Anda di halaman 1dari 20

Human Relations

http://hum .sagepub.com /

A Unified Model of Turnover from Organizations


Allen C. Bluedorn

Human Relations 1982 35:

135
DOI: 10.1177/001872678203500204
The online version of this article can be found at:
http://hum.sagepub.com/content/35/2/135

Published by:

SAGE
http://www.sagepublications.com

On behalf of:

THE
TAVISTOCK
INSTITUTE
The Tavistock Institute

Additional services and information for Hum an Relations can be found at:
Email Alerts: http://hum.sagepub.com/cgi/alerts
Subscriptions: http://hum.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://hum.sagepub.com/content/35/2/135.refs.html

>> Version of Record - Feb 1, 1982


What is This?
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Human Relations, Volume 35, Number 2, 1982, pp. 135-153

A Unified Model of Turnover from


Organizations1
Allen C. Bluedorn
University o f Missouri Columbia

A model o f the turnover process is developed by synthesizing three turnover


models: those o f Price and M obley and the model which has developed
around the organizational commitment variable. This m odel is tested via
path analysis and is generally supported. An attem pt to cross-validate the
new model also provided reasonable support fo r it.

INTRODUCTION
Interest in the topic o f turnover has never been higher than has been
demonstrated over the last five years. In addition to a great deal of em
pirical study, the development o f causal models including variables from
many different domains has been a major theoretical activity. While many
of these models emphasize different parts of the turnover process, they
tend to be more complementary than contradictory.
The present study is an attempt to develop a more complete under
standing o f the turnover process by synthesizing three o f these recently
developed turnover models. The three models to be synthesized are (i) the
causal model constructed by Price (1977), (ii) the model which has de
veloped around the cpocept o f organizational commitment, and (iii) the

This research was funded by a grant from the Research Projects Fund, Center for Research,
College of Business Administration, The Pennsylvania State University and by a Faculty
Summer Fellowship from the College o f Business Administration, The Pennsylvania State
University. Parts of this manuscript were presented at the 40th Annual Meeting of the
Academy of Management, Detroit, Michigan, August 9-13, 1980.
2Requests for reprints should be sent to Allen Bluedorn, Department of Management, College
of Business and Public Administration, University of MissouriColumbia, Columbia,
Missouri 65211.
135
0 0 1 8 -7 2 6 7 /8 2 /0 2 0 0 -0 1 3 5 $ 0 3 .0 0 /l 1982 T avistock Institute o f H um an R elations

Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014


from the SAGE Social Science Collections. All Rights Reserved.

Bluedorn

136

model developed by Mobley (1977) which specifies the linkages between


job satisfaction and turnover. Following the construction o f the integrated
model, an empirical test of it will be presented. No attempt will be made
to comprehensively review the turnover literature. The reader is referred
to Muchinsky and Tuttle (1979), Mobley, Griffeth, Hand, and Meglino
(1979), and Bluedorn (1982) for the most recent comprehensive reviews of
this literature.
THEORETICAL INTEGRATION
The Price M odel
Price (1977, pp. 66-91) developed a model o f the turnover process
which portrays this process as beginning with a series o f structural and
individual determinants o f job satisfaction (i.e., centralization, pay, com
munication, etc.). The individuals satisfaction level then determines the
probability of an individual staying in or leaving the organization, con
tingent upon the state o f the economy. That is, satisfaction and the op
portunity structure (the state o f the economy) should interact, and turnover
should be most likely for very dissatisfied people in economies of high
opportunity. Few people, regardless o f satisfaction level, should leave in
periods o f low opportunity. A final premise o f this model is that individual
demographic characteristics (age, length o f service, education, etc.) should
not have independent causal impacts once the variables in the model have
been taken into account.
This model has been tested empirically by Price and Bluedorn (1979),
Bluedorn (1979), Martin (1979), Dickson (1977), and Price and Mueller
(1979). These five tests (on nurses, U.S. Army officers, white collar ad
ministrative-clerical-professional workers, nurses, and nurses again,
respectively) have produced a series o f uniform results. All five tests have
rejected the interaction between job satisfaction and the opportunity
structure. All have, however, supported the positioning o f opportunity,
first suggested by Bluedorn (1976), as causally prior to satisfaction in the
model, and effecting turnover (or turnover intentions) indirectly through
job satisfaction.
A second uniform finding has been that the model does not eliminate
the independent effects o f demographic variables. Age, length o f service,
and other demographics frequently retained powerful independent effects,
despite the explanatory power o f the other variables in the model.
Overall, the basic structure of Prices model has been supported
in five studies with two fundamental changes: (i) the elimination of the
job satisfaction X opportunity interaction with opportunity now specified
as one o f a series of determinants o f satisfaction which impact turnover
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Unified Model of Turnover

137

indirectly through satisfaction; and (ii) the recognition that the model
does not eliminate all demographic variables as important causes o f
turnover.

