Sie sind auf Seite 1von 30

APPLIED PSYCHOLOGY: AN INTERNATIONAL REVIEW, 1990.

39 (1) 47-76

Job Performance and Turnover:


A Review and Meta-analysis
Peter Bycio
Xavier University, Cincinnati, U.S. A.

R c k D. Hackett
McMaster University, Canada

Kenneth M. Alvares
Nichols Institute, U.S.A .

Les ramifications theoriques et appliquees du couple performance au travail


et turn-over du personnel furent identifikes et une meta-analyse (cf. Hunter,
Schmidt, et Jackson, 1982) reahsee 21 partir des rksultats d e 45 etudes.
Principalement, les partants tendaient ?J rtaliser des performances moyen-
nes, mais la correlation etait specialement faible et variable quand le depart
volontaire ktait invoque. La correlation en m comgke entre la notation de
Iencadrement et le dipart volontaire etait d e -0.17 (variance = 0.033),
tandis que Iequivalent en cas de depart involontaire etait d e -0.52 (variance
= 0.012). Des resultats semblables etaient obtenus avec des indices d e
performance qualitatifs. Erreur dkchantillonnage, variations dans la fidelitk
d e la performance e t ajustement des diffkrences dans le taux de turn-over
recommand6 par Kemery, Dunlap, et Griffeth (1988) rarement releverent
pour plus de 50% d e la variance parmi 1es rksultats.

The theoretical and applied ramifications of the job performance-employee


turnover relationship were identified, and a meta-analysis (cf. Hunter,
Schmidt, & Jackson, 1982) of data from 45 studies was performed. Overall,
leaven tended to be below average performers, but the relationship was
especially weak and variable when voluntary turnover was involved. T h e
average corrected correlation involving supervisory ratings and voluntary
turnover was -0.17 (variance = 0.033) whereas the equivalent involuntary
turnover finding was -0.52 (variance = 0.012). Similar findings were
obtained with non-rating performance indices. Sampling error, variations in
performance reliability, and the adjustment for differences in the rate of

Requests for reprints should be sent to Peter Bycio, Management Department, Xavier
University, 3800 Victory Parkway, Cincinnati, Ohio 45207. U.S.A. We gratefully acknowl-
edge the help of June Hahn and John Hunrer on various aspects of this manuscript.
~~~

01990 International Association of Applied Psychology


48 BYCIO. HACKElT, ALVARES

turnover recommended by Kemery. Dunlap. and Griffeth (1988) rarely


accounted for more than 50% of the variance among t h e findings.

INTRODUCTION
Despite more than 50 years of employee turnover research, Mobley (1982,
p. 134) found that t h e literature exhibits a dearth of conceptual or
empirical treatments of performance as either an antecedent or conse-
quence in the turnover process. Since then, considerable empirical and
theoretical work on the topic has appeared. In this paper, a review and
meta-analysis (cf. Hunter, Schmidt, & Jackson, 1982) of the performance-
turnover literature is presented.

THEORETICAL RE LEVANCE
Job performance has been incorporated into some recent models of the
turnover process (see, for example, Farris, 1971; Jackofsky, 1984; Steers &
Mowday, 1982). Expectations regarding the direction of the relationship
differ depending on whether the focus is on voluntary or involuntary
turnover. Steers and Mowday (1982) deal only with voluntary turnover
which, they argue, results from the complex interaction of job perform-
ance, job attitudes, organisational experiences, and job expectations.
According to their model, the poor performer tries to rationalise his or her
sub-par performance by harbouring negative attitudes towards the job
(this is a lousy job anyway). Employees perceptions of pay and prom-
otional opportunities are also thought to become gradually unfavourable,
to the point where the supervisor takes punitive action. Sanctions, in turn,
are hypothesised to exacerbate the negative feelings of the poor performer
and sometimes prompt voluntary turnover. Steers and Mowday (1982)
further argued (as did Farris, 1971), that organisations find ways to keep
their above average performers, largely by meeting their heightened
expectations regarding organisational rewards. Thus, both Steers and
Mowday (1982). and Farris (1971). expect a negntive relationship between
performance and voluntary turnover. Poor performers will tend to leave,
and high performers will tend to stay.
Jackofsky (1984) also proposed a model of turnover which included job
performance as a variable. The model was intended to explain intrnorgani-
sational movement (i.e. job changes resulting from lateral transfers and
promotions), along with the more typically studied inrerorganisational
movement including both voluntary and involuntary turnover. In contrast
to the predictions of others (cf. Farris, 1971; Steers & Mowday, 1982),
Jackofsky (1984) hypothesised that performance and voluntary turnover are
positivefy related, i.e. that high performers tend voluntarily to leave
JOB PERFORMANCE AND TURNOVER 49

their organisations. This prediction was based largely on the belief that
above average performance buys enhanced alternative job opportunities,
and that high performers receive a greater number of unsolicited job offers
than their lower performing counterparts. Above average personnel were
viewed as being more likely to leave voluntarily simply to take advantage
of the attractive job alternatives that below average performers lack
(Jackofsky, 1984).
Though theoretical differences exist concerning the expected direction
of performance and voluntary turnover, the predictions involving perform-
ance and involuntary turnover have been uniform. Involuntary turnover,
not employee initiated, is viewed by most investigators as negatively
related to job performance (Jackofsky, 1984; Stumpf & Dawley, 1981;
Wanous, Stumpf, & Bedrosian, 1979). Poor performers are thought to be
dismissed from their jobs, whereas moderate to high performers are
allowed, or possibly encouraged, to remain.

APPLlE D RELEVANCE
The direction and magnitude of the job performance-turnover rela-
tionship is important from an applied perspective because t h e conse-
quences of turnover for an organisation are largely dependent on who stays
and who goes (Mobley, 1982). For example, if companies tend to lose their
best employees, the costs associated with a poorer performing replacement
must be absorbed along with those of additional recruiting, selection, and
training. On the other hand, if the majority of voluntary leavers are poor
performers, and easily replaceable, it may be that the negative effects of
turnover have been greatly exaggerated. A high turnover rate might even
be regarded as functional (Dalton, Krackhardt, & Porter, 1981; Hollen-
beck & Williams, 1986).

THE JOB PERFORMANCE-TURNOVER


LITERATURE
In two early reviews of the performance-turnover literature, Price (1977),
and Martin, Price, and Mueller (1981), tentatively concluded that high
performers were most likely to leave their jobs. This conclusion was based
on a small number of studies (cf. Price, 1977) in which performance had
often been poorly measured (cf. Martin et al., 1981). More recently
Jackofsky (1984), citing 16 studies, concluded that the performance-
turnover relationship was inconsistent. To identify conceptual explana-
tions for the inconsistencies, Jackofsky (1984) devoted most of her paper to
developing a model of turnover that incorporated job performance as a
variable.
50 BYCIO, HACKElT, ALVARES

The latest review of the performance-turnover literature by McEvoy


and Cascio (1987) went beyond the previous ones in several import-
ant ways. First, they used meta-analysis as proposed by Hunter et al.
(1982) to: (1) examine the size of the performance-turnover relation-
ship in addition to its direction, and (2) evaluate the contribution that
statistical artefacts (sampling error and unreliability in job performance)
had on the variability of the findings. Their review of 18 articles yielded an
average corrected correlation between performance ratings and voluntary
turnover of -0.31, whereas the equivalent finding for involuntary turnover
was -0.51. Thus, below average performers tended to leave regardless of
the type of turnover involved, but the relationship was much stronger for
involuntary cases. McEvoy and Cascio (1987) also found that sampling
error accounted for only a small percentage (11'/0 overall) of the variance
in the performance ratings-turnover correlations. They concluded that
unemployment rates, the amount of time turnover was measured for, and
differences in the relative proportions of stayers and leavers involved in t h e
samples were responsible for the remaining variability among the studies.
This review expands on the previous literature in three ways. First, a
more comprehensive and updated database was used. We cover 37 studies
that were not included in previous reviews. Our larger database should
lessen second-order sampling error (cf. Schmidt et al., 1985) and increase
the level of confidence in the accuracy of the meta-analytic results (cf.
