Professional Documents
Culture Documents
703
704 Academy of Management Journal December
(belief that the outcome in question is necessary and sufficient for achieve-
ment of a second outcome).
Two types of valence exist in Vroom's formulation. The valence of
possible outcomes (such as status and recognition, which are usually
positively valent, and working long hours, which is usually negatively val-
ent) is labelled Vki the valence of actual job performance is labelled Vj. The
valence of job performance is hypothesized to be the result of a multiplicative
interaction between (a) the individual's subjective perception that certain
job activities will facilitate achievement of desired outcomes and (b) those
outcomes viewed as desirable by the individual. Specifically, the interaction
is as follows:
VJ =J^
where Vj = the valence of performance level j
Ijk = the instrumentality of outcome j for the attainment
of outcome k
Vk = the valence of outcome k
n = the number of outcomes
Expectancy is defined as a momentary belief of an individual that an act
on his part will be followed by a given outcome. The value of the ex-
pectancy associated with any action-outcome pair may range from 0 (no
relationship perceived) to + 1 (complete certainty that the performance of
the act will result in the outcome). Vroom hypothesizes that expectancy
combines multiplicatively with the valence of job performance as follows:
Recently, a number of writers (2, 15, 16, 24) have questioned many of
the features and assumptions inherent in Vroom's basic expectancy model.
Criticisms of the theory range from concerns about the manner in which
the variables are conceptualized to questions about the descriptive accuracy
of the assumptions underlying expectancy theory. The research described
here pursues the line of questioning the descriptive accuracy of the norma-
tive postulates underlying expectancy theory.
Both Vroom's original expectancy statement and the elaborations and
refinements proposed by others place heavy emphasis on the expectancy
"core," i.e., the hypothesized 2(Ei]Vj) relationship which purportedly
predicts effort. This central core is actually a work motivation variant of
the subjective expected utihty (SEU) theory of decision making (20) and
assumes that individuals systematically analyze the value of performance and
the likelihood that it can be achieved before coming to a level-of-effort
decision. The mathematical similarity between SEU theory and expectancy
theory is evident when the formal statement of each is presented:
METHOD
The subjects were 54 lower and middle level managers taking a manage-
ment development course at a Canadian university. Questionnaires de-
signed to operationalize the basic expectancy variables were administered
seven times over a three month period as follows:
Instrumentality
Subjects were asked to indicate on seven point summated scales the
relationship they perceived between successful completion of the course
and the attainment of both positively and negatively valent outcomes, with
—3 stated as "I will definitely (valent outcome inserted) even if I do not
successfully complete this course" and + 3 stated as "I will definitely (valent
outcome inserted) if I successfully complete this course."
Valence
Each subject was initially presented with a list of 17 outcomes that have
been used in other expectancy studies and asked to indicate his or her six
most valent (four positive and two negative) outcomes. Each individual's
choice pattern was recorded and a unique questionnaire was generated for
each of the 54 subjects for the seven subsequent data gathering points. As
noted above, it is important when measuring valence to ensure that de-
sirabihty, not importance, is assessed. Accordingly, valence was operation-
alized on seven point summated scales, with —3 stated as "This outcome
would displease me greatly if it occurred," and +3 stated as "I would be
greatly pleased if this outcome occurred." Test-retest reliabilities were
computed across all valences (range—.76 to .97; median—.91); but since
each subject picked different outcomes, the meaning of the reliability coef-
ficient is not clear. Computation of test-retest reliabilities could be done us-
ing common subsets of valence choices; this procedure would result in
coefficients higher than those indicated.
Expectancy
Each subject was asked to indicate his or her subjective perception that
four specific behaviors (e.g., regularly reading the text, regularly attending
class, etc.) would lead to successful completion of the course. The subjects
were asked to indicate their perceptions in a "chances in 10" format which
allowed conversion of subject responses to probability assessments even
though many of the subjects had no understanding of formal probability
theory. The internal reliability of these four questions was assessed by com-
puting a Spearman-Brown split half reliability coefficient which equalled
.75.