The Organizational Commitment M odel


Organizational commitment is the strength o f an individuals identi
fication with and involvement in an organization (Porter, Steers, Mowday,
& Boulian, 1974). As a variable which has come to generate considerable
interest, organizational commitment has been studied as both an in
dependent and a dependent variable (Alutto, Hrebiniak, & Alonso, 1973;
Hrebiniak& Alutto, 1972; Kidron, 1978; Marsh & Mannari, 1971; Schoenherr & Greeley, 1974; Stevens, Beyer, & Trice, 1978). Several findings are
important for the theoretical issue under discussion.
First, organizational commitment has been significantly and consis
tently related to turnover (Koch & Steers, 1978; Porter et al., 1974; Porter,
Crampon, & Smith, 1976; Steers, 1977). Secondly, a number of the same
variables which have been found to effect satisfaction have also been
related to organizational commitment. Third, Porter, et al. (1974) found
that job satisfaction and organizational commitment were distinct concepts
empirically. Furthermore, Marsh and Mannari (1977) hypothesized that
job satisfaction was a determinant o f commitment and empirically sup
ported this hypothesis in their research. Price and Mueller (1979) have
also made this suggestion theoretically and provided some empirical
support for it. Overall, organizational commitment has been consistently
related to turnover, and may itself be caused by several o f the determinants
of satisfaction in Prices model, as well as by satisfaction itself.
The M obley M odel
Mobley (1977) has developed a very detailed model of the linkages
between job satisfaction and turnover, an area basically undeveloped in
either Prices model or in the work on organizational commitment. For
the purposes o f the current theory development effort, Mobleys model
will be simplified to the following sequence: dissatisfaction leads to job
search which leads to an intent to quit or stay which leads to the individuals
actual staying or quitting behavior.
While there is little work, aside from Mobley, Horner, and Hollings
worth (1978), which empirically tests the first two linkages, there is a
large body o f research confirming the final linkage between intentions to
leave or stay and the actual behavior of staying or leaving. This material
is reviewed in Porter and Steers (1973) and Bluedorn (1976; 1982).
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Bluedorn

138

The Unified Model


Figure 1 presents a diagram o f the model which incorporates the three
previous models. The variables leading directly to job satisfaction were
suggested in the previous models or in the work empirically testing the
models. The overall satisfaction linkage was present in the original Price
model. The work o f Marsh and Mannari (1977) suggests the position o f

Promotion
+
opportunities----Centralization
Formalization

Job
satisfaction

Instrumental
communication

Equity

----------

Pay ---------------Routinization

Member -------
integration

Organizational
commitment

Job
search

Intent
to leave

Environmental
opportunities

Foregone
environmental
opportunities

Turnover

Role
_
conflict -------- Length
of service-------Age --------------Education -------Marital
status------------Fig. 1. The unified model of turnover.

Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Unified Model of Turnover

139

organizational commitment, and Mobleys (1977) model suggests the


position o f job search and intent to leave. Table I presents the variable
definitions and the hypothesized effects of each variable on turnover.
This model is formulated in the integrative mode suggested by
Roberts, Hulin, and Rousseau (1978) in that it includes individual (attitudinal and demographic), organizational, and environmental variables.
Although the empirical test to follow deals with individual perceptions
of or reactions to the organizational and environmental factors, future
research can use this model to follow the strategy suggested by Roberts
et al. (1978) more precisely.
Pierce and Dunham (1978) have discussed the convergence of many
individual and organizational level measures and how variable names
change with the level o f analysis. In order to maintain continuity, the
following discussion will retain the organizational and environmental
level variable labels even though the analysis will be performed at the
individual level.

METHOD
Sample
Data were gathered from the employees, largely women (92% and
94% in two samples, respectively), in the operations division o f a large
insurance company on two separate occasions. The first occasion consisted
o f a questionnaire administration which was primarily a pilot study for
use in the development o f a basic questionnaire. The second question
naire administration took place four months later and utilized a refined
version o f the original questionnaire. Data on the divisions employees
were also obtained from company records on both occasions.
Table I. Definitions of the Variables in the Unified Model and Anticipated Overall Relation
ships with Turnover

Variable
Promotion
opportunity
Centralization

Definition
The probability that an individual will
be able to occupy roles within the orga
nization that offer greater rewards
The extent to which power is con
centrated in a social system (Price
& Bluedorn, 1980)

Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Anticipated
overall
relationship with
turnover