Sackett, Harris, & O r r , 1986). Second, the correction procedure recom-
mended by Kemery, Dunlap, and Griffeth (1988) was used to adjust the
correlations to a common turnover base rate of 50%. This adjustment was
necessary because correlations involving different proportions of stayers
and leavers are not directly comparable (cf. Kemery et al., 1988). Finally,
we reviewed findings involving non-rating measures of performance (such
as counts of actual work performed) in addition to those involving super-
visory ratings. This made it possible to determine if t h e ratings-turnover
relationships identified by McEvoy and Cascio (1987) extend to non-rating
indices. Different relationships might reasonably be expected because
ratings are often poorly correlated with other performance indices (cf.
Cascio & Valenzi. 1978; Seashore, Indik, & Georgopoulos, 1960) and are
subject to various rater errors (cf. Bernardin & Beatty. 1984).

METHOD
Literature Covered in this Review
Dara Gathering. Our interest was confined to studies of interorganisa-
tional turnover where the individual employee was the unit of analysis.
Thus, job changes resulting from transfers, promotions, and the like were
JOB PERFORMANCE AND TURNOVER 51

excluded as were studies using departments or organisations as the unit of


analysis. This differs somewhat from McEvoy and Cascio (1987) in that
they reviewed both intra and interorganisational findings. We chose not to
because the mechanisms underlying intraorganisational movement (see,
for example, Forbes, 1987, on promotions) are likely very different from
those associated with leaving an organisation entirely.
Several search methods were used to attempt retrieval of all relevant
studies done prior to 1989. As a first step, all volumes of the following
journals were searched: the Journal of Applied Psychology, Organizational
Behavior and Human Performance (now Organizational Behavior and
Human Decision Processes), Personnel Psychology, the Academy of Man-
agement Journal, Administrative Science Quarterly, Human Relations, and
the Journal of Vocational Behavior. These periodicals were hand searched
(as opposed to computer searched) because they were the ones most likely
to contain empirical data.
Many studies were located which contained the variables of interest, but
not the required correlation(s). Sometimes the correlation(s) could be
derived from other descriptive statistics in the paper itself (cf. Ferguson,
1976 pp. 415-418). If the correlation(s) could not be derived, a letter
requesting the data was directed to the author whenever there was reason
to believe that the correlation might have been calculated as part of the
original analysis. Letters asking for unpublished data were also sent to
those who frequently contributed to the turnover literature. Finally, a
computer search was conducted using the ABIIINFORM, Psychological
Abstracts, and the Dissertation Abstracts International databases.

Data Sorting. The correlations were grouped according to whether the


investigator claimed to be measuring voluntary or involuntary turnover.
Because some investigators did not separate voluntary from involuntary
turnovers, a total turnover grouping was also created. When there was
doubt as to which form of turnover was involved, the finding was placed in
the total turnover grouping. Using Cascio (1987, p. 52) as a guide, the
findings were also sorted to reflect the type of performance measure
involved: (1) supervisory ratings or rankings; (2) self-ratings; (3) output
measures, that is. actual counts of work performed; (4) quality measures,
such as number of errors made or the number of complaints registered; and
( 5 ) promotability indices, such as the number of promotions or pay
increases given.

Srraregies for Handling Multiple Findings. Meta-analytic formulas


require the assumption that the correlations in each analysis are statisti-
cally independent (Hunter et al., 1982). Because failure to meet this
assumption can result in an under-correction for sampling error, correla-
52 BYCIO, HACKETT, ALVARES

tions from the same sample which involved only slightly different perform-
ance measures were averaged to form a single value. This strategy was used
(for example) when investigators presented multiple performance-
turnover correlations based on highly correlated performance rating
dimensions.

An a Iysi s

Meta-analytic Corrections. Using the artefact distributions method


developed by Hunter et al. (1982, pp. 73-83) all the analyses involved
correcting the correlations for: (1) sampling error; (2) differences in the
reliability of job performance (except when non-rating promotional indices
were used); and (3) a common rate (50%) of turnover (cf. Kemery et al.,
1988).
When supervisory ratingsjrankings were involved, the correction for
unreliability in measurement was based on the mean (0.82) and variance
(0.01) of the square root of 10 inter-rater reliability estimates collected
during the literature search. The equivalent correction involving self-rated
performance was based on a mean (0.92) and variance (0.00) of the square
root of three estimates. Similarly, the mean (0.87) and variance (0.01) of
the square root of six estimates was used in analyses of non-rating output.
Finally, the adjustment for measurement unreliability in non-rating quality
measures was based on the square root of three estimates (mean = 0.86;
variance = 0.00).
No adjustment for unreliability was made when non-rating promotional
indices were involved because traditional notions of reliability do not
readily apply to such low-frequency events. Nor was a correction for
unreliability made to the turnover measure because the determination of
whether an employee has stayed or has left was viewed as virtually error
free (except for occasional clerical mistakes). Of course measurement
error probably does occur when inferences are made as to whether the
turnover was voluntary or involuntary (cf. Jackofsky, 1983, among others)
but there is no obvious way to correct for this.
The adjustment for differences in the proportion of stayers and leavers
required knowledge of the number of stayers and leavers the correlations
were based on. This was explicitly reported (or at least estimable) in 40 of
the 45 articles and was the basis for the correction to both the mean and
variance of each distribution (cf. Hunter & Schmidt. Note 1). The value of
the conversion factor depended on the particular set of correlations
involved (cf. Kemery et al., 1988).

T h e Situational Speclficify Hvpothesis. The sample-size weighted mean


(uncorrected) of each distribution of performance-turnover correlations
JOB PERFORMANCE AND TURNOVER 53

was calculated. The corrections described earlier were then applied to


obtain the corrected sample-size weighted mean, which is the best estimate
of the relationship between the two underlying constructs (cf. Hunter et
al., 1982).
The sample-size weighted variance (uncorrected) of each distribution
was also calculated, as was the variance due to statistical artefacts (such as
sampling error). The observed (uncorrected) variance was then compared
to the variance accounted for by artefacts. When the variance accounted
for by artefacts was 75% or more of the observed variance, the situational
specificity hypothesis was rejected, implying that the relationship under
study was relatively uniform across settings. When artefacts failed to
account for at least 75% of the observed variance, the situational specificity
hypothesis was not rejected, implying that the size of the relationship does
vary across settings due to as yet unspecified situational and/or rnethod-
ological factors (cf. Hunter et al., 1982).

RESULTS AND DISCUSSION

Results of the Literature Search


Twenty-three studies mentioned turnover and performance as variables,
but did not report the correlation(s) between them. Through letter writing,
authors of seven studies were able to provide the requested correlation(s).
Thus, some (or all) of the data from Beyer and Trice (1984), Farr,
O'Leary, and Bartlett (1971), Kanfer, Crosby, and Brandt (1988), Para-
suraman and Alutto (1984), Parsons, Herold, and Leatherwood (1985),
Turnage and Muchinsky (1984), and Wanous et al. (1979) were obtained
directly from the authors. and have not been published previously. Ques-
tions concerning published material sometimes resulted in obtaining other
unpublished work (see Seybolt & Ross, Note 2; Summers & Hendrix, Note
3). Clarifications concerning published data could also sometimes be
obtained. Four of ten authors were helpful in this regard. General
enquiries to well-known investigators for unpublished data were not very
useful, however. Eight such enquiries yielded only one see Bluedorn &
Abelson, Note 4) study.
The results of the performance-turnover literature search are shown in
Tables A1 to A3 (see Appendix A ) . The findings have been grouped with
regard to the particular measures involved, and are listed in chronological
order. Unless otherwise indicated. all correlations are point biserial. The
signs have been changed in some cases so that a negative correlation always
implies that the stayers in the sample performed better than the leavers.
References to the 45 reviewed studies are shown in Appendix B.
Inspection of Appendix A reveals: (1) a broad range of occupations is
54 BYCIO, HACKE'TT, ALVARES

represented; (2) non-rating measures are relatively infrequent, so the


results of those analyses should be viewed cautiously because the 75% rule
lacks power when the number of studies is small (cf. Sackett et al., 1986;
Spector & Levine, 1987); and (3) 91% (83 of 91) of the correlations are
negative.

Meta-analytic Findings for Job Performance and Turnover. Meta-


analytic results pertaining to 15 sets of performance-turnover correlations
are presented in Table 1. Findings, involving each type of performance
measure and each form of turnover, are presented provided more than one
study representing the measurement combination was found.
Table 1 displays the number of correlations (No. of rpb's) and the total
sample size each analysis was based on. These are followed by the
sample-size weighted means ( i p b )which,, because of the sample-size
weighting involved, have effectively been corrected for sampling error.
