Sie sind auf Seite 1von 5

Journal of Applied Psychology

VOL. 35, No. 5 OCTOBER, 1951

An Index of Job Satisfaction *


Arthur H. Brayfield
Kansas State College

and
Harold F. Rothe
Hospital Supply Corporation

Increasingly, business and industrial con- As a working approach for this study it was
cerns are studying the job satisfaction and assumed that job satisfaction could be inferred
morale of their employees. It has been recog- from the individual's attitude toward his work.
nized that the effectiveness of selection, train- This approach dictated the methodology-
ing, and supervisory programs should be attitude scaling.
gauged in part by their effect on employees' An attitude scale elicits an expression of
satisfaction with their work and that specific feeling toward an object. It may be used
personnel techniques and procedures should be directly with an individual to obtain such an
validated against a job satisfaction criterion. expression. It permits quantification of the
Unfortunately, adequate indices of job satis- expression of feeling. These characteristics
faction are difficult to obtain. This report suggest the utility of attitude scaling method-
describes the construction and validation of ology in developing an index of job satisfaction.
a quantitative index of job satisfaction which The following requirements were formulated
has been used as a criterion measure in sub- as desirable attributes of an attitude scale
sequent personnel studies to be reported designed to provide a useful index of job
elsewhere. satisfaction:
Construction 1. It should give an index to "over-all" job
A careful survey of the literature reveals satisfaction rather than to specific aspects
that attempts to identify and estimate job of the job situation.
satisfaction have preceded precise definition. 2. It should be applicable to a wide variety
Employee satisfaction and morale are often of jobs.
equated but seldom defined (4). Hull and 3. It should be sensitive to variations in
Kolstad aptly summarize the state of affairs: attitude.
"Although the term 'employee morale' is widely 4. The items should be of such a nature (in-
used, it remains a more or less undefined con- teresting, realistic, and varied) that the
cept whose meaning, usually, is simply taken scale would evoke cooperation from both
for granted. Such definitions as have been management and employees,
offered are of little help to the psychologist in 5. It should yield a reliable index.
the construction of items designed to measure 6. It should yield a valid index.
morale. Thus it was (is) necessary to proceed 7. It should be brief and easily scored.
on the basis of subjective judgment" (3: 350).
At the time this study was undertaken the
* A fuller account may be found in Brayfield's Ph.D. two most widely known and used attitude
dissertation (1946) on file in the University of Minne- scaling techniques were those of Thurstone
sota library. Special thanks are due Professor Donald
G. Paterson under whose advisorship this study was (10) and Likert (5). Initially the choice be-
conducted as one portion of the dissertation. tween them was made on the basis of a practical
307
308 A. H. Brayfield and H. F. Rothe

consideration. Since it would have been ex- the investigators, items were judged as to
tremely difficult to obtain employed persons as specificity. Items referring to specific aspects
subjects for the item analysis required by the of a job were eliminated since an "over-all"
Likert technique, the Thurstone method was attitudinal factor was desired; thus items re-
chosen. The latter method requires a number garding pay, working conditions, etc., were
of judges to sort items. On the basis of other eliminated even though it might be argued that
studies it was assumed that employed persons they reflect a general attitude. Finally,
were not necessarily required as judges but acceptability to employees and management
that almost any mature person could make a as judged by the investigators and manage-
judgment regarding the "value" of a statement ment representatives was a criterion. For
on an attitude continuum regardless of his example, the item "I am tempted to use illness
own job situation (1, 8). as an excuse to stay home from this job" was
The construction of this scale was made a typical of those rejected because they seemed
class project in Personnel Psychology1 for to reflect unfavorably upon the individual.
members of an Army Specialized Training Next a preliminary scale containing eighteen
Program in personnel psychology at the Uni- selected statements was administered to 10
versity of Minnesota in the summer and fall of employed female office workers and a rank
1943. Seventy-seven men cooperated. Their order correlation was computed for the odd
average age was around 30 years. The versus even items paired according to Thur-
majority had had at least several years occupa- stone's directions. The resulting rho was .31
tional experience ranging from unskilled labor which was converted to an estimated product
to professional occupations. The number of moment r and boosted by the Spearman-Brown
judges appears adequate (7, 11). formula to .48.
The class was given instructions similar to This finding indicated a shift in method.
those outlined by Thurstone for the construc- Since Likert had found that his method of
tion of items. Approximately 1,000 state- scoring attitude scales gave a higher reliability
ments were turned in by the class and an than Thurstone's, his scoring technique was
additional 75 by the investigators. This col- adopted and a second experimental scale was
lection was edited and the resulting 246 developed (6).
statements were mimeographed, sorted into As a result of experience with the preliminary
sets, and distributed to the ASTP men for administration, additional comments of man-
judging. Each judge sorted the statements agement, and to replace two "neutral" items
under supervision according to the instructions 9 new items were substituted. The resulting
suggested by Thurstone. After tabulating the blank contained 18 items with Thurstone scale
results the scale and the Q values for each values ranging from 1.2 to 10.0 with approxi-
statement were determined graphically. Care- mately .5 step intervals. The items were not
ful checks were made for accuracy. arranged in order of magnitude of scale values.
Four specific criteria determined the in- The Likert scoring system consisting of five
clusion or exclusion of items for the preliminary categories of agreement-disagreement was ap-
scale. First, it was desired to have items cover- plied to each item. From the Thurstone scale
ing the entire range of the attitude continuum value it was known in what direction to apply
at approximately .5 step intervals although the the new scoring method so that a low total
statements at the ends of the continuum were score would represent the dissatisfied end of
eliminated as being too extreme to be practical. the scale and a high total score the satisfied end.
Second, the Q value, which is based on the The items were selected so that the satisfied end
degree of uniformity in the sorting of state- of the scale was indicated by Strongly Agree
ments, was used as an objective measure of and Agree for one-half the items and by
ambiguity in accordance with Thurstone's Strongly Disagree and Disagree for the other
recommendations. Consequently no item was half. The neutral response was Undecided.
selected which had a Q value of 2.00 or above. The Likert scoring weights for each item ranged
Third, from a purely subjective appraisal by from 1 to 5 and the range of possible total
1
Class time was made available for this project scores now became 18 to 90 with the undecided
hrough the cooperation of Dr. H. P. Longstaff. or neutral point at 54.
An Index of Job Satisfaction 309