Bluedorn

140
Table I. Continued

Variable
Formalization

Instrumental
communication
Equity

Pay

Routinization

Member integration

Environmental
opportunity
Environmental
opportunities
foregone

Potential role
conflict

Length of service
Age
Education
Marital status
Job satisfaction

Organizational
commitment

Definition
The degree to which the norms of a
social system are explicit (Blau &
Scott, 1962)
The extent to which information about
role performance is transmitted to orga
nization members (Price & Bluedorn, 1980)
The extent to which an individuals job
inputs and/or outcomes are perceived as
equivalent to the inputs and/or outcomes
of those in the individuals reference
group [adopted from Adams (1963) de
finition of inequity ]
The money, fringe benefits, and other
commodities that have financial value
which organizations give to employees
in return for their services (Lawler, 1971)
The extent to which role performance
in an organization is repetitive (Price &
Bluedorn, 1980)
The extent to which a member partici
pates in primary and/or quasiprimary
relationships within the organization
(Price & Bluedorn, 1980)
The number and quality of unoccupied
roles in an organizations environment
(Bluedorn, 1979)
The number and quality of unoccupied
occupational roles in an organizations
environment that an organizational
member has bypassed or rejected
(Bluedorn, 1979)
The probability that two or more sets of
pressures will occur such that compliance
with one set will make more difficult com
pliance with the other(s) [adopted from
Kahn, Wolfe, Quinn, and Snoeks (1964)
definition of role conflict]
The amount of time an individual has been
a member of an organization
An individuals chronological age
The amount of an individuals formal
schooling or training
Whether an individual is married or not
The affective orientations of individuals
to the work roles which they occupy
(Vroom, 1964)
The strength of an individuals identifica
tion with and involvement in an organiza
tion (Porter et al., 1974)
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Anticipated
overall
relationship with
turnover
-

"

141

Unified Model of Turnover

Both questionnaires were also filled out by about 20 supervisors


and managers, and the vice-president who headed the division. These
individuals were not included in the analysis, not because position in the
organizational hierarchy is not a potentially important variable, but because
the small number o f these individuals would make claims about hierarchical
position unconvincing. Since these individuals could influence the results,
the strategy suggested by Kerlinger (1973, p. 309) o f making the sample
homogeneous was adopted, and these individuals were excluded. Analyses
including these respondents revealed essentially the same results as pre
sented in Fig. 2 with the one exception being that marital status becomes a
statistically significant exogenous variable.

Instrumental 32
information

77/

Job
jf
Satisfaction

Promotion
Opportunities
Routinization
Education
Foregone
environmental
opportunities
Environmental
opportunities
Fig. 2. A path analysis o f the unified model. Note: The matrix of correlations
between the exogenous variables is available from the author.
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Bluedorn

142

Measures
Several o f the measurement instruments included on both question
naires are modified versions o f the scales developed and reported by Price
and Bluedorn (1979), and which have also been used by Martin (1979)
and Price and Mueller (1979). The measures (instrumental communication,
member integration, routinization, centralization, equity, and environ
mental opportunity) were modified, except for member integration, by
creating a common 7-point response format for the items in each scale.
The measures o f formalization, foregone envrionmental opportunities,
promotion opportunities, role conflict, and leaving or staying intentions
also used the same 7-point response format.
Formalization was measured with five questions suggested by the
scale developed by Hage and Aiken (1969) asking about the extent to
which rules and procedures were established and known in the company
and work unit.
Foregone environmental opportunities were measured with the same
basic questions used in the environmental opportunity scale, but with the
questions directing the respondent to evaluate his or her job possibilities
at the time the respondent first came to work with the insurance company.
Gutek (1978) has demonstrated the accuracy of this type o f retrospective
questioning.
Potential role conflict was measured by two questions asking the
respondent to rate the chances that he or she would quit the company if
asked to do so by a spouse.
Staying or leaving intentions were measured with six questions from
the Staying or Leaving Index (SLI) (Bluedorn, 1982). The six questions
formed two groups: three questions which asked respondents to rate
their chances o f still working for the company three, six, and twelve
months from now, and three other questions, placed in a different part
o f the questionnaire, which asked respondents to rate their chances o f
quitting the company three, six, and twelve months from now. Recoding
the first three questions, the six items are summed to form the index.
The SLI as reported in Bluedorn (1982) actually consists of eight
questions. The two questions asking the respondent to forecast over a two
year period were deleted because data on actual turnover was being col
lected for one year following each questionnaire administration.
Satisfaction was measured with an index formed from two questions
designed to measure overall job satisfaction. Respondents were asked:
What is your opinion o f (the company) as a place to work? They
could then select one response from a list o f 21 adjectives which ranged
from absolutely perfect to disgusting. The second question asked
respondents, Which o f the phrases would you pick to describe this
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