Next, sample-size weighted means ( ~ 5 reflective
) ~ ~ of the adjustment to a
50-50 split between stayers and leavers are shown. The ppb's have also
been corrected for unreliability in job performance, except in cases where
non-rating promotional indices were involved.
Table 1 also contains the observed variance charactensing each distribu-
tion of correlations along with the amount and percentage of that variance
attributable to statistical artefacts [artefactual variance (percentage
accounted for)]. Next, the variance of the corrected correlations is shown.
These values are usually quite high because reliance on statistical formulas
to correct for artefacts (differencc ;in turnover base rate and measurement
unreliability) increases sampling error tremendously (Hunter & Schmidt,
Note 1). The last column shows 95% credibility intervals around the
corrected mean. These are indicative of the extent to which the size of the
relationship can be expected to vary across situations (cf. Hunter et al.,
1982).

Vohnrary Turnover. The mean correlations between performance and


voluntary turnover are all negative. However, the size of the relationship is
modest at best and highly variable. For example, ppb for supervisory
ratings (-0.17) converts to a standardised mean difference ( d ) in perform-
ance between stayers and voluntary leavers of approximately 0.33 (cf.
Hunter et al., 1982; McEvoy & Cascio, 1985). This implies that the average
stayer exceeds the rated performance level of only 63% of those who leave
(cf. Glass, McGaw, & Smith, 1981. pp. 102-103). Alternatively, if the
strongest performance-voluntary turnover relationship (non-rating qual-
ity, p p h = -0.31) is used to estimate t h e standardised mean performance
difference ( d = 0.62) then the average stayer exceeds the performance
level of approximately 73% of those who leave.
TABLE 1
Meta-analytic Results for Job P e r f o r m a n c e a n d Turnover

Arrefacruul
NO. Observed Variance (% Corrected 95% Credibiliry
i'erformarice of 'pb's n jpb bpb Variance Accounied for) Vuriurice lnrerval
-______
Across all Forms of Turnover
All types" 91 27.312 -0.19 0.037 0.003 (H) 0.034
Voluntary Turnover
Supervisory ratings 24 10.260 -0.12 -0.17 0.020 0.003 (13) 0.033
Supervisory ratingsb 23 6274 -0.18 -0.26 0.022 0.004 (19) 0.034
Self-ratings 3 274 -0.04 -0.05 0.013 0.011 (84) 0.003
Non-rating output 3 420 -0.18 -0.22 0.011 0.008 (71) 0.005
Non-rating quality 7 1444 -0.23 -0.31 0.015 0.006 (44) 0.016
Non-rating promotion 7 2185 -0.17 -0.18 0.063 0.003 (5) 0,067
Involuntary Turnover
Supervisory ratings 10 2744 -0.38 -0.52 0.017 0.010 (62) 0.012 -0.74 < ppb < -0.31
Non-rating quality 5 105 -0.41 -0.61 0.028 0.014 (51) 0.031 -0.96 < ppb < -0.27
Non-rating promotion 3 583 -0.57 -0.59 0.033 0.002 (7) 0.033 -0.94 < ppb < -0.23
Total Turnover
Supervisory ratings 17 4888 -0.18 -0.25 0.011 0.005 (42) 0.012 -0.47 < pPb < -0.03
Self-ratings 3 1320 -0.12 -0.15 0.002 0.002 (100)'
Non-rating output 3 770 -0.51 -0.61 0.045 0.006 (12) 0.056
Non-rating quality 2 31 1 -0.01 -0.01 0.003 0.006 (loo)'
Non-rating promotion 3 98 1 0.02 0.03 0.014 0.003 (23) 0.020

Nore. Only those performance-turnover combinations where more than one study was found are shown.
'This analysis across all of the performance-turnover correlations involves a correction for sampling error only. When this distribution IS corrected
for differences in the turnover base rate. the corrected mean = -0.22. and the variance accounted for by the artefacts is 0.004 (ll"/o).
?his analysis excludes the Ofsanko (1979) finding.
cn
'As artefacts accounted for all the variance among the correlations in this analysis, a credibility interval was not called for.
56 BYCIO, H A C K E T , ALVARES

It is worth noting that our corrected mean involving supervisory ratings


and voluntary turnover (-0.17) is considerably lower than the equivalent
McEvoy and Cascio (1987) finding (-0.31). Moreover, our estimate of the
corrected variance associated with the relationship is much higher (0.033
versus 0.008). These differences are traceable to several judgement calls in
meta-analysis (cf. Wanous, Sullivan, & Malinak, 1989). First, our estimate
was based on 24 studies whereas theirs was based on 6. Our review
contains one study in particular (cf. Ofsanko, 1979) with a small restricted
performance-voluntary turnover correlation (rph, = -0.02) and a huge
sample size ( n = 3986) which helps explain much of the difference in the
Pphs. With Ofsanko (1979) removed, our mean estimate of the supervisory
ratings-voluntary turnover relationship (-0.26, see Table 1) is more
similar to that of McEvoy and Cascio (1987). Even without Ofsanko
(1979), though, our estimate of the corrected variance remains very high
(0.034). Given this large corrected variance and the resulting wide credi-
bility interval, it appears that McEvoy and Cascio (1987) were correct in
not ruling out the possibility of a positive (but weak) relationship between
supenisory ratings and voluntary turnover.
Despite the variance among the voluntary turnover correlations (even
after adjustments for statistical artefacts), the preponderance of evidence
supports Steers and Mowday (1982) and Farris (1971), that voluntary
leavers will tend to be below average performers in most cases. Still, we
need to know more about the circumstances under which strong re-
lationships involving performance and voluntary turnover will emerge.
Steers and Mowday (1982) implied that the passage of time might be
important in this regard. They proposed that below average performers
decide to leave only after becoming entangled in a gradual degenerative
self-reinforcing cycle (p. 127) of poor attitudes and negative organisa-
tional experiences. Consistent with (though not proof of) the value of this
time-sensitive perspective, we found that investigators who measured
voluntary turnover the longest (in months) tended to report larger correla-
tions involving supervisory ratings and voluntary turnover
( r ( n = 23) = -0.33; P < 0.04). Even though this finding failed to attain
statistical significance ( r ( n = 22) = -0.17; P < 0.44) with an outlier
removed (Dreher, 1982, measured turnover for 15 years), closer attention
to the time element as reflected by recent interest in survival analysis (cf.
Morita. Lee, & Mowday, 1989; Peters & Sheridan, 1988) might increase
our understanding of the performance-turnover relationship. Part of the
ongoing progressively more negative experience for the below average
performer might be (for example) the receipt of less support and feedback
from the supervisor (cf. Farris & Lim, 1969; Parsons et al.. 1985; Wells &
Muchinsky, 1985). Poor performers might also receive less than positive
JOB PERFORMANCE AND TURNOVER 57

feedback resulting from self-comparisons against average and above aver-


age co-workers (Kanfer et al., 1988). Thus, performance could play a role
in the voluntary turnover process without the employee ever experiencing
a formal appraisal meeting.
Because the possibility of a weak positive relationship between volun-
tary turnover and performance cannot be ruled out, the mechanisms
thought to underlie it (cf. Jackofsky, 1984) demand some consideration.
Basically the argument for a positive relationship rests on the hypothesis
that superior job performance yields increased attractive job alternatives
relative to lower performing colleagues. Importantly, however, Dreher
(1982) pointed out that high performers cannot always translate their
outstanding achievements into enhanced job opportunities because they
lack the concrete evidence needed to document their past record convin-
cingly in the job market. He argued that above average performers may
not enjoy increased mobility because prospective employers have difficulty
equating performance levels in one organisation with those in another.
Following this reasoning, perhaps the positive relationship Jackofsky
(1984) argued for would receive greater support in studies of intraorganisa-
tional movement where (presumably) information about employees past
performance would be more readily available and interpretable to prospec-
tive bosses. When interorganisational movement is involved, though, there
is evidence suggesting that attractive job options are available to both low
and high performers. For example, in the Stumpf and Dawley (1981) study
most of the bank tellers who voluntarily quit were poorly regarded by their
employer, yet the reason most gave for leaving was to accept another
position. Similarly, in a study involving accountants, Knod (1974) found
even the majority of involuntary leavers got new jobs at an improved rate
of pay. Thus, positive relationships between performance and voluntary
turnover may be uncommon partly because superior past performance
does not guarantee especially attractive job options.

