward the organiza-
absenteeism and lateness (Beehr & Gupta, 1978; Herz- tion; and once one has positive attitudes toward the orga-
berg et al., 1957). Our results show that employees' level nization, one will be less willing to leave the organization.
of OCB is a valid predictor of turnover. Second, reports This theory (or reasoning) can find support from the find-
of the three critical components of the study—turnover ings in the behavior-attitude literature ( Aronson & Mills,
intention, OCB, and actual turnover—were all from dif- 1959; Axsom, 1989; Lawler, Kuleck, Rhode, & Sorensen,
ferent sources, thus the results do not suffer from common 1975) and from studies on the relationship between job
method variance problems (Podsakoff & Organ, 1986). altitudes and turnover (Horn & Griffeth, 1991; Mowday,
In particular, turnover intention was based on employee Porter, & Steers, 1982).
self-report data, OCB measures were obtained from em- We recognize, however, that there are alternative expla-
ployees' immediate supervisors, and the actual turnover nations for our findings. One alternative explanation is that
data were collected from each company's records after the the different levels of OCB may be caused by individual
survey. The fact that the variance explained by turnover differences. For example, people who score high on the
intention on OCB and the variance explained by OCB on conscientiousness dimension of the Big Five (see Costa &
actual turnover were not from a common method variance MaCrae, 1985) may present higher levels of the conscien-
gives us more confidence about the reliability and validity tiousness OCB. Indeed, some studies have shown that
of our findings. dispositional variables (e.g., affectivity or optimism) in-
Third, these results have significant implications for fluence levels of OCB (Mount, Barrick, & Strauss, 1994),
future research in organizations. The present study exam- although other studies have not demonstrated such effects
ined die relationship between OCB and turnover at the (Organ & Ryan, 1995). Because we did not measure these
individual level. That is, it examined the relationship be- personality variables in this study, the alternative explana-
tween OCB and turnover by the individuals who exhibit tion cannot be excluded entirely. However, the Big Five
these OCBs. However, another interesting question to ex- measures personality whereas OCB measures behaviors;
amine is the relationship between OCB and turnover at hence, even though they are related, these two concepts
the group or organizational level. For example, we may are rooted in different domains.
expect that groups (or organizations) that have higher Another point worth discussing is the sample used in
levels of OCB will have lower levels of turnover because this study. People may argue that these findings are subject
interactions among employees who exhibit high levels of to cultural factors and thus are not likely to be generalized.
OCB are likely to foster group attractiveness and cohesive- Virtually any study conducted in a specific setting has
ness and subsequently to decrease voluntary turnover. this problem. For example, findings from a sample of
Moreover, the finding that OCB is related to turnover nurses may be difficult to generalize to a sample of steel
appears to complement the recent research findings re- workers, and findings from North American organizations
ported by Karambayya (1989), Podsakoff, Ahearne, and may be difficult to generalize to Southern Chinese organi-
 OCB AND TURNOVER 929