143

Unified Model of Turnover

company as a place to work if you could not use the phrase you picked in
question (the previous question)?
Organizational commitment was measured with the organizational
commitment scale of Porter et al. (1974). A comprehensive review of this
scales properties has been made by Mowday, Steers, and Porter (1978).
Job search was measured by the single question, How many times
have you looked for another job during the last three months? Only count
the times since you have been working for this company if you have worked
less than three m onths. Although this measure asks the respondent to
report on job search behavior in the recent past, it is assumed that this
response will also reflect the respondents relative amount of search be
havior in the near future.
Marital status was measured with the question, What is your
marital status? The respondent then checked one o f a number of cate
gories. The categories have been combined to form the dichotomy: married
(coded 1) or unmarried (coded 0).
Age, length o f service, pay and education were obtained from
company records.
Turnover data (stayers = 0; leavers = 1) were obtained by monthly
examinations o f the divisions termination reports and roster of employees.
The termination reports were used to distinguish involuntary (including
pregnancy terminations) from voluntary separations as suggested by
Bluedorn (1978). Only voluntary terminations were utilized.
The study was designed to track employees for a one year period
following each questionnaire administration. This was accomplished for
respondents in the first sample, but respondents were only traced for
ten months following the second sample. The reason these respondents
were not tracked for the full 12 months was that a new vice-president,
who did not wish to continue the study, was placed in charge o f the division
near the end o f the tracking period.
The reliabilities o f the indexes as measured by Cronbachs (1951)
alpha coefficient, as well as the means and standard deviations of all
variables, are presented in Table II.

STATISTICAL ANALYSIS
Multiple Regression
The first phase of the statistical analysis was a series of five fullequation multiple regressions (presented in Table III) which utilized data
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Bluedorn

144
Table II. Descriptive Statistics of the Measures Used in the Study

Variable
Centralization
Instrumental
information
Environmental
opportunity
Foregone environ
mental
opportunities
Promotion
opportunity
Member
integration
Equity
Routinization
Potential role
conflict
Job satisfaction
Organizational
commitment
Intent to leave
Job search
Marital status
Pay
Length of service
Age
Education
Turnover

Mean,
standard deviation,
number of items in
index

Reliabilities (Cronbachs alpha)


---------------------------------------------------First survey
Second survey Mean*2

(8.30; 3.62; 3)

.92

.74

.86

(60.24; 9.41; 12)

.91

.88

.90

(15.30; 4.19; 3)

.94

.91

.93

(13.60;4.25; 3)

.94

.89

.92

(18.90; 6.17; 5)

.93

.89

.91

(15.07; 5.73; 4)
(18.09;6.46; 5)
(11.86; 4.53; 3)

.69
.79
.63

.73
.94
.91

.71
.89
.81

(9.16; 3.83; 2)
(29.20; 5.27; 2)

.81
.85

.85
.87

.83
.86

(68.39; 15.69; 15)


(15.40; 9.21; 6)
(0.34; 0.99; 1)
(1.53; 0.50; 1)
(5 68.8;676.23;/?^)
(24.64; 34.68;/?)
(27.48; 10.06;/?)
(12.42; 1.77;/?)
(0.21; 0.41; R)

.90
.85

.90
.95

.90
.91

^Calculated via Fishers (1921) Z r transformation.


bR indicates data obtained from company records.

from the second sample (the first sample will be used for a cross valida
tion procedure reported later). The purpose o f these multiple regressions
was to reduce the number o f variables involved in the analysis. Therefore,
only those variables with significant (p < .05) direct effects on any of the
five criterion variables (satisfaction, organizational commitment, job
search, intent to leave, turnover) were selected for inclusion in the sub
sequent analyses.
During this phase o f the analysis, it was discovered that the oper
ationalizations o f formalization and instrumental communication were
producing a partialling fallacy (Gordon, 1968). A factor analysisprinciple
factor method (Harman, 1976)revealed that the combined items o f the
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Criterion variables
Variables

Job
satisfaction

Organizational
commitment
Beta

Beta
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Promotion opportunity
Centralization
Instrumental information
Routinization
Equity
Pay
Member integration
Environmental opportunities
Foregone environmental opportunities
Potential role conflict
Length of service
Age
Education
Marital status
Job satisfaction
Organizational commitment
Job search
Intent to leave
R2
Adjusted R 2

(.42)
(.10)
(.43)
(-.2 9 )
(.50)
(-.0 1 )
(-.0 2 )
(-.2 3 )
(-.1 9 )
(-1 0 )
(-.0 2 )
(.19)
(-0 2 )
(.08)

.15
-.0 8
.32*
-.0 9
.34e
.12
.05
.00
-.0 9
-.0 6
-.1 0
.23^
.05
.01
-

(-75)
(-.2 3 )
(-.3 7 )
.46
.40

(.42)
(.20)
(.40)
(-3 9 )
(.37)
(.03)
(.01)
(-.3 0 )
(-2 5 )
(-2 3 )
(.10)
(.28)
(-.2 5 )
(.21)
(.75)