Involuntary Turnover. Regardless of the performance measure


involved, the mean involuntary turnover relationships were negative and
statistically significant. Using supervisory ratings as an example,
GPb= -0.52) the equivalent d-statistic (approximately 1.04) implies that
the average stayer performs better than 85% of the involuntary leavers.
These findings are consistent with the expectations of most investigators
(cf. Jackofsky. 1984; Stumpf & Dawley, 1981; Wanous et al., 1979).
As with voluntary turnover, statistical artefacts failed to explain all the
variance among the involuntary turnover findings. Only in one of the
distributions (supervisory ratings) did differences in the proportions of
stayers and leavers account for a large proportion (about 50%) of the
58 BYCIO, HACKETT, ALVARES

observed variance. This was atypical because this artefact normally


explained only 3-7% of the variance in a given set of correlations. Thus,
even though the uncorrected (observed) variance among the 10 reviewed
correlations (0.017) was much larger than the equivalent McEvoy and
Cascio (1987) finding (0.O008, based on 2 correlations) o u r corrected
means and variances are similar to theirs, due to the large impact the
artefact adjustment had in this particular case.
Some of the variance among the involuntary turnover correlations may
reflect the inclusion of varying numbers of cases whose reason for separa-
tion (i.e. death. lay-off, mandatory retirement) had little to do with their
performance per se. There may also be individual differences among
supervisors in their disposition towards firing people even given sufficient
cause (O'Reilly & Weitz, 1980). In fact Beyer and Trice (1984) found that
most supervisors resort to other disciplinary strategies such as informal and
written warnings and temporary suspeiisions prior to using the dismissal
option. Thus, the strength of the performance-involuntary turnover rela-
tionship might depend on the supervisor and on the degree to which formal
disciplinary procedures short of dismissal exist. Union versus non-union
environments might also affect the relationship (Dalton. Todor, & Krack-
hardt, 1982).

Total Turnover. Except for the findings involving non-rating pro-


motional indices, the means involving total turnover are negative and
statistically significant. Our result for supervisory ratings based on 17
studies Ujpb= -0.25; corrected variance = 0.012) is much smaller than
the equivalent hlcEvoy and Cascio (1987) finding (ppb = -0.53) based on
only one study. Statistical artefacts (especially sampling error) accounted
for all the variance among the self-ratings and non-rating quality results,
though both distributions had only 2 or 3 correlations to begin with.
It is noteworthy that the mean ratings-turnover relationship
(jpb = -0.25) lies between the comparable voluntary ( p = -0.17) and
?b
involuntary (ppb= -0.52) findings. This raises the possibility that some of
the larger performance-voluntary correlations might actually reflect a high
proportion of mutual agreement pacts wherein employees quit to avoid
being fired (cf. Jackofsky, 1984). Misclassification of these cases would
artificially inflate the resulting so-called performance-"vo1untary"turnover
correlations and therefore might partly account for some variance among
them.
JOB PERFORMANCE AND TURNOVER 59

CONCLUSIONS

Implications for Organisations


Organisations must weigh the costs of losing even their below average
personnel against those of obtaining and training replacements (cf. Boud-
reau & Berger, 1985). Despite the slight tendency for voluntary leavers to
be below average performers, a high turnover rate should not be ignored
because: ( I ) even sub-par performance in absolute ferms may be more than
sufficient for organisational goal attainment (Wells & Muchinsky, 1985);
and (2) many above average personnel are lost also. Moreover,
replaceability must also be considered-a below average employee may be
desirable to a vacancy (Dalton et al., 1982).
These qualifications aside, the potential practical importance of per-
formance and turnover was nicely illustrated by HoIlenbeck and WiIliams
(1986), using department store retail sales staff. They estimated that the
store could expect an increase in sales of $112,000 per year for each
occasion an average employee replaced a below average leaver. For this
store at least the implication was clear: find ways to retain above average
performers and encourage the departure of below average personnel. This
requires reliable and valid performance appraisal, a human resource
information system that keeps track of who leaves, and strategies speci-
fically geared to the reduction of avoidable turnover among superior
personnel. With the possible exception of attitudes towards ones immedi-
ate supervisor (Miller, 1988), few consistent differences in the predictors of
turnover have emerged between low and high performing groups (cf.
Farris, 1971; Knod, 1975; Ofsanko, 1979; Seybolt & Ross, 1985). Logic,
though, suggests that the retention of above average employees rests with
performance-contingent rewards (cf. Dreher, 1982; Johns, 1989) facilitated
by goal-oriented behaviourally based performance feedback (cf. Latham &
Wexley, 1981).

Implications for Research


Very few of the 45 reviewed studies contained the same independent
variables or offered information on the host of theoretically important
issues that might be expected to moderate performance and turnover.
Progress in accounting for the variability among the findings will be made
only when the research becomes more focused. T o this end, future
investigators should examine the following variables for their possible
impact: (1) the specific reasons for departure, i.e. the degree of avoidabil-
ity from the perspective of both the employee and the organisation (cf.
Abelson, 1987; Dalton et al., 1981); (2) the nature of the reward system
and the ease with which past performance can be communicated to (and
60 BYCIO, HACKETT, ALVARES

verified by) prospective employers (cf. Dreher, 1982); (3) job tenure (cf.
Kanfer et al., 1988); and (4) local, state, and national unemployment rates
during the study (cf. McEvoy & Cascio, 1987).
We, like McEvoy and Cascio (1987) also advocate testing hypotheses
about performance and turnover in the context of multivariate models to
clarify (for example) the role of job satisfaction in performance and
turnover (cf. Lance, 1988; Rosse, Note 5). Also, in particular, the Steers
and Mowday (1982) formulation deserves more testing because important
links in their model involving performance, job satisfaction, and turnover
have been supported at the univariate level by this and other recent
meta-analytic work (see Iaffaldano & Muchinsky, 1985; Steel & Ovalle,
1984). Although an initial test revealed that only a small percentage of
variance in turnover was explained by the model (cf. Lee & Mowday,
1987), an idiographic-longitudinal approach similar to that used in the
study of absenteeism by Hackett, Bycio and Guion (1989) might increase
. its explanatory value.

Manuscript received February 1988


Revised manuscript received June 1989

REFERENCES
Abelson, M. A . (1987). Examination of avoidable and unavoidable turnover. Journal of
Applied Psychology, 72. 382-386.
Bernardin. H. J & Bcatty, R. W. (1984). Performance appraisal: Assessing human behavior
a[ work. Belmont, Calif.: PWS-Kent.
Beyer. J. M. & Trice, H. M. (1984). A field study of the use and perceived effects of
discipline in controlling work performance. Academy of Managemenr Journal, 27. 743-
764.
Boudreau. J . W. & Berger, C. J. (1985). Decision-theoretic utility analysis applied to
employee separations and acquisitions (Monograph). Journal of Applied Psychology. 70,
581-612.
Cascio. W . F. (1987). Applied psychology in personnel managemenr (3rd ed.). Englewood
Cliffs, N.J.: Prenticc-Hall.
Cascio. W . F. & Valenzi, E. R. (1978). Relationships among criteria of police performance.
Journal of Applied Psychology. 63-22-28.
Dalton. D. R . , Krackhardt. D. M.,& Porter, L. W . (1981). Functional turnover: An
empirical assessment. Journal of Applied Psychology, 66. 716-721,
Dalton. D . R.. Todor. W . D.. & Krackhardt. D. M. (1982). Turnover overstated: The
functional taxonomy. Academy of Managemenf Review. 7 , 117-123.
Dreher. G . F. (1982). The role of performance in the turnover process. Academy of
.Uanagemenr Journal. 25. 137-147.
Farr. J . L . , OLeary. B. S . . & Bartlett. C. J . (1971). Ethnic group membership as a
moderator of the prediction of job performance. Personnel Aychology. 2 4 , 609-636.
Farris. G . F. (1971). A predictive study of turnover. Personnel Psychology. 2 4 , 31 1-328.
Farris. G . F. & Lim. F. G . (1969). Effects of performance on leadership. cohesivenesr.
influence. satisfaction. and subsequent performance. Journal of Applied Psychology. 5.J,
490-497
JOB PERFORMANCE AND TURNOVER 61

Ferguson. G . A . (1976). Srarisrical analysis in psychology and educarion (4th ed.). New
York: McGraw-Hill.
Forbes, J . B. (1987). Early intraorganizational mobility: Patterns and influences. Academy
of Management Journal, 30, 110-125.