708 Academy of Management Journal December
Report
1.50
1.50
OS'l
OS'l
0.00
1.50
1.50
1.50
1.00
8
1.00
1.50
1.00
Self-
Effort
-0.50
o
o o 8 o o
>n
o o o o o in 8 o oo oin
fc 2i 0 >n
o r) r) r- 0o0 '~- o o
140
0 aCfc)
16
CTv r- o r)
8 CTV 00 00
oo 00
1 1 1 1 1 1 1 1 1 1
r-) m VO r- >n r. ro VO
Report Subfect
No.
o
in
8
.50
.00
.00
.50
.50
.50
.50
.50
00
00
00
.00
.00
Self-
Effort
o o o o
= oin o m o
8 8 o 8 8 8 8
00
00
o in
«n TJ- o m
Tj- o CT\ OO oc VO fO
CTv
1 1 1 1 1 1 1 1 1 1 1 1
i-~
8 8 8
.50
.50
50
50
50
50
50
50
.50
.00
00
o o
o
o 8 in
r- r-l o
o o
o o 8 »n o
m
m m
in
OC m o CTv o o
80
00 00 VO
.0 aifc! oo m r-l ?i 00
VO VO
1 1 1 1 1 1 1 1 1
No. Week
-n VO tn VO r) rt
Subfect
rj
o o o o o o o o o o O O §
•Ao^ m o o o m 8 o o m O
o o o o o
1 1
o o o o
8 8 8 8 8 8
75
o o o
muted
f- t--
Com-
t~ in
Effort
oo o
80
00 CTv
!S
28
m VO CTv
VO
1 1 1 1 1
Week
VO >n — VO •n
s o
1975 Volume 18, Number 4 709
Effort
A self-report (seven point scale ranging from —3 to -|-3) of effort
expended by each subject on a week by week basis was used. Subjects were
asked to indicate the amount of time they had spent during the preceding
week on class-related activities such as reading the text, taking class notes,
etc.
As the internal and test-retest reliabilities indicate, the difficulties often
encountered in using survey techniques to test expectancy theory as noted
by DeLeo and Pritchard (5) did not arise in this study. Correlation analysis
was used to assess the extent to which actual individual decision patterns
conformed to the transitivity and independence postulates.
Independence Postulate
The data generated to test the descriptive accuracy of the independence
postulate are presented in Table 2. In order to determine the extent to which
the independence postulate is an accurate representation of individual be-
havior, the covariation between individual preferences for various outcomes
and the expectancy that these could be attained was analyzed. Table 2
indicates the relationship found between both types of valence (Vj's and
Vk's) and expectancy. Since the Vj component is a sum, only one figure is
generated for each person, while there are six Vk's because there are six
second level outcomes. The expectancy (Eij) component is the sum of the
four specific behaviors noted above divided by 4. (To be conceptually
precise, the only correlation which explicitly tests the independence postu-
late is that between Eij and Vj; however, the correlations between Eij and
the individual Vk's are also included.)
An inspection of the data contained in Table 2 indicates a fairly general
distribution of significant correlations between both types of valence and
expectancy. Although at an intuitive level this is detrimental to expectancy
theory, it is more important to come to some conclusions about individual
cases. When this is done, it becomes apparent that some individuals do not
violate the postulate at all; others show multiple violations. Of the 54 re-
spondents, 27 exhibit at least one significant correlation between one Vk
and Eij; 15 respondents show a lack of independence for the summed Vk's
( V J ) . In addition to the significant correlations generated, a substantial
number of coefficients are quite large (although not statistically significant
because the A^ for each subject is only seven).