zations. We contend, however; that this should not under- of internal mobility on employee turnover: Multiple field as-
mine the theoretical contribution and practical implication sessments. Journal of Management, 13, 705-711.
of our findings, although, at the same time, we suggest a Farh, J. L., Barley, P. C., & Lin, S. C. (1997). Impetus for ac-
tion: A cultural analysis of justice and organizational citizen-
test of the boundary conditions of our findings because
ship behavior in Chinese society. Administrative Science
we believe that generalizability of findings from one cul-
Quarterly, 42, 421-444.
ture to another is important in application, especially as
Festinger, L. (1957). A theory of cognitive dissonance. Evans-
organizations are becoming less and less constrained by ton, IL: Row, Peterson.
national or geographical boundaries. Fishbein, M., & Ajzen, I. (1974). Attitudes toward objects as
predictors of single and multiple behavioral criteria. Psycho-
References logical Review, 81, 59-74.
Herzberg, P., Mausner, B., Peterson, R., & Capwell, D. (1957).
Arnold, H. J., & Feldman, D. C. (1982). A multivariate analysis Job attitudes: Review of research and opinion. Pittsburgh,
of the determinants of job turnover. Journal of Applied Psy- PA: Psychological Service of Pittsburgh.
chology, 67, 352. Horn, P. W., & Griffeth, R. W. (1991). Structural equations
Aronson, E., & Mills, J. (1959). The effect of severity of initia- modeling test of a turnover theory: Cross-sectional and longi-
tion on liking for a group. Journal of Abnormal and Social tudinal analyses. Journal of Applied Psychology, 76, 350-
Psychology, 12, 16-27. 366.
Axsom, D. (1989). Cognitive dissonance and behavior change Horn, P. W., & Griffeth, R. W. (1995). Employee turnover. Cin-
in psychotherapy. Journal of Experimental Social Psychology, cinnati, OH: International Thomson.
25, 134-252. Hom,P. W.,&Hulin,C. L. (1981). Acompetitive test of predic-
Barrick, M. R., & Mount, M. K. (1991). The Big Five personal- tion of reenlistment by several models. Journal of Applied
ity dimensions and job performance: A meta-analysis. Person- Psychology, 66, 23-39.
nel Psychology, 44, 1-26. Hulin, C. L. (1991). Adaptation, persistence, and commitment
Bateman, T. S., & Organ, D. W. (1983). Job satisfaction and the in organizations. In M. D. Dunnette & L. M. Hough (Eds.),
good soldier: The relationship between affect and employee Handbook of industrial and organizational psychology (Vol.
"citizenship." Academy of Management Journal, 26, 587- 2, pp. 445-505). Palo Alto, CA: Consulting Psychologists
595. Press.
Beehr, T. A., & Gupta, N. (1978). A note on the structure of Hulin, C. L., Roznowski, M., & Hachiya, D. (1985). Alternative
employee withdrawal. Organizational Behavior and Human opportunities and withdrawal decisions: Empirical and theo-
Performance, 21, 73-79. retical discrepancies and an integration. Psychological Bulle-
Benson, P., & Pond, S. (1987). An investigation of the process tin, 97, 233-250.
of employee withdrawal. Journal of Business and Psychol- Karambayya, R. (1989). Contexts for organizational citizenship
ogy, 1, 218-229. behavior: Do high performing and satisfying units have better
Borman, W. C., & Motowidlo, S. J. (1993). Expanding the crite- "citizens"? (%rk University working paper). North %rk,
rion domain to include elements of contextual performance. Ontario, Canada: Author.
In N. Schmitt & W. C. Borman (Eds.), Personnel selection Katz, D., & Kahn, R. L. (1966). The social psychology of orga-
in organizations (pp. 71-98). San Francisco: Jossey-Bass. nizations. New \ork: Wiley.
Brislin, R. W. (1980). Translation and content analysis of oral Katz, D., & Kahn, R. L. (1978). The social psychology of orga-
and written material. In H. C. Triandis & J. W. Berry (Eds.), nizations (2nd ed.). New Tfork: Wiley.
Handbook of cross-cultural psychology, Vol. 2: Methodology Lawler, E. E., Ill, Kuleck, W. J., Fr., Rhode, J. G., & Sorensen,
(pp. 349-444). Boston: Allyn & Bacon. J. E. (1975). Job choice and post-decision dissonance. Orga-
Camman, C., Fichman, M., Jenkins, D., & Klesh, J. (1979). nizational Behavior and Human Performance, 13, 133—145.
The Michigan Organizational Assessment questionnaire. Un- March, J. G., & Simon, H. A. (1958). Organizations. New York:
published manuscript, University of Michigan, Ann Arbor. Wiley.
Carsten, J. M., & Spector, P. E. (1987). Unemployment, job Marsh, R., & Mannari, H. (1977). Organizational commitment
satisfaction and employee turnover: A meta-analytic test of and turnover: A predictive study. Administrative Science
the Muchinsky model. Journal of Applied Psychology, 72, Quarterly, 22, 57-75.
374-381. McEvoy, G.M., & Cascio, W. G. (1987). Do good or poor
Costa, P. T, Jr., & McCrae, R. R. (1985). The NEO Personality performers leave? A meta-analysis of the relationship between
Inventory manual. Odessa, FL: Psychological Assessment performance and turnover. Academy of Management Journal,
Resources. 30, 744-762.
Coverdale, S., & Terborg, J. R. (1980). A re-examination of the Michaels, C. E., & Spector, P. E. (1982). Causes of employee
Mobley, Homer & Hollingsworth model of turnover: A useful turnover: A test of the Mobley, Griffeth, Hand, and Meglino
replication (Tech. Rep. No. 80-4). Arlington, VA: Office of model. Journal of Applied Psychology, 67, 53-59.
Naval Research, Organizational Effectiveness Research Miller, H. E. (1981). Withdrawal behaviors among hospital em-
Program. ployees. Unpublished doctoral dissertation, University of Illi-
Dalton, D. R., & Todor, W. D. (1987). The attenuating effects nois at Urbana-Champaign.
930 CHEN, HUI, AND SEGO