(-.3 0 )
(-.5 3 )

Job search

.74
.71

.18<*
.00
.05
-A 9d
- .I S C
-.0 7
.03
-.0 8
-.0 9
- .I K
.02
-.0 7
-.2 3 *
.10
.61*
-

Intent to leave

Beta
(-2 7 )
(-.1 6 )
(-.0 6 )
(.25)
(-2 8 )
(-.0 3 )
(.07)
(.20)
(-00)
(.09)
(-0 5 )
(-.1 7 )
(.12)
(16)
(-.2 3 )
(-.3 0 )

-.1 5
-.0 2
.08
.12
-.0 7
-.0 5
.05
.29
-.2 1 *
.01
.08
-.1 2
.07
-.0 8
-.0 2
-.0 8

(.31)
.22
.11

Turnover

Beta
(-.2 6 )
(-.1 2 )
(-.2 8 )
(.26)
(-.2 3 )
(.00)
(-.0 4 )
(.18)
(.10)
(.18)
(-.0 3 )
(-1 5 )
(.15)
(-2 0 )
(-.3 7 )
(-.5 4 )
(.31)
.34
.25

.01
.02
-.1 1
.04
-.0 4
-.01
-.0 6
.00
-.0 2
.09
.05
.03
.01
-.1 0
.10
-.4 6
.16
-

Beta
(-.0 2 )
-.0 2
- .0 9
(-1 5 )
(-0 4 )
.06
.13
(.18)
(-0 3 )
.16
(-0 5 )
.05
(.04)
.06
.27*
(.17)
(.02)
- .0 9
(.10)
.08
(-1 9 )
-.1 8
(-2 4 )
-.1 5
-.0 2
(.08)
- .0 2
(-1 4 )
-.1 9
(-1 1 )
(-.1 4 )
.30
(.19)
.04
(.28) ,2Sd
.22
.11

a Minimum pair-wise TV =141.


^The numbers in parentheses are Pearson Product Moment Correlation coefficients. The betas are standardized partial regression coefficients.
Cp < .05.
<*p < .01.

*p < .001.

Unified Model of Turnover

Table III. Multiple Regression Solutions to Reduce the Number of V a ria b le s^

Bluedorn

146

two scales formed an acceptable construct. The one factor solution ac


counted for 43% o f the total variance and all items had a loading o f
.43 or higher. This construct is labeled instrumental information since both
operationalizations involve information concerning the performance o f
role activities.
As is revealed in Table III, centralization, pay, member integration,
length o f service, and marital status were not significantly related to any
o f the criterion variables. Consequently, they will be excluded from
further consideration in the evaluation of the synthetic model.
A test for Prices (1977) often rejected job satisfaction X environ
mental opportunity interaction was also made during this phase utilizing
the procedure verified by Allison (1977). The proposed interaction Was
not significant statistically ( p < .05).
Path Analysis
The variables found to be significantly related to the criterion vari
ables in the multiple regression analyses were utilized in a path analysis o f
the model depicted in Fig. 1. The resulting path diagram is presented in
Fig. 2.
The path diagram has been simplified by deleting all paths which
were not significant at the .05 level. Thus, a total o f 37 out o f a possible
55 direct paths were eliminated. One of the advantages o f path analysis
over ordinary multiple regression is the explication o f one variables total
causal impact on another. Table IV presents the total causal effects of
the exogeneous variables (and intervening variables, where appropriate)
on the five criterion variables. An examination o f this table indicates the
most important determinants o f turnover are, in order: environmental
opportunity, intentions to stay or leave, routinization, and age. As is
revealed in this table, it is very difficult for a variable to have a large impact
on another when a direct path is absent. This is illustrated by the negligible
total effects (on turnover) o f instrumental information, equity, potential
role conflict, and promotion opportunities.

Cross Validation
Muchinsky and Tuttle (1979) have recommended that predictive
turnover research should be cross-validated from one sample to another.
An attempt to cross-validate the synthetic turnover model developed in this
research was made by using the prediction equations developed from the
path analysis reported earlier (based on data from the second questionDownloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Table IV. Total Effects of the Predictor Variables on Each Criterion Variable*7
Criterion variable
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Job
satisfaction
Variable

lE b

Instrumental information
Equity
Age
Potential role conflict
Promotion opportunity
Routinization
Education
Environmental opportunity
Environmental opportunity foregone
Job satisfaction
Organizational commitment
Job search
Intent to leave

Organizational
commitment

Intent to leave

DE c

TE^

IE

DE

TE

IE

.32
.34
.20

.32
.34
.20

.19
.20
.12

-.1 6
.12
-.1 0
.18
-.2 1
-.2 1

.19
.04
.24
-.1 0
.18
-.2 1
-.2 1

-.1 0
-.0 2
-.1 2
.05
-.0 9
.11
.11

Job search

DE

TE

-.1 0
-.0 2
-.1 2
.05
-.0 9
.11
.11

IE

.60

.60

.30
-

-.5 0
.17

a A discussion of total effects can be found in Alwin and Hauser (1975) and Lewis-Beck (1974).
^IE = Indirect effect.
CDE = Direct effect.
^TE = Total effect.