712 Academy of Management Journal December
TABLE 2
Correlations Between Both Types of Valence and Expectancy
Correlations
Re- Between Valence
spond- Correlations Between Valence (Vt) (V,) and
ent and Expectancy (En) Expectancy (En)
b^ V, V, V, Ve v>
02 -.34 .29 -.67* — .22 -,52 .90**
03 .00 -.51 -.44 -.38 — .83**
04 .00 .48 -.52 .57 — -,68*
05 — —
07 -.23 — .84** .23
08 .42 .56 ,14 .64 .10 .25
09 .05 -.18 .50 .25 .28 .27
1 f\
10
.40 -.06 .25 -.40 .54 .71*
11
12 .57 .15 -.13 -.13 .03 -.19 .47
13 .00 -.73* -.38 .27 — -.27 .11
14 ,60 .12 -.25 .19 — -.42
15 .38 .30 .00 — -,38 — -.55
16 .89** .89** .93** .99** -.89**
17 .19 .13 -.70* ,71*
18 -.89** .00 .79* .34 -.68* -.58
19 -.42 — .24 -.05 ,47 -.16 .10
20 .76* .99* .64* ,47 -.11 -.64* .86**
21 —.33 .04 -.22 -.02 -.49 -.80** .64
22 .5(, -.27 .21 -.45
23 .27 .40 .78** .27 -.85**
24 .58 -.56 .51 .41 ,24
25 — .47 .14 .84** .73* .14 .17 .61
26 .96** ,62 .67* -.63 .72*
27 .88** .88** .41 ,55 ,35 -.08
28 .85** -.70* .11 -.33 .18
29 .34 .64* .52 -.65** .24
30 -.35 -.02 -.01 -.18 -.30 -.30
31 .85** .62 .37 .37 .58 — — .81**
32 .08 .00 -.50 .44 -.65* -.50 -.44
33 -.54 .25 ,26 -,65* -,67*
34 .68* .26 .34 .00 .02
35 .25 -.37 .25 .03 -.28
36 -.40 .40 — .41 -.40 -.28
37 .44 .72* .63 .06 .38 .69*
38 -.75* -.34 -.29 -,51 — .31
39 .35 -.35 .17 .36 — -.32
40 .32 -.49 .03 .77* .77* .55 -.55
41 -.05 .75* — ,17
42 — — —
43 — — —
44 — — —
45 -.83** -.43 .73* .63 — .27
46 -.51 .39 — -.87**
47 -.17 -.68* -.48 -.48 .68* -.57
48 -.55 -.26 .26 .21 -.26 -.07 -.20
49 -.34 .64* -.58
50 .55 ,14 -.29
52 .04 .73*
53 -.25 .63 -.21 .12 -.29
54 -.07 -.17 ,22 ,22 -.50
55 .71* .06 .87** -.40 — .74*
56 .28 — — — — -.26
57 — — — — — —
*p < .05 **p < .01
1975 Volume 18, Number 4 713
One major difficulty exists when analyzing data relating to the independ-
ence postulate. Specifically, some subjects—Numbers 5, 10, 42, 43, and
44—exhibited no variation over the seven time periods for one or both of the
variables of interest. The blank spaces in the matrices indicate this difficulty,
noting that a correlation coefficient in cases of this type is not conceptually
meaningful. When all cases of this type are removed from the analysis, 200
correlations remain. Of these, 43 are statistically significant. This, in com-
bination with the many coefficients that exceed .40 (but are not statistically
significant), throws considerable doubt on the descriptive accuracy of the
independence postulate as it relates to expectancy theory.
CONCLUSION
The research strategy of testing the postulates underlying expectancy
theory rather than testing the theory itself is, at this writing, in an exploratory
stage. The evidence generated in this study indicates that numerous in-
dividuals do not exhibit behavior patterns that are consistent with two of the
normative postulates of expectancy theory.
With respect to the transitivity postulate, none of the subjects exhibited
effort preference patterns consistent with the basic expectancy model. It is
possible that the significance of this finding is overstated (since the criterion
of agreement was very strict), yet an examination of the data contained in
Table 1 leads to the conclusion that few systematic changes in effort oc-
curred as proposed in expectancy theory. When the difficulty of the calcula-
tions assumed in expectancy theory is taken into consideration, these in-
consistencies are not surprising. It seems highly unlikely that individuals will
routinely spend the time required to make decisions in a systematic fashion.
Rather, they likely will engage in a process which only roughly approximates
an SEU-type analysis, with time and cognitive constraints deterring maxi-
mizing behavior in most routine decisions.
In further research testing of the transitivity postulate, it will be im-
portant to examine the effect of different numbers of second level outcomes
and the impact this has on determining what an "acceptable" level of in-
transitivity is.
Many of the same general observations can be made about the in-
dependence postulate. At this point, the two major questions are: (a)
In a given sample, how many individuals can exhibit a lack of independence
between Vj and Eij before the independence postulate is clearly violated
and (b) How do individual differences mediate the rate of nonindependent
responses? Answers to these questions undoubtedly will be forthcoming as
research progresses in the area of testing postulates.