Millet, H. E. (1982, August). Some evidence concerning the research. In G. R. Ferris (Ed.), Research in personnel and
progression of withdrawal hypothesis. Paper presented at the human resources management (Vol. 11, pp. 1-40). Green-
Annual Meeting of the Academy of Management, New 'York. wich, CT JAI Press.
Millet; H. E., Katerberg, R., & Hulin, C. L. (1979). Evaluation Podsakoff, P. M., MacKenzie, S. B., Moorman, R. H., & Fetter,
of the Mobley, Horner, and Hollingsworth model of employee R. (1990). Transformational leader behaviors and their ef-
turnover. Journal of Applied Psychology, 64, 509-517. fects on followers' trust in leader, satisfaction, and organiza-
Mobley, W. H. (1977). Intermediate linkages in the relationship tional citizenship behaviors. Leadership Quarterly, 1, 107-
between job satisfaction and employee turnover. Journal of 142.
Applied Psychology, 67, 237-240. Podsakoff, P. M., & Organ, D. W. (1986). Self-reports in orga-
Mobley, W. H., Griffeth, R. W., Hand, H. H., & Meglino, B. M. nizational research: Problems and prospects. Journal of Man-
(1979). Review and conceptual analysis of the employee turn- agement, 12, 531-544.
over process. Psychological Bulletin, 86, 493-522. Porter, L. W., Steers, R. M., Mowday, R. T, & Boulian, P. V.
Mobley, W. H., Horner, S. O., & Hollingsworth, A. T. (1978). (1974). Organizational commitment, job satisfaction, and
An evaluation of precursors of hospital employee turnover. turnover among psychiatric technicians. Journal of Applied
Journal of Applied Psychology, 63, 408-414. Psychology, 59, 603-609.
Mount, M. K., Barrick, M. R., & Strauss, J. P. (1994). Validity Price, J. L., & Mueller, C. W. (1986). Absenteeism and turnover
of observer ratings of the Big Five personality factors. Journal of hospital employees. Greenwich, CT: JAI Press.
of Applied Psychology, 79, 272-280. Rhodes, S. R. (1983). Age-related differences in work attitudes
Mowday, R. T., Koberg, C. S., & McArthur, A. W. (1984). The and behavior: A review and conceptual analysis. Psychologi-
psychology of the withdrawal process: A cross-validational cal Bulletin, 93, 328-367.
test of Mobley's intermediate linkage model of turnover in Rosse, J. G. (1988). Relations among lateness, absence, and
two samples. Academy of Management Journal, 27, 79-94. turnover: Is there a progression of withdrawal? Human Rela-
Mowday, R., Porter, L., & Steers, R. (1982). Employee-organi- tions, 41, 517-531.
zation linkages: The psychology of commitment, absenteeism, Rosse, J. G., & Hulin, C. L. (1985). Adaptation to work: An
and turnover. New 'fork: Academic Press. analysis of employee health, withdrawal, and change. Organi-
Mowday, R., Steers, R., & Porter, L. (1979). The measurement zational Behavior and Human Decision Processes, 36, 324-
of organizational commitment. Journal of Vocational Behav- 347.
ior, 14, 79-94. Rosse, J. G., & Miller, H. E. (1984). Relationship between ab-
Muchinsky, P. M., & Morrow, P. C. (1980). A multi-disciplin- senteeism and other employee behaviors. In P. S. Goodman &
ary model of voluntary turnover. Journal of Vocational Be- R. S. Atkin (Eds.), Absenteeism: new approaches to under-
havior, 17, 263-290. standing, measuring, and managing employee absence. San
Organ, D. W. (1988). Organizational citizenship behavior: The Francisco: Jossey-Bass.
good soldier syndrome. Lexington, MA: Lexington Books. Rusbult, C. E., & Farrell, D. (1983). A longitudinal test of
Organ, D. W., & Ryan, K. (1995). A meta-analytic review of the investment model: The impact on job satisfaction, job
attitudinal and dispositional predictors of organizational citi- commitment, and turnover of variations in rewards, costs,
zenship behavior. Personnel Psychology, 48, 775-802. alternatives, and investments. Journal of Applied Psychology,
Park, O.-S., & Sims, H. P., Jr. (1989). Beyond cognition in 68, 429-438.
leadership: Prosocial behavior and affect in managerial Salancik, G. R. (1983). Commitment and the control of organi-
judgment. Unpublished manuscript. zational behavior and belief. In B. M. Stow (Ed.), Psycholog-
Podsakoff, P.M., Ahearne, M., & MacKenzie, S. B. (1997). ical foundations of organizational behavior (pp. 202-206).
Organizational citizenship behavior and the quantity and qual- Glenview, IL: Scott, Foresman.
ity of work group performance. Journal of Applied Psychol- Scarpello, V, & Campbell, J. P. (1983). Job satisfaction: Are
ogy, 82, 262-270. all the parts there? Personnel Psychology, 36, 577-600.
Podsakoff, P.M. & MacKenzie, S. B. (1994). Organizational Smith, C. A., Organ, D. W., & Near, J. P. (1983). Organiza-
citizenship behavior and sales unit effectiveness. Journal of tional citizenship behavior: Its nature and antecedents. Jour-
Marketing Research, 31, 351-363. nal of Applied Psychology, 68, 653-663.
Podsakoff, P. M., & MacKenzie, S. B. (1997). The impact of Smith, P. C., Kendall, L. M., & Hulin, C. L. (1969). The mea-
organizational citizenship behavior on organizational perfor- surement of satisfaction in work and retirement: A strategy
mance: A review and suggestions for future research. Human for the study of attitudes. Chicago: Rand McNally.
Performance, 10, 133-151. Steers, R. M., & Mowday, R. T. (1981). Employee turnover and
Podsakoff, P. M., MacKenzie, S. B., & Bommer, W. H. (1996). post-decision accommodation processes. Research in organi-
Transformational leader behaviors and substitutes for leader- zational behavior (Vol. 3, pp. 235-281). Greenwich, CT
ship as determinants of employee satisfaction, commitment, JAI Press.
trust, and organizational citizenship behaviors. Journal of Thibaut, J. W., & Kelley, H. H. (1959). The social psychology
Management, 22, 259-298. of groups. New \brk: Wiley.
Podsakoff, P. M., MacKenzie, S. B., & Hui, C. (1993). Organi- Walz, S. M., & Niehoff, B. P. (1996). Organizational citizen-
zational citizenship behaviors and managerial evaluations of ship behaviors and their effect on organizational effectiveness
employee performance: A review and suggestions for future in limited-menu restaurants. In J. B. Keys & L. N. Dosier
 OCB AND TURNOVER 931

