Professional Documents
Culture Documents
Values, Status, and Professors
Values, Status, and Professors
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Sociological Association is collaborating with JSTOR to digitize, preserve and extend
access to Sociometry
Washington University
* This report was supported in part by the Office of Naval Research, Contract Nonr
816 (11). The data were analyzed through support provided by the Washington Uni-
versity Computer Center under NSF grant G-22296.
1 P. M. Blau, Exchange and Power in Social Life, New York: Wiley, 1964.
2 G. C. Homans, Social Behavior: Its Elementary Forms, New York: Harcourt, Brace,
and World, 1961. Actually, we use the concept status differently than does Homans. He
uses status much the way others have used the concept status symbols. We define status
as the expression of feelings of approval, respect, or esteem. The latter terms are used
by both Homans and Blau in describing status exchanges. Actually, a concept such as
status can be defined in many useful ways, but the present conceptualization has the
advantage of being close to common usage.
183
visor) and various indicators of status, for example, contacts received.8 Some
investigators would have been satisfied with the correlation as it was, but
Blau's analysis of two of the three deviant cases established the rule. Two
very competent agents were given low status. Why? Evidently because they
refused to help the less competent agents who needed their assistance badly.
In other words, status was apparently exchanged for competent help; if an
agent did not give competent help because he was unable or unwilling he
did not receive high status, as gauged by the several indicators.
While the theory does appear to account for the data, particularly in the
instances cited by Homans and Blau, this does not mean that their status
exchange theory is demonstrated to be true generally, or that it cannot be
improved or refined. It would be a great disservice to Homans and Blau to
enshrine or to ignore the theory in its formative stages.
In further analyzing status exchange theory, a number of questions could
be considered, but an important one involves the action and feeling dimen-
sions of status. This distinction, which may be crucial, has been blurred
except in Blau's discussion of simulated status where he observes that the
giving of status via acts of approval, praise, or deference may not be gen-
uine.5 When status is simulated, the feelings of the giver do not coincide with
his actions-for example, he might praise a colleague for a lecture, yet pri-
vately feel it to be a waste of time and effort. Such simulated status-giving is
typical of etiquette and ordinarily it is recognized as an attempt to get along,
to feign reciprocation without being rude. Yet simulated status-giving is not
to be confused with actual status-giving which involves genuine feelings of
approval, respect, or esteem and an accurate communication of those feelings.
Genuine status-giving is problematic because having feelings of approval,
respect, or esteem for someone appears to be beyond the individual's direct
choice. In fact, the evidence suggests that all feelings, including those of
approval, respect, or esteem, are part of a class of non-voluntary responses.
As with all non-voluntary responses, these feelings are presumably controlled
by the unconditioned or the conditioned stimuli which elicit them. Appar-
ently, an individual must provide the valued attributes and behavior which
produce in the other the feelings of approval, respect, or esteem; then and
only then may these feelings be communicated as genuine status.
There are many other classes of non-voluntary responses besides emotions
or feelings, however, a few of which are the activities of many internal organs
such as the beat of the heart, the various reflexes, and interestingly enough,
the sensations produced in sensory receptors by physical stimuli. The latter
are an important class because we know so much about their relationship to
the stimuli which produce them. The magnitude of a psychological sensation
4I' always increases as a power function of the magnitude of the physical
stimulus b, as in Stevens' law: B-con, or its logarithmic equivalent: log
*-log c+n log 0, where c and n are parameters which vary from one
physical stimulus to another.6
Although limited, the evidence thus far suggests the power function may
turn out to be the general form of the relationship between the magnitude
of all non-voluntary responses R and the magnitude of the stimuli S which
produce them as in the following equations:
R=c Sn
or
log R=log c+n log S
or esteem should vary from group to group or from time to time as interests
and values change.
(2) The magnitude of an individual's estimated status in a group, since
it is based on feelings, should increase as a power function of the magnitude
of the related stimulus or stimuli-the individual attributes or behavior that
are valued in that group.