Although the apparent violation of these necessary postulates of ex-
pectancy theory does not eliminate the possibility that individuals may
arrive at levels of work effort through some combination of the variables
considered in expectancy theory, it does indicate that, if these results are
supported in future studies, the specific combination of them hypothesized
714 Academy of Management Journal December
REFERENCES
1. Adams, J. S. "Toward an Understanding of Inequity," Journal of Abnormal Social
Psychology, Vol. 67 (1963), 422-426.
2. Behling, Orlando, and F. Starke. "The Postulates of Expectancy Theory," Academy of
Management Journal, Vol. 16 (1973), 373-388.
3. Campbell, J. P., M. D. Dunnette, E. E. Lawler, K, E. Weick. Managerial Behavior,
Performance and Effectiveness (New York: McGraw-Hill, 1970).
4. Dachler, H, Peter, and William H. Mobley. "Construct Validation of an Instrumentality-
Expectancy-Task-Goal Model of Work Motivation: Some Theoretical Boundary
Conditions," Journal of Applied Psychology Monograph, Vol. 58 (1973), 397-418,
5. DeLeo, Phillip J., and Robert Pritchard. "An Examination of Some Methodological
Problems in Testing Expectancy-Valence Models with Survey Techniques," Organiza-
tional Behavior and Human Performance. Vol. 12 (1974), 143-148,
6. Graen, G. "Instrumentality Theory of Work Motivation: Some Experimental Results
and Suggested Modifications,'" Journal of Applied Psychology Monograph, Vol. 53
(1969), 1-25.
7. Herzberg, F., B. Mausner, and B. Snyderman. The Motivation to Work (New York:
Wiley, 1959).
8. House, R. "A Path-Goal Contingency Theory of Leadership" (Unpublished paper, 1971).
9. House, Robert J., H. Jack Shapiro, and Mahmoud Wahba, "Expectancy Theory as a
Predictor of Work Behavior and Attitude: A Re-evaluation of Empirical Evidence,"
Decision Sciences, Vol. 5 (1974), 54-77,
10. Lawler, E. E. "A Correlational-Causal Analysis of the Relationship Between Expectancy
Attitudes and Job Performance," Journal of Applied Psychology, Vol, 52 (1968), 462-
468.
11. Lawler, E, E. Pay and Organization Effectiveness: A Psychological View (New York:
McGraw-Hill, 1971).
12. Lawler, E. E., and L, Suttle. "Expectancy Theory and Job Behavior," Oragnizational
Behavior and Human Performance, Vol, 9 (1973), 460-482.
13. Luce, R. D. "Psychological Studies of Risky Decision Making," in G. Strother (Ed.),
Social Science Approaches to Business Behavior (Homewood, 111,: Irwin, 1962), pp, 141-
161.
14. Maslow, A. H. "A Theory of Human Motivation," Psychological Review, Vol. 50 (1943),
370-396,
15. Mitchell, T. "Instrumentality Theories: Conceptual and Methodological Problems"
(Technical Report 71-19, Department of Psychology, University of Washington, 1971).
16. Mitchell, Terrence R. "Expectancy Models of Job Satisfaction, Occupational Preference
and Effort," Psychological Bulletin, Vol, 81 (1974), 1053-1077.
17. Mitchell, T. R., and A, Biglan. "Instrumentality Theories: Current Uses in Psychology,"
Psychological Bulletin, Vol. 76 (1971), 432-454.
18. Porter, L. W., and Lawler, E. E. Managerial Attitudes and Performance (Homewood,
111.: Irwin, 1968).
19. Pritchard, R. D,, and M. S, Sanders. "The Influence of Valence, Instrumentality, and
Expectancy on Effort and Performance," Journal of Applied Psychology, Vol, 57 (1973),
55-60.
20. Savage, L. The Foundations of Statistics (New York: Wiley, 1954),
21. Sheridan, John E,, Max D. Richards, and John Slocum. "The Descriptive Power of
Vroom's Expectance Model of Motivation," Proceedings, Academy of Management,
1973, pp. 414-420.
22. Sheridan, J. E,, J. W. Slocum, and Max D. Richards. "Expectancy Theory as a Lead
Indication of Job Behavior," Decision Sciences, Vol, 5 (1974), 507-522,
23. Vroom, V. Work and Motivation (New York: Wiley, 1964).
24. Wahba, M,, and R, House, "Expectancy Theory in Work and Motivation: Some Logical
and Methodological Issues" (Unpublished paper, 1972).