-.3 0
-.5 0
.17

DE

.28
-.2 2

Turnover
TE

.28
-.2 2

IE
-.0 3
.01
-.0 3
.01
-.0 2
.03
.03
.01
.01
-.0 8
-.1 4
-

DE

TE

-.0 3
.01
-.2 3
.01
-.0 2
.23
.03
.30
.01
-.0 8
-.1 4

-.2 0
-

.20
_
.29
-

.27

.27

148

Bluedorn

naire administration and its attendant turnover) to predict the same five
criterion variables in the data collected in the first questionnaire ad
ministration. Only variables with direct effects on a variable were used to
predict it.
The prediction equations used are as follows:
*

= .363*6 + .198*7 + .9 6 4 * 8

X2 =

2.02 *

+ .368

X 9-

.686

X l0 -

(1)
.304 * 6

- .506 * n - 2 .1 2 2 * 12

(2)

* 3 = .8 6 2 * 3 - .5 6 3 * 4

(3)

(4)

= 1.571

X5=

* 3

.2 8 5 *

.115*4 + .1 2 6 * 3 + .9 4 6 ^ 2 + .7 7 5 *

where:
*

= Job satisfaction

* 2 = Organizational commitment
*3

= Job search

= Intent to leave

* 5 = Turnover
* 6 = Equity
*

= Organizational information

* 8 = Age
* 9 = Promotion opportunities
* 10 = Potential role conflict
* i = Education
* 2

= Routinization

* 3 = Environmental opportunities
* 4 = Foregone environmental
opportunities
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

(5)

149

Unified Model of Turnover


Table V. Results of the Cross-Validation Attempted

Predicted Variable

O riginal/?2

R 2 in the
cross
validation

Job satisfaction
Organizational commitment
Job search
Intent to leave
Turnover

.40
.69
.06
.30
.12

.06
.47
.13
.17
.03

Original
explained
variance
retained, %
15
68
217
57
25

and the coefficients are the unstandardized regression coefficients from


the prediction equations produced in the second data set. Table V presents
the results o f the cross-validation.
The cross-validation was most successful in reproducing the original
explained variance for Job Search, Organizational Commitment, and
Intent to Leave, reproducing over one half o f the original explained vari
ance in each case, and actually explaining more o f the job search variation
than was explained originally. The attempts to repredict job satisfaction
and turnover were less successful as 15% and 25% o f their original ex
plained variances were reproduced, respectively.

DISCUSSION
Overall, the form o f the general model developed in the theoretical
synthesis was supported in its empirical examination. Some modifications
are suggested, however. First, the position o f the job search variable was
not confirmed where it was placed in either Mobleys (1977) original model
or in the present theoretical construction. Instead, it appears to be related
neither to organizational commitment nor to job satisfaction. It appears to
be related to the individuals perception o f past and present environmental
opportunities.
A second result which was not specified in the synthetic model is
the presence o f the direct paths from the exogenous variables to organiza
tional commitment and turnover. Although Price and Mueller (1979) have
suggested some direct paths from the exogenous variables to organizational
commitment, the empirical confirmation of these paths in their research
is weakened by their operationalization o f it as intent to leave.
Of greater interest are the direct paths from age, routinization, and
environmental opportunities to turnover. In several other studies these
variables have been found to have direct effects on turnover, independent
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Bluedorn