Glass, G . V . , McGaw, B.. & Smith. M . L. (1981). Mera-analysis in social research. Beverly
Hills: Sage.
Hackett, R . D . , Bycio, P . , & Guion. R . M. (1989). Absenteeism among hospital nurses: A n
idiographic-longitudinal analysis. Academy of Managemenr Journal, 32, 424-453.
Hollenbeck, J. R . & Williams, C. R . (1986). Turnover functionality versus turnover fre-
quency: A note on work attitudes and organizational effectiveness. Journal of Applied
Psychology, 71, 606-61 1.
Hunter. J . E.. Schmidt. F. L . , &Jackson. G . B. (1982). Mera-analysirt Cumularing research
findings acrms srudies. Beverly Hills: Sage.
laffaldano, M . T. & Muchinsky. P. M. (1985). Job satisfaction and pcrformance: A meta-
analysis. Psychological Bullerin, 97. 251-273.
Jackofsky, E . F. (1984). Turnover and job performance: An integrated process model.
Academy of Managernenr Review, I , 74-83.
Johns, G. (1989). Performance and turnover cognitions among managers. Canadinn Journal
of Adminisrrarive Sciences, 6 , 37-42.
Kanfer. R . , Crosby. J . V., & Brandt. D. M . (1988). Investigating behavioral antecedents of
turnover at three job tenure levels. Journal of Applied Psychology, 73, 331-335.
Kemery. E. R.. Dunlap. W. P . , & Griffeth, R . W. (1988). Correction for variance restric-
tion in point-biserial conclations. Journal of Applied Psychology, 73, 688-691.
Knod, E. M., Jr. (1975). The development and testing of a model of employee turnover.
Dirserrarion Absrracu fnfernarional,35. 4243B (University Microfilms N o . 75-3434).
Lance, C. E. (1988). Job performance as a moderator of the satisfaction-turnover intention
relation: An empirical contrast of two perspectives. Journal of Organizarional Behavior.
9. 271-280.
Latham. G . P. & Wexley. K . N. (19x1). Increating producrivlry rhrough performance
appraisal. Reading: Addison-Wesley.
Lee, T. W. & Mowday, R . T. (1987). Voluntarily leaving an organization: An empirical
investigation of Steers and Mowday's model of turnover. Academy of Managemenr
Journal. 30. 721-743.
McEvoy. G . M . & Cascio, W. F. (1985). Strategies for reducing employee turnover: A
meta-analysis. Journal of Applied Psychology. 70. 342-353.
McEvoy, G . M. & Cascio. W. F. (1987). Do good o r poor performers leave? A meta-
analysis of the relationship between pcrformance and turnover. Academy of Managemenr
Journal. 30, 744-762.
Martin, T. N., Price. J . L., & Mueller. C . W. (1981). Job performance and turnover.
Journal of Applied Psychology, 66. 116-1 19.
Mlller. S. J . (1988). Employee turnover: An investigation of attitudinal differences between
functional and dysfunctional leavers. Dlsrerrarion A bsrracrr fnrernarional. 49. 29058
(University Microfilms No. 88-10381).
Mobley. W. H . (1982). Employee rumover: Causes, consequences, and conrrol. Reading,
Mass.: Addison- Wesley.
Morita. J . G . . Lee. T. W . . & Mowday, R . T. (1989). Introducing survival analysis to
Organizational researchers: A selected application to turnover research. Journal of
Applied Psychology. 74, 280-292.
Ofsanko. F . J . (1979). Employee turnover by job performance level. Dlsserrarion Absrracrs
Inre-rnarional. 40, 2419B (University Microfilms No. 79-24567).
O'Reilly. C. A , . I l l . & Weitz, B A . (1980). Managing marginal employces: The use of
warnings and dismissals. Adminlsrrarive Science Quarrerly. 25. 467-484.
Parasuraman. S . & Alutto. 1. A . (1984). Sources and outcomes of stress in organizational
settings: Toward the development of a structural model. Academy of Managemenf
Journal. 27. 330-350.
Parsons. C. K.. Herold, D. M . . & Leathemood. M. L. (1985). Turnover during initial
employment: A longitudinal study of the role of causal attributions. Journal of Applied
Psychology, 70. 337-341.
Peters, L. H . & Sheridan, J . E. (1988). Turnover research methodology: A critique of
traditional designs and a suggested survival model alternative. In G . F. Ferris & K. M.
Rowland (Eds.), Research in personnel and human resources managernen:: A research
a n n u l (pp. 231-262). Greenwich: JAI Press.
Price. J . L. (1977). The srudy offurnover. Ames, Iowa: Iowa State Press.
Sackett. P. R . , H a m s , M. M . , & Orr, J . M. (1986). O n seeking moderator variables in the
meta-analysis of correlational data: A Monte Carlo investigation of statistical power and
resistance to Type I error. Journal of Applied Psychology. 71, 302-310.
Schmidt. F. L.. Hunter, J . E.. Pearlman. K . . Hirsh. H. R., Sackett, P. R., Schmitt N. . .
Tenopyr. M. L.. Kehoe. J . . & Zedeck, S . (1985). Forty questions about validity gener-
alization and meta-analysis with commentaries. Personnel Psychology, 32, 257-281.
Seashore, S . E . , Indik, B. P., & Georgopoulos. B. S. (1960). Relationships among criteria
of job performance. Journal of Applied Psychology, 44, 195-202.
Spector, P. E. & Levine. E. L. (1987). Meta-analysis for integrating study outcomes: A
Monte Carlo studv of its susceptibility to Type I and Type 11 errors. Journal of Applied
P~ychology,7 2 . 3-9.
Steel. R. P. & Ovalle. N . K. (1984). A review and meta-analysis of the relationship between
behavioral intentions and employee turnover. Journal of Applied Psychology, 69, 673-
686.
Steers, R. M . & Mowday. R. T. (1982). Employee turnover and the postdecision accom-
modation processes. In R. T. Mowday. R . M. Steers, & L. W . Porter. Employee-
organizational linkages: The psychology of commirmenr. absenreeirrn. and rurnover (pp.
123-131). New York: Academic Press.
Stumpf. S. A . & Dawley, P. K . (1981). Predicting voluntary and involuntary turnover using
absenteeism and performance indices. Academy of Managemenr Journal, 24. 148-163.
Turnage. J . J . & Muchinsky. P. M. (1984). A comparison of the predictive validity of
assessment center evaluations versus traditional measures in forecasting supervisory job
performance: Interpretive implications of criterion distortion for the assessment para-
digm. Journal of Applied Psychology, 6 9 , 595-602.
Wanous. J . P.. Stumpf. S . A . . & Bedrmian. H. (1979). J o b survival of new employees.
Personnel Psychology, 32. 651-662.
Wanous. J . P . . Sullivan. S . E.. & .Malinak. J. (1989). The role of judgment calls in
meta-analysis. Journal of Applied Psychology, 74. 259-264.
Wells. D. L. & Muchinsky. P. M. (1985). Performance antecedents of voluntary and
involuntary managerial turnover. Journal of Applied Psychology, 70. 329-336.

REFERENCE NOTES
1 . Hunter. J. E. & Schmidt, F. L. (1987). Dichorornizing confinuorcr variables: The implica-
rioru for mera-unolysu Unpublished manuscript. Michigan State University. Psychology
Dcpanment.
2 . Scybolt. J . W . & Ross, S . C. (1985). Work role design and rhe rerenrion of high
performing personnel. A sunrval analysir. Unpublished manuscript. Cnivcrsity of Utah,
Department of Management.
JOB PERFORMANCE AND TURNOVER 63

3. Summers, T. & Hendrix, W. (1988). Modeling the role of pay equity: A field study.
Manuscript in preparation, Clemson University. Management Department.
4. Bluedorn, A. C. & Abelson,M. A. (1980). Employeeperformance and withdrawalfrom
work. Unpublished manuscript. The Pennsylvania State University, College of Business
Administration.
5. Rosse, J. G. (1986). The role of job performance in models of turnover. Paper presented
at the 46th Annual Meeting of the Academy of Management, Chicago.
Appendix A

TABLE A1
S t u d i e s Involving Job P e r f o r m a n c e a n d Voluntary Turnover

.supervisory KairngslKunkmgs
Knod (1Y7.5) Accountants at a large public Partner potenlial versus not partner 164 -0.18"
accounting firm. potential. N o reliability data.