JOB QUESTIONNAIRE

Some jobs are more interesting and satisfying than others. We want to know how people feel about different
jobs. This blank contains eighteen statements about jobs. You are to cross out the phrase below each statement
which best describes how you feel about your present job. There are no right or wrong answers. We should
like your honest opinion on each one of the statements. Work out the sample item numbered (0).
0. There are some conditions concerning my job that could be improved.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
1. My job is like a hobby to me.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
2. My job is usually interesting enough to keep me from getting bored.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
3, It seems that my friends are more interested in their jobs.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
4. I consider my job rather unpleasant.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
5. I enjoy my work more than my leisure time.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
6. I am often bored with my job.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
7. I feel fairly well satisfied with my present job.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
8. Most of the time I have to force myself to go to work,
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
9. I am satisfied with my job for the time being.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
10. I feel that my job is no more interesting than others I could get.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
11. I definitely dislike my work.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
12. I feel that I am happier in my work than most other people.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
13. Most days I am enthusiastic about my work.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
14. Each day of work seems like it will never end.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
15. I like rny job better than the average worker does.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
16. My job is pretty uninteresting.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
17. I find real enjoyment in my work.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
18. I am disappointed that I ever took this job.
STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE
FIG. 1. Revised job satisfaction blank.
310 A. H. Bray field and H. F. Rothe

The new scale was administered to 8 addi- be placed at such and such a point along such a
tional employed female office workers and a feeling continuum."
rank order correlation computed for the odd A more rigorous requirement for validation
versus even items. The resulting rho of .61 is to use an outside criterion. An attempt was
was converted to an estimated product-moment made to provide such a criterion. The job
r and boosted by the Spearman-Brown formula satisfaction blank was administered to 91
to .77. This was believed to be satisfactory for adult night school students in classes in
further experimentation and the revised scale Personnel Psychology at the University of
was then printed (see Figure 1). Minnesota2 during 1945 and 1946. Responses
were anonymous. The group included 49
Reliability males and 42 females. The age range was
The revised scale was administered sub- from 22 to 54 with a median of 35 years.
sequently as part of a larger study to 231 Practically the entire membership was engaged
employed female office employees in positions in either clerical, semi-professional and pro-
including entry, typing and stenographic, low fessional, or managerial and supervisory oc-
and high skill level machine clerical, and cupations. The range of job satisfaction scores
accounting jobs. The blanks were signed for this sample was 29-89. The mean score
along with other test materials. One of the was 70.4 with an S.D. of 13.2.
investigators personally administered the tests The common denominator for the members
to employees in small groups. Typically, the of this sample was their enrollment in an
subjects were young, unmarried girls without evening class in Personnel Psychology. En-
dependents. The average girl in the sample rollment in the class was considered to be an
had completed 12 years of schooling. She had overt expression of their interest in personnel
been on her present job for more than one year work. Some indication of the strength of this
and had been employed by the company for interest is afforded by their continued attend-
one and three-fourths years. The range of ance in a night class (after a full day's work) for
job satisfaction scores for this sample was a full semester at the end of which time the
35-87. The mean score was 63.8 with an blank was administered.
S.D of 9.4. Given this circumstance, the following
The odd-even product moment reliability assumption may be made: Those persons in the
coefficient computed for this sample was ,77 class employed in occupations appropriate to
which was corrected by the4 Spearman-Brown their expressed interest should, on the average,
formula to .87. be more satisfied with their jobs than those
Validity members of the class employed in occupations
inappropriate to their expressed interest in
The nature of the individual items is partial, personnel work.
although not crucial, evidence for the validity This assumption seemed reasonable to the
of the scale. This is an appeal to "face" investigators and provided a test of the
validity. validity of the job satisfaction blank.
Additional evidence of a like nature is The 91 persons accordingly were divided
furnished by the method of constructing the into two groups (Personnel and Non-Personnel)
scale. In developing the scale an attitude with respect to their employment in a position
variable was specified. This was job satisfac- identified by pay-roll title as a personnel
tion which was to be inferred from verbal function. Four occupationally unidentified
reactions to a job expressed along a favorable- persons were placed arbitrarily in the Non-
unfavorable continuum. The statements used Personnel group. In all, 40 persons comprised
in this scale uniformly had small Q values the Personnel group; 51 the Non-Personnel. A
which indicates a marked consistency among comparison was then made between the mean
the judges. With the attitude variable speci- scores for the two groups on the job satisfaction
fied, 77 adult judges were consistent in saying, blank. The mean for the Personnel group was
"This statement expresses a feeling of satis- 2
These subjects were made available through the
faction or dissatisfaction with a job and should cooperation of Dr. H. P. Longstaff.
An Index of Job Satisfaction 311