150

o f the intervening variable satisfaction, and in one case, the intervening


variables satisfaction and intent to leave. Dickson (1977) and Price and
Bluedorn (1979) report direct paths from age to turnover. Dickson also
reported a direct path from routinization to turnover. A direct path from
environmental opportunities to turnover was found by Price and Mueller
(1979) and to intent to stay or leave by Martin (1979) and Bluedorn (1979).
Martin also found a direct path between age and intent to stay or leave.
Given the appearance o f these direct paths in the present research,
and the frequency of their occurrence in other research, it is reasonable to
add the specification to the model that direct paths, as well as indirect,
can be expected to occur from routinization, environmental opportunities,
and age to turnover. Although it did not occur in the present study, length
o f service, independent o f age, has been found to produce a direct path
to turnover (Dickson, 1977; Price & Bluedorn, 1979; Price & Mueller,
1979) and intentions to stay or leave (Bluedorn, 1979) in a similar manner.
The specification of this direct path as a likely occurrence is another reason
able modification of the unified model.
The attempted cross-validation of the model by predicting the crite
rion variables in another data set seems to have produced reasonable sup
port for the model. Most successful in reproducing the explained variance
in organizational commitment, job search, and intent to leave, the model
did, nonetheless, reproduce one-fourth of the variance originally ex
plained in turnover.
One reason for the shrinkage in the amount of turnover explained
is that some o f the turnover originally explained in the second sample was
also recorded for members o f the first sample as turnover. This could
happen as some individuals were in both samples and quit after they took
the second questionnaire. Mobley, Griffeth, Hand, and Meglino (1979)
have suggested that the predictive power o f intent to leave statements
should decay as the time period o f the prediction increases. Shenk (1972)
has demonstrated this effect in a large sample o f U.S. Air Force officers.
Thus, using an equation derived from predictions much closer to the actual
turnover behavior of many respondents would be expected to result in
substantial shrinkage when used to predict behavior from an earlier data
set (as in the case of using the equation from the second sample to predict
behavior from the first). This appears to have been what happened in
this case.

CONCLUSION
A model o f turnover from organizations developed by synthesizing
three earlier models was generally supported by a path analysis and a
Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

151

Unified Model of Turnover

subsequent cross-validation in another data set. The repositioning o f job


search was, however, suggested in the empirical results. In addition,
because o f their frequent appearance in other research and their reoccur
ence in the present study, the specification of direct paths from environ
mental opportunity, routinization, age, and length o f service to turnover
is suggested in addition to their indirect paths. The model seems generalizable to other types of organizations, and therefore appears to deserve
additional testing and refinement by other researchers.
REFERENCES
ADAMS, J. S. Toward an understanding of inequity. Journal o f Abnormal and Social
Psychology, 1963, 67, 422-436.
ALLISON, P. D. Testing for interaction in multiple regression. American Journal o f So
ciology, 1977, 83, 144-153.
ALUTTO, J. A., HREBINIAK, L. G., & ALONSO, R. C. On operationalizing the concept
of commitment. Social Forces, 1973, 51, 448-454.
ALWIN, D. F., & HAUSER, R. M. The decomposition of effects in path analysis. American
Sociological Review, 1915,40, 37-47.
BLAU, P. M., & SCOTT, W. R. Formal organizations. San Francisco: Chandler, 1962.
BLUEDORN, A. C. A taxonomy of turnover. Academy o f Management Review, 1978, 3,
647-651.
BLUEDORN, A. C. A causal model of turnover in organizations (Doctoral dissertation,
University of Iowa, 1976). Dissertation Abstracts International, 1977, 37, 8004-A (Uni
versity Microfilms No. 77-13,057, 196).
BLUEDORN, A. C. Structure, environment, and satisfaction: Toward a causal model of
turnover from military organizations. Journal o f Political and Military Sociology,
1979, 7, 181-207.
BLUEDORN, A. C. The theories of turnover: Causes, effects, and meaning. In S. Bacharach
(Ed.), Perspectives in Organizational Sociology: Theory and Research (Vol. ').
Greenwich, Conn.: JAI Press, 1982.
CRONBACH, L. J. Coefficient alpha and the internal structure of tests. Psychometrika, 1951,
16, 297-334.
DICKSON, M. E. A test o f a causal model o f turnover in organizations. Unpublished Masters
Thesis, University of Iowa, 1977.
FISHER, R. A. On the probable error of a coefficient of correlation deduced from a small
sample. Metron, 1921, 1, 3-32.
GORDON, R. A. Issues in multiple regression. American Journal o f Sociology, 1968, 73,
592-616.
GUTEK, B. A. On the accuracy of retrospective attitudinal data. Public Opinion Quarterly,
1978, 42, 390-401.
WAGE, J., & AIKEN, M. Routine technology, social structure, and organizational goals.
Administrative Science Quarterly, 1969, 14, 366-376.
HARMAN, H. H. M odem factor analysis (3rd ed.). Chicago: University of Chicago Press,
'
1976.
HREBINIAK, L. G., & ALUTTO, J. A. Personal and role-related factors in the development
of organizational commitment. Administrative Science Quarterly, 1972, 18, 555-573.
KAHN, R. L., WOLFE, D. M., QUINN, R. P ., & SNOEK, J. D. Organizational stress:
Studies in role conflict and ambiguity. New York: Wiley, 1964.
KERLINGER, F. L. Foundations o f behavioral research (2nded.). New York: Holt, Rinehart,
and Winston, 1973.
KIDRON, A. Work values and organizational commitment. Academy o f Management
Journal, 1978, 21, 239-247.

Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Bluedorn

152

KOCH, J. L., & STEERS, R. M. Job attachment, satisfaction, and turnover among public
sector employees. Journal o f Vocational Behavior, 1978, 12, 119-128.
LAWLER, E. E. Pay and organizational effectiveness. New York: McGraw-Hill, 1971.
LEWIS-BECK, M .S. Determining the importance of an independent variable: A path analytic
solution. Social Science Research, 1974,5, 95-107.
MARSH, R. M., & MANNARI, H. Lifetime commitment in Japan: Roles, norms, and values.
American Journal o f Sociology, 1971, 76, 795-812.
MARSH, R. M., & MANNARI, H. Organizational commitment and turnover: A prediction
study. Administrative Science Quarterly, 1977, 22, 57-75.
MARTIN, T. N., Jr. A contextual model of employee turnover intentions. Academy o f
Management Journal, 1979, 22, 313-324.
MOBLEY, W. H. Intermediate linkages in the relationship between job satisfaction and
employee turnover. Journal o f Applied Psychology, 1977, 62, 237-240.
MOBLEY, W. H ., GRIFFETH, R. W., HAND, H. H ., & MEGLDMO, B. M. Review and
conceptual analysis of the employee turnover process. Psychological Bulletin, 1979,
86, 493-522.
MOBLEY, W. H ., HORNER, S. O., & HOLLINGSWORTH, A. T. An evaluation of pre
cursors of hospital employee turnover. Journal o f Applied Psychology, 1978, 63,
408-414.
MOWDAY, R. T., STEERS, R. M., & PORTER, L. W. The measurement o f organizational
commitment: A progress report. Technical Report No. 15. Eugene, Oregon: Graduate
School of Management, University of Oregon, 1978.
MUCHINSKY, P. M., & TUTTLE, M. L. Employee turnover: An empirical and meth
odological assessment. Journal o f Vocational Behavior, 1979, 14, 43-77.
PIERCE, J. L., & DUNHAM, R. B. An empirical demonstration of the convergence of
common macro- and micro-organization measures. Academy o f Management Journal,
1978, 21, 410-418.
POSTER, L. W., & STEERS, R. M. Organizational, work, and personal factors in employee
turnover and absenteeism. Psychological Bulletin, 1973, 80, 151-176.
PORTER, L. W., STEERS, R. M., MOWDAY, R. T., & BOULIAN, P. V. Organizational
commitment, job satisfaction, and turnover among psychiatric technicians, Journal
o f Applied Psychology, 1974,59, 603-609.
PORTER, L. W., CRAMPON, W. J., & SMITH, F. J. Organizational commitment and
managerial turnover: A longitudinal study. Organizational Behavior and Human
Performance, 1976, 15, 87-98.
PRICE, J. L. The study o f turnover. Ames: Iowa State University Press, 1977.
PRICE, J. L., & BLUEDORN, A. C. Test of a causal model of turnover from organizations.
In D. Dunkerley & G. Salaman (Eds.), International Yearbook o f Organization Studies
1979. London: Routledge & Kegan Paul, 1980, 217-236.
PRICE, J. L., & MUELLER, C. W. A causal model o f turnover estimated fo r nurses. Paper
presented at the 39th National Academy of Management Meeting, Atlanta, 1979.
ROBERTS, K. H., HULIN, C. L., & ROUSSEAU, D. M. Developing an interdisciplinary
science o f organizations. San Francisco: Jossey-Bass, 1978.
SCHOENHERR, R. A., & GREELEY, A. M. Role commitment processes and the American
Catholic priesthood. American Sociological Review, 1974, 39, 407-426.
SHENK, F. Predictability o f expressed career intent. Air Force Human Resources Laboratory
Technical Report No. 72-25, Lackland Air Force Base, Texas, 1972.
STEERS, R. M. Antecedents and outcomes of organizational commitment. Administrative
Science Quarterly, 1977, 22, 46-56.
STEVENS, J. M., BEYER, J. M., & TRICE, H. M. Assessing personal, role, and orga
nizational predictors of managerial commitment. Academy o f Management Journal,
1978, 21, 380-396.
VROOM, V. H. Work and motivation. New York: Wiley, 1964.

Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Unified Model of Turnover

153

BIOGRAPHICAL NOTE
ALLEN C. BLUEDORN is an Assistant Professor of Management at the University of
Missouri-Columbia. He received his BS, MA, and PhD from the University of Iowa. His
published work has appeared in journals and books such as the Academy o f Management
Review, Business Horizons, Journal o f Political and Military Sociology, Sociology and Social
Research, and Middle Range Theory and the Study o f Organizations. He is currently con
ducting research on the impact of turnover on organizations.

Downloaded from hum.sagepub.com at TUFTS UNIV on September 28, 2014

Anda mungkin juga menyukai