La Kocco. Pugh. & Guiiderson Navy enlisted men aboard combat M o s ~recent performance marks. 462 -0.07'
(1Y77) ships No reliability data.
Sheridan & Vredenburgh (1978) Feniale nurses and nurses' aides. Kating summed over 5 dimensions. 216 -0.02
u = 0.77.
Ofsanko (IY7Y) Managerial, administrative. and Above average performers versus 3986 -0.02"
professional employees at a average ones. No reliability data.
profit-making utility.
Waiious, Stunipf, & Bedrosian Low wage workers hired via a state Kating of overall job performance. 1140 -0.36'
(197')) employment agency. u = 0.96.
Bluedorn & Abelsoii (1980) Clerical employees at an insurance Kating of overall job performance. 158 0.03
company, 95% female. No reliability data.
Larson ( 1982) Sales deparlmerit staff at a Raring of overall j o b performance. 95 -0.01
pharmaceutical house, 80% male. No reliability data.
Martin, Price, & Mueller (1981) Female nurses at a voluntary Average rating across 6 dimensions. I62 0.13
hospital. Split-half reliability = 0.98.
Spencer & Steers (1981) liospital employees who held a Kating summed over 1 1 dimensions. 2Y5 -0.07
variety o f positions. a = 0.82.
Stumpf & Dawley (1981) Full-time bank tellers hired Trichotomy reflecting the average 189 -0.30'
between 1970and 1976. ratiiig oblained by a feller during
hidher entire tenure.
Stumpf & Dawley (1981) Full-time bank tellers hired Trichotomy reflecting the average 188 -0.21'
between 1977 and 197X. rating obtained by a teller during
hidher entire tenure.
Dreher (1982) Exempt professional. managerial Composite of rankings and ratings. 468 -0.41'
and technical employees from an No reliability data.
oil company.
Beyer & Trice (1984) Employees of a manufacturing firm Average rating a c r o s 8 dimensions. 435 -0.14.
who were identified as ' a = 0.91.
disciplinary problems.
Larson & Fukami (1984) Male blue-collar employees in the Rating summed over 6 dimensions. 108 -0.31'
transportation department of a a = 0.86.
newspaper.
Parasuraman & Alutto (1984) Mainly clerical and blue-collar Rating summed over 3 dimensions 217 -0.W
workers at a food processing u = 0.85.
plant. 67% male.
Parsons. flerold. & Leatherwood Female hotel room cleaners. Rating summed over 4 dimensions. 31 -0.21
(1985) u = 0.84.
Seybolt & Ross (1985) Registered nurses. Linear combination of 12 dimensions. 312 -0.13
a = 0.96.
Sheridan (1985) Recently hired nurses and nurses' Hating summed over 5 dimensions. 84 -0.28'
aides, 96% female. N o reliability data.
Wells & Muchinsky (1985) Managers of branch offices of a retail Rating o n each of 12 dimensions. 280 -0.31b
credit institution. %O/O male. No reliability data.
Jackofsky. Ferris. & Breckenridge Male accountants at a public Rating of overall job performance. 169 -0.29'
(1986) accounting firm. Inter-rater reliability = 0.62.
Sager (1986) Sales staff at a national consumer Average rating across 87 items. 89 -0.29'
goods manufacturer. a = 0.98.

(Continued)
Copeet al. (19x7) Employees at a mental health Rating of overall job performance. 202 -0.07
facility, 38% male. No reliability data.
Lee & i
Mowday (1987) Full-time employees at a financial Hating of overall job performance. 445 -0.01
institution, 45% male. No reliability data.
Mossholder et al. (1988) First-line supervisors at a textile Rating uf overall job performance. 365 -0.oY
firm. 07% male. n = 0.93.
Self-rarrngr
Rluedorn & Abelson (1980) Clerical employees at an insurance Hating of overall job performance. I58 0.05
company. 95% feinale. No reliability data.
Stunipf & liartnian (IYM) Individuals who used an on-cainpus Average rating across 6 dimensions. SS -0.20'
university placement service. a = 0.85.
Parsons, tierold. & Leatherwood Female hotel room cleaners. Ratings of performance quality and 31 -0. l o b
(1985) quantity. u = 0.77.
Non-ruling Oulpur indices
Jackofsky, Ferris, & Breckenridge Owner-operator truck drivers employed Monthly revenue earned by the driver. 107 -0.35'
(1Y86) by a trucking company. Test-rerest reliability = 0.62.
IIollenbeck & Williams Department store salespersons, Average sales volume over three I12 -0.07
(lY86) 6Y% female. months. Test-retest reliability =
0.63.
Kaiifer, <'roshy. 51 Uraridt (1YR8) Three groups of plant operators each Average '7'0 of standard achieved daily. 20 I - 0 . IS"
differing in job tenure, No reliability data.
74% female.
Non-raring Qualiry Indices

Mirvis & Liiwlcr (1077) Bank tellers, 6% male Number of teller shortages over each 1W -0.15d
o f three months. Average stability
coefficient = 0.71.
Stumpf & Dawley (1981) Full-time bank tellers hired Trichotomy reflecting the number 189 -0.41'
between 1970 and 1976. and amount of cash imbalances
over the entire tenure of a teller.
Stumpf & Dawley (1981) Full-time bank tellers hired Trichotomy reflecting the number 188 -0.12
between 1977 and 1978. and amount of cash imbalances
over the entire tenure of a teller.
Beyer & Trice (1984) Employees of a manufacturing firm who Average of findings o n whether written 461 -0.17"
were identified as disciplinary warnings or suspensions were given.
problems.
Larson & Fukami (1984) Male blue-collar employees in the Number of official administrative 108 -0.41'
transportation department o f a warnings given.
newspaper.
Kanfer, Crosby, & Brandt (1988) Three groups of plant operators Number of products mis-processed 201 -0.12''
each differing in j o b tenure. No reliability data.
74% female.
Summers & t k n d r i x (1988) Managers at various levels in a Controllable costs divided by 137 -0.41.
restaurant organisation. total revenue.
Non-raring Proniolional Indices
La Kocco, Pugh. & Gunderson Navy enlisted men aboard combat ships. Pay grade achieved. 543 0.00
(1977)
L a n o n (1982) Sales department staff at a % merit salary increase. 113 -0.06
pharmaceutical house, 80% male.
Stumpf & Dawley (1981) Full-time bank tellers hired Number of promotionslpay increases 189 - 0.33'
between 1970 and 1976. receivpd per year.
Stunipf & Dawley (1981) Full-time bank tellers hired Number of proniotiondpay increases 188 -0.81'
between 1Y77 and 1978. received per year.
Q) - ____. _ _ _ ____ .-
d
(Continued)
TABLE A1
(Continued)

I n vesrigaror(s) SubjecrslSefring Performance Measure n rpb


__ ___--
Dreher (1Y82) Exempt professional. managerial Difference between the initial and 484 -0.21'
and technical employees from an highest j o b level attained.
oil company. divided by the number of years
of continuous service.
Gruhbs (1987) Employees of a corporation between Average of promotional rate and 467 -0.21'
1980 and 1985.25% male. average monthly pay increases.
Kanfer. Crosby. & Ijrandt (1988) Three groups of plant operators Whether o r not a pay increase had 201 0.26"'
each differing in j o b tenure. been given.
74% female.

Nore. Unless otherwise indicated. all the correlations shown are point biserial. They have been standardised so that a ncgarive value implies that
the slayers i n the sample performed better than the leavers.
' P < 0.05. or better.
"As the performance measure involved was a dichotomy, the resulting correlation is a phi coefficient as opposed to a point biserial.
"I'his is a mean correlation. obtained by averaging across the performance-rating dimensions.
'This is a mean correlation, obtained by averaging across the three tenure groups. using the latest performance appraisal information available for
each group.
d.
l'his is the median o f three correlations that were each based o n a single month of data collection.
TABLE A2
S t u d i e s Involving J o b Performance a n d Total Turnover

In ves/igaror(s) su bjeclsl~ielring Performance Measure n 'P b

Supervisory RaringslRankings
Lopez (1966) Female toll collectors. Alternation ranking. 82 -0.11
N o reliability data.
Farris (1971) Engineers employed by an electronics High versus low performers. lnter- 192 -0.03"
firm. rater reliability = 0.70.
Farris (1971) Scientists employed by a High versus low performers. Inter- 203 -0.24'.
pharmaceutical firm. rater reliability = 0.70.