76.9 with an S.D. of 8.6 as compared to a mean nel group was 22.2 with an S.D. of 2.6; the
of 65.4 with an S.D. of 14.02 for the Non- mean for the Non-Personnel group was 19.2
Personnel group. This difference of 11.5 with'an S.D. of 4.0. The difference between
points is significant at the 1% level; the the means is significant at the 1% level; the
difference between the variances also is sig- variances do not differ significantly.
nificant at the 1% level. Since Fisher and
Behrens' d-test is appropriate when significant Summary
differences are found both between two means
and their respective variances, it was applied An attitude scale which purports to be an
and the difference between the means was index of job satisfaction was constructed by a
found to be significant at the 5% level (9). combination of Thurstone and Likert scaling
Sukhatme's tables give only the 5% level methods. A reliability coefficient of .87 was
values. obtained for one large group. Evidence for
If the original assumption as to the differ- the high validity of the blank rests upon the
ential significance of membership in one or the nature of the items, the method of construction,
other of the two groups is accepted, these data and its differentiating power when applied to
furnish evidence for the validity of the job two groups which could reasonably be assumed
satisfaction blank. to differ in job satisfaction. Scores on this
blank were highly correlated with scores on
Comparison with the Hoppock Blank the Hoppock blank in the sample studied.
Perhaps the most systematic attempt to Received November 10,1950.
develop an index of job satisfaction was the
one made by Hoppock in the early 1930's (2). References
Originally Hoppock tried out a series of simple
attitude scales as part of an interviewing study 1. Hinckley, E. B. The influence of individual opin-
ion on construction of an attitude scale. /. soc.
of 40 employed adults. These scales were PsycM., 1932, 3, 283-296,
revised to consist of four items each with seven 2. Hoppock, R. Job satisfaction. New York: Harper
responses at step intervals. Values of 1 to 7 and Brothers, 1935.
were assigned arbitrarily to the responses in 3. Hull, R. L., and Kolstad, A. Morale on the job.
In G. Watson (Ed.), Cwttian Morale. New
each item, the smaller numbers being assigned York: Reynal and Hitchcock, 1942.
to the responses indicating dissatisfaction. 4. Kornhauser, A. Psychological studies of employee
The range of possible total scores was 4 to 28. morale. /. consult. PsycM., 1944, 8, 127-143.
This system of scoring correlated .997 for 301 5. Likert, R. A technique for the measurement of atti-
cases with a system of scale values assigned on tudes, Archives of Psychology, No. 140. New
York: Columbia University, 1932.
the basis of z-scores. The corrected split-half 6. Likert, R., Roslow, S., and Murphy, G. A simple
reliability coefficient for the scale for the same and reliable method of scoring the Thurstone
301 cases was reported to be .93. The scale scales. /. soc. Psychol., 1934, 5, 228-238.
evidently has been assumed to have "face" 7. Nystrom, G. H. The measurement of Filipino
attitudes toward America by use of the Thur-
validity. s.tone technique. /. soc. Psychol., 1933, 4, 249-
The night school classes completed both 252.
blanks. The product-moment correlation be- 8. Pintner, R., and Forlano, G. The influence of
tween scores on the Hoppock blank (Form 11) attitude upon scaling of attitude items. /. soc.
and on the Brayfield-Rothe blank was .92. Psychol., 1937, 8, 39-45.
9. Sukhatme, P. V. On Fisher and Behren's test of
Although the two blanks were developed by significance for the difference in means of two
different methods and contain items which normal samples. Sankhya: the Indian Journal
over-lap only slightly they give results which of Statistics, 1938, 4, 39-48.
are highly correlated. 10. Thurstone, L. L., and Chave, E. J. The measure-
ment of attitude. Chicago: University of Chicago
A comparison was made between the means Press, 1929.
of the Personnel and Non-Personnel groups on 11. Uhrbrock, R. S. Attitudes of 4430 employees.
the Hoppock blank. The mean for the Person- /. soc. Psychol, 1934, 5, 368-377.

Das könnte Ihnen auch gefallen