(Eds.), Academy of Management best papers proceedings nese subjects: A review of the effects of Western scales. Jour-
(pp. 307-311). nal of Chinese Psychology, 29. 113-132.
Wells, D. L., & Muchinsky, P. M. (1985). Performance anteced- Yang, K. S. (1981). The personality and behavior of a Chinese:
ents of voluntary and involuntary managerial turnover. Jour- Formation and change. Journal of Chinese Psychology, 23,
nal of Applied Psychology, 70, 329-336. 39-55.
Werbel, J. D., & Bedeian, A. G. (1989). Intended turnover as
a function of age and job performance. Journal of Organiza- Received March 27, 1997
tional Behavior, JO, 275-281. Revision received May 7, 1998
Yang, C. P., & Chiu, C. Y. (1987). The paradox faced by Chi- Accepted May 11, 1998 •

i | AMERICAN PSYCHOLOGICAL ASSOCIATION

F
SUBSCRIPTION CLAIMS INFORMATION Today'some:
We provide this form to assist members, institutions, and nonmember individuals with any subscription problems. With the
appropriate information we can begin a resolution . If you use the services of an agent, please do NOT duplicate claims through
them and directly to us. PLEASE PRINT CLEARLY AND IN INK IF POSSIBLE.

PRINT FULL NAME OR KEY NAME OF INSTITUTION MEMBERORajSTCMES.NUMBER(MAYBEFOUNDON ANYPASTLSSUELABEL)

ADDRESS DATE YOUR ORDER WAS MAILED (OR PHONED)

PREPAID CHkCK CHARGE

CHECK/CARD CLEARED DATE:
CITY STATE/COUNTRY ZIP
(If potable, Kndacopy, front and back, of your cancelled check to betpui In our research
ofvomdatm.)
YOUR NAME AND PHONE NUMBER ISSUES: MISSING DAMAGED

TITLE VOLUME OR YEAR NUMBER OR MONTH

Thank you. Once a claim is received and resolved, delivery of replacement issues routinely lakes 4-6 weeks.

DATE RECEIVED: DATE OF ACTION!

ACTION TAKEN! INV. NO. ft DATE:
STAFF NAME! LABEL NO. ft DATEs

Send this farm to APA Subscription Claims, 750 First Street, NE, Washington, DC 20002-4242

PLEASE DO NOT REMOVE. A PHOTOCOPY MAY BE USED.