Knod (1Y75) Accountants at a large public Partner potential versus not partner 186 -0.32"'
accounting firm. potential. N o reliability data.
Brief & Aldag (1976) Nursing aides and assistants. Percentage of items checked as 77 -0.32'
strengths. N o reliability data.
Katz (1978) Government employees from a variety Average rating across 2 items. 80 -0.33'
of job categories. Internal consistency = 0.81
Larson (1982) Sales department staff at a Rating of overall job performance. 104 -0.08
pharmaceutical house, 80% male. N o reliability data.
Rasmussen (1Y81) Hospital aides. 60% male. Rating of overall job performance. 267 -0.17'
No reliability data.
Keller (1984) Professional. supervisory, clerical, Rating summed over 5 dimensions. 1y0 -0.26'
and hourly staff at a GI = 0.90.
manufacturing firm.
Turnage & Muchinsky (1Y84) Manufacturing supervisors who had Rating of overall job performance. 750 -0.10
been promoted after participation No reliability data.
in an assessment centre.
O'Connor et al. (1984) First. second, and third level Rating summed over between 8 and 11 1450 -0.20'
managers at a convenience store dimensions. a = 0.94.
organisation.
U TABLE A2
0
(Continued)

S U b]ecrrl.~rrling Perforniance h l e u u r e II

Parsons. Herold. & Leathewood Female hotel room cleaiiers Rating summed over 4 dimensions. 47 --0.05
(1~x5) u = 0.84.
Wells & hluchinsky (1985) Managers of branch offices of a Rating on each of 12 dimensions. 420 -0.38
retail credit institution. No reliability data.
86% male.
Oldhani et a1 (1986) Electronic data processing personnel Rating of overall job performance. 230 -0.02
at a state government. 70% male No reliahility data.
ROSW ( iwo) Newly hired hospital nursing and Rating of overall job performance. 42 -0.24
housekeeping personnel. H3% fenialc. Test-retest reliability = 0.63.
Cope et al. (IYX7) Employees at a mental health facility. Kating of overall ]oh performance. 2x8 -0.17
38% male. No reliability data.
Colarelli. Dean. & Koiista~is Entry-level accountants at Big Composite of 3 items. 280 0.w
(IY87) Eight firms. 61% male. = 0.82.

St,/f-rurings
Brief & Aldag (1Y70) Nursing aides and assistants. Single-item rating of performance 77 -0.02
quality.
Marsh & Mannari (1977) Mostly assembly line personnel at a Composite of 5 items 1033 -0.14
Japanese electronics factory. u = 0.57.
50% male.
Fisher (198.5) Newly graduated nurses at various Rating summed over 7 items. 210 -0.05
hospitals, 5% male. Test-retest reliability = 0.90.
Vincent & Dugan ( l Y 6 2 ) Life insurance salesinen hired Average number of sales points per 35 I -0.39
in 1958. month. Reliability (two 6-month
periods) = 0.84.
Vincent & Dugan (1962) Life insurance salesmen hired Average number of sales points per 310 -0.76
in IY5Y. month. Reliability (two 6-month
periods) = 0.84.
Kingstrom & Mainstone (1985) Sales personnel for a large YO of dollar sales quota attained 109 -0.20'
manufacturer of data processing over a 10-month period. N o
equipment. rcliability data.
Non-raring Qualify Indices
Lopez (1966) Female toll collectors. Composite of cash shortages. ticket 182 -0.06
shortages, and axle count errors.
Farr, O ' l x a r y , Cli Bartlett Female toll collectors. Cash imbalances and axle count errors. 129 O.Mb
(1971) Spearman-Brown reliability = 0.77.
Non-raring Promorioiial Itidices
Larson (1982) Sales department staff at a % merit salary increase. 122 -0.03
pharmaceutical house. 80% male.
Turnage & Muchinsky (198.1) Manufacturing supervisors who had Number of promotions past foreinan 750 -0.02
been promoted after participation level.
in an assessment centre.
Kingsrrom & Mainstone (1985) Sales personnel for a large Whether or not a promotion had 109 0.35'"
manufacturer of data processing k e n given over a 2-year period.
equipment.

Nore. Unless otherwise indicated, all the correlations shown are point biserial. They have been standardised s o that a negarive value implies that
the slayers in the sample performed better than the leavers.
' P < 0.05, or better.
"As the performance measure involved is a dichotomy. the resulting correlation is a phi coefficient as opposed to a point biserial.
"This is an average o f two findings. one involving cash imbalances, the other, axle count errors.
4 TABLE A3
N S t u d i e s Involving Job Performance a n d Involuntary Turnover
~ ~ - _ _
I n vesriguror(s) It
Sublet rslSerring Perforniuritr Meusure 'pb
- -~ __----__-- _--_ ~ ______ -
Supervisory HurinRslHaiikmKs
-
h o d (1Y75) Accountants at a largc public Partner potential versus not partner 152 -0.40'"
accou n t I ng f i rni , partner potential. No reliability
data.
La Rocco, Pugh, & Gunderson Navy enlisted men aboard combat Most recent performance marks. 148 -0.45**
(1Y77) ships. No reliability data.
Wanous. Stumpf. & Bedrosian Low wage workers hired via a state Rating of overall job performance. 1110 -0.40'
( 1Y79) employment agency. a = 0.Y6.
Stumpf & Dawley (1Y81) Full-time bank tellers hired Trichotomy reflecting the average 258 -0.36'
between 1970 and 1976. rating obtained by a teller during
hidher entire tenure.
Stumpf & Dawley (IY81) Full-time bank tellers hired 'Trichotomy reflecting the average 151 -0.66'
between 1977 and 1978. rating obtained by a teller during
hisiher entire tenure.
13eyer & Trice ( 1984) Employees of a nianufacturing firm Average rating across 8 dimensions. 432 -0.13'
who were identified as u = 0.91.
disciplinary problems.
Larson & Fukdml (1984) Male blue-collar cmployces in the Rating summed over 6 dimensions, 95 -0.42'
transportation department of a u = 0.86.
newspaper.
Wells Sr Muchinsky (1Y85) Managers of branch offices o l a Rating o n each of 12 dimensions. 280 -0.49''
retail credit institution, No reliabilily data.
X6% male.
Parsons. Herold. & Leatherwood Female room cleaners. Rating summed over 4 dimensions. 32 -0.23
(1985) u = 0.84.
Cope et al. (1987) Employees at a niental health Rating of overall job performance. 86 -0.50'
facility, 38% male. No reliability data.
Self-ratings
I'arsons, Ilerold, & Leatherwood Female room cleaners. Katings of performance quality 37 -O.OSb
(1985) and quantity. Q = 0.77.
Non-rating Qualiry Indices
Stumpf & Dawley (1981) Full-time bank tellers hired between Trichotomy reflecting the number and 258 -0.50'
1970 and 1976. amount of cash imbalances over the
entire tenure of a teller.
Stumpf & Dawley (1981) Full-time bank tellers hired between Trichotomy reflecting the number and 151 -0.04
1977 and 1978. amount of cash imbalances over the
entire tenure of a teller.
Heyer & Trice (1984) Employees of a manufacturing firm Average of findings o n whether written 460 -0.52"
who were identified as disciplinary warnings or suspensions were given.
problems.
Larson & Fukarni (1984) Male blue-collar employees in the Number of official administrative 95 -0.35'
transportation department of a warnings given.
newspaper.
Summers & Ilendrix (1988) Managers at various levels in a Controllable costs divided by 131 -0.29.
restaurant organisation. total revenue.
Non-raring Promorionul Indices
La Kocco. Pugh. & Gunderson Navy enlisted men aboard combat Pay grade achieved. I74 -0.60'
(1977) ships.
Stumpf & Dawley (1981) Full-time bank tellers hired between Number of promotiondpay increases 258 -0.39.
1Y70 and 1976. received per year.
Stumpf & Dawley (1981) Full-time bank tellers hired between Number of promotiondpay increases 151 -0.84'
1977 and 1978. received per year.

Nore. Unless otherwise indicated, all the correlations shown are point biserial. They have been standardised so that a negarivc value implies that
the slayers in the sample performed better than the leavers.
' P < 0.05, or better.
"As the performance measure involved is a dichotomy, the resulting correlation is a phi coefficient as opposed to a point biserial.
J. bThis is a mean correlalion, obtained by averaging across the performance-rating dimensions.
w
74 BYCIO, HACKElT, ALVARES

Appendix B
Srudies Included in [he Performance-Turnover Mera-Analyses
Beycr. J. M. & Trice, H . M . (1984). A field study of the use and perceived effects of
discipline in controlling work performance. Academy of Managemenr Journal. 27. 743-
764.
Bluedorn, A. C. & Abelson, M. A . (1980). Employee performance and withdrawal from
work. Unpublished manuscript. The Pennsylvania State University. College of Business
Administration.
Brief. A . P. & Aldag. R . J . (1976). Correlates of role indices Journal of Applied Psychol-
ogy. 61. 468-472.
Colarelli. S . M..Dean. R . A . . & Konstans, C. (1987). Comparative effects of personal and
situational influences o n job outcomes of new professionals. Journal of Applied Psychol-
ogy, 72.558-566.
Cope, J. G . . Grossnickle, W. F.. Covington. K. B., Durham, T. W.. & Zaharia, E. S.
(1987). Staff turnover as a function of performance in a public residential facility.
American Journal of Menial Dcficiency, 92. 151-154.
Dreher. G . F. (1982). T h e role of performance in the turnover process. Academy a/
Managemenr Journal. 25, 137-147.
F a n , J. L.. OLeary. B. S . . & Bartlett. C. J . (1971). Ethnic group membership as a
moderator of the prediction of job performance. Personnel Psychology, 2 4 . 609-636.
Farns. G . F . (1971). A predictive study of turnover. Personnel Psychology. 2 4 . 311-328.
Fisher. C . D. (1985). Social support and adjustment to work. A longitudinal study. Journal
O f Managemenr. 11. 39-53.
Grubbs. M. R . (1987). T h e prediction of voluntary employee turnover for a commercial
bank. Disserratron Absrracts Inrernarional. 48, 2387.4 (University Microfilms No. 87-
28233).
Hollenbeck. J . R . & Williams. C . R. (1986). Turnover functionality versus turnover fre-
quency: A note on work attitudes and organizational effectiveness. Journal of Applied
Psychology, 71, 606-611.
Jackofsky. E. F.. F e m s . K. R.. & Breckcnridge, B. G. (1986). Evidence of a curvilinear
relationship between j o b performance and turnover. Journal of Managemenr, 12. 105-
111.
Kanfcr. R . . Crosby, J . V . . & Brandt. D. M. (1988). Investigating behavioral antecedentsof
turnover at three job tenure levels. Journal of Applied Psychology. 73, 331-335.
Katz, R. (1978). The influence of job longevity o n employee reactions to task characteristics.
Human Relarionr. 8. 703-725.
Keller, R. T. (1984). T h e role of performance and absenteeism in the prediction of turnover.
Academy of Managemenr Journal. 27. 176-183.
Kingstrom, P. 0. & Mainstone. L. E . (1985). An investigation of the rater-ratee acquaint-
ance and rater bias. Academy of Managemenr Journal. 28. 641-653.
Knod, E. M.. J r . (1975). T h e development and testing of a model of employee turnover.
Dirsenation Abstracts Inrernational. 35, 42438 (University Microfilms No. 75-3434).
La Rocco, J . M.. Pugh, W . M.. & Gunderson. E . (1977). Identifying determinants of
retention decisions. Personnel Psychology. 30. 199-215.
Larson. E . W . (1982). Employee commitment to an organization and the effects of per-
ceived ease of movement. Dirserrarion Absrracts Inrernarional. 43, 246A (University
Microfilms No. 82-14202).
Larson, E. W. & Fukami. C . V . (1984). Relationships between worker behavior and
commitment to the organzation and union. Proceedings of rhe 44rh Annual Meering of [he
Academy of Managemenr. 222-216.
Lee. T. W. & blowday R . T. (1Y87). Voluntanly leaving an organization: An empirical
JOB PERFORMANCE AND TURNOVER 75
investigation of Steers and Mowdays model of turnover. Academy of Management
Journal. 30. 721-743.
Lopez, F. M., Jr. (1966). Current problems in test performance of job applicants: I.
Personnel Psychology, 19, 10-18.
Marsh, R. M. & Mannari, H. (1977). Organizational commitment and turnover: A predic-
tion study. Administrative Science Quarterly. 22, 57-75.
Martin, T. N., Price, J. L.. & Mueller, C . W. (1981). Job performance and turnover.
Journal of Applied Psychology, 66. 116-119.
Minis, P. H. & Lawler, E . E.. 111. (1977). Measuring the financial impact of employee
attitudes. Journal of Applied Psychology, 62. 1-8.
Mossholder, K. W.. Bedeian. A . G.. Norris, D. R., Giles, W. F.. & Feild, H. S. (1988). Job
performance and turnover decisions: Two field studies. Journal of Managernenl, 24,
403-4 14.
OConnor, E. J., Peters, L. H., Pooyan, A., Weekley. J.. Frank, B., & Erenkrantz, B.
(1984). Situational constraint effects o n performance. affective reactions, and turnover:
A field replication and extension. Journal of Applied Psychology, 69. 663-672.
Oldham. G . R.. Kulik, C. T., Ambrose. M. L.. Stepina. L. P., & Brand, J. F.
(1986). Relations between job facet comparisons and employee reactions. Organizational
Behavior and Human Decision Processes, 38, 28-47.
Ofsanko. F. J. (1979). Employee turnover by job performance level. Disserrarion Absiracrr
Inrernarional, 40, 24198 (University Microfilms No. 79-24567).
Parasuraman, S. & Alutto, J . A . (1984). Sources and outcomes of stress in organizational
settings: Toward the development of a structural model. Academy of Managemenr
Journal, 27, 330-350.
Parsons, C. K., Herold. D. M . , & Leathenvood. M. L. (1985). Turnover during initial
employment: A longitudinal study of the role of causal attributions. Journal of Applied
Psychology, 70, 337-341.
Rasmussen, B. R. (1981). The relation between training type, role relations. performance,
absenteeism and turnover among state hospital aides. Dirsertation Abstracts Inrernational.
42, 3M2B-3M3B (University Microfilms No. 81-28679).
Rosse, J. G. (1986). The role of job performance in m0del.s of rumover. Paper presented at
the 46th Annual Meeting of the Academy of Management, Chicago.
Sager, J . K. (1986). Towards a managerial model of salespeoples turnover behavior:
Development and testing of a turnover model using consumer goods sales force data,
Dlssertation A bstracts Internarional, 47, 266lA (University Microfilms No. 86-25438).
Seybolt, J . W. & Ross, S . C. (1985). Work role design and rhe retention of high performing
personnel: A survival analysis. Unpublished manuscript, University of Utah, Management
Department.
Sheridan, J . E . (1985). A catastrophe model of employee withdrawal leading to low job
performance, high absenteeism. and job turnover during the first year of employment.
Academy of Management Journal, 28.88-109.
Sheridan. J. E. & Vredenburgh. D. J . (1978). Usefulness of leadership behavior andsocial
power variables in predicting job tension, performance, and turnover of nursing
employees. Journal of Applied Psychology, 63. 89-95.
Spencer, D . G. & Steers, R. M. (1981). Performance as a moderator of the job satisfaction-
turnover relationship. Journal of Applied Psychology. 66, 511-514.
Stumpf. S. A . & Dawley, P. K. (1981). Predicting voluntary and involuntary turnover using
absenteeism and performance indices. Academy of Managemenr Journal. 24, 148-163.
Stumpf, S. A . & Hartrnan. K. (1984). Individual exploration to organizational commitment
or withdrawal. Academy of Management Journal. 27,308-329.
Summers. T. & Hendrix, W. (1988). Modeling the role of pay equiry: A field study.
Manuscript in preparation. Clemson University. Management Department.
76 BYCIO, HACKETf, ALVARES

Turnage. J . J . & Muchinsky. P. M. (1984). A comparison of the predictive validity of


assessment center evaluations versus traditional measures in forecasting supervisory job
performance: Interpretive implications of criterion distortion for the assessment para-
digm, Journal of Applied Psychology. 69. 595-602.
Vincent. N . L. & Dugan, R. D. (1%2). Validity information exchange. Personnel Psychol-
ogy. IS. 223-226.
Wanous. J . P., Stumpf, S. A . . & Bedrosian, H . (1979). Job survival of new employees.
Personnel Psychology, 32, 651-662.
Wells, D. L. & Muchinsky. P. M. (1985). Performance antecedents of voluntary and
involuntary managerial turnover. Journal of Applied Psychology. 70. 329-336.

Das könnte Ihnen auch gefallen