Although the first prediction may not be obvious, we will not investigate
it here. However, we do want to determine whether or not the relationship
between estimated status and the related stimuli can be accurately described
by a power function. To answer this question successfully, it is necessary
first to isolate accurately the various stimuli which produce status in an
ongoing social system.
For convenience and inherent interest, we decided to study part of the
status-exchange between professors and graduate students in a small aca-
demic department. Specifically, we decided to try to isolate the characteristics
of professors which earn them status among graduate students-or to put it
differently, to find out what behaviors and attributes of professors are valued
and thus might earn the approval, respect, or esteem of the graduate students
in the department in question.
Many behaviors and attributes could be valued and perhaps are valued
in one department or another. However, the pervasive notion now is that
academic status increases as (1) the merit of a professor's publications in-
creases, and (2) the merit of his teaching increases. (Although it has been
popular in some circles to say that good teaching is negatively associated
with academic status, this should not be true for the graduate student group.)
Academic status may also increase with (3) professional age. Further, Ellis
has assumed that status increases with personal attributes such as (4) pro-
fessorial appearance, (5) cordiality, and (6) popularity (being liked as a
person).8 Some would argue that an increase in formal (7) professorial rank
increases the status of an individual beyond what might be expected as a
function of merit of publication, etc. Finally, we assumed that a professor's
status might be related to the degree to which the evaluator knew the pro-
fessor and his work.9 This last may not be a value, but perhaps a condition
that should be held constant. Although thus far we have talked about status
as a unitary concept, several theorists 10 have pointed to status of two types:
status in the local department, and status in the wider profession and dis-
cipline. We have not been able to find quantitative data indicating that these
two types of status are earned in a different way, although it usually is as-
sumed that they are. In the present investigation, however, we have no spe-
cific hypotheses as to differences. We simply assumed that local and pro-
fessional status would be different, and that each could be determined by any
combination of the above variables (and perhaps others not listed above
and thus not operationalized in the present investigation).
One final theoretical problem: the stimulus-response power hypothesis is
relevant only to the bivariate case where one variable is the response, and
the other the stimulus which produces variations in the response. Yet the
evidence we have suggests that the status response is a multivariate phe-
nomenon, that a number of attributes and behaviors may be valued and thus
may become determinants of status in a given social system. If a multivariate
power function describes the status relationship, what form might the general
equation take? The simplest is as follows:
This is, of course, the multivariate extension of the stimulus response law
as may be noted by its logarithmic transformation:
Social Theory and Social Structure, Glencoe: Free Press, 1957; and A. W. Gouldner,
"Cosmopolitans and Locals: Toward an Analysis of Latent Social Roles," Administra-
tive Science Quarterly, 2 (December, 1957-58), pp. 281-306; 444-480.
uate students. We assume that they are, with the one exception noted above
(that is, the degree to which the evaluators knew the professor and his work).
(3) However, if both assumptions 1 and 2 hold, and if the feelings of
approval, respect, or esteem are, in fact, non-voluntary responses which may
be conditioned, then:
(4) We predict that the magnitude of estimated status, be it local or
professional, will be a multivariate power function of the magnitudes of a
subset of our independent variables as specified above. Furthermore, if we
have been successful in accurately isolating the independent variables, we
expect to account for circa 98 per cent of the variance, as is typically done
with non-voluntary behavior.
(5) We consider this last a crucial aspect of status exchange theory since
it involves the detailed relationship of output to input.
METHOD
All of the twelve graduate students who were used as observers had been in
residence in the chosen department more than one and one-half years. Since
the department was medium to smallish in size (thirteen faculty members
counting the marginal members), a year and a half gave the students some
opportunity to become acquainted with the various professors as well as with
the status system.
To see if the results were stable, we decided to do the study twice, in
January and March, 1965. In January, our observers estimated the magnitude
of the status variables and the various possible determinants of status for
eight professors, and in March for all thirteen. With the exception that our
measure of merit of publication was changed in the second study, the essen-
tials of the procedure were the same.
In all cases the measurements began with teaching the observers how to
use lines to estimate magnitudes:
First, note this line which represents ten units. Now draw a line twenty units long;
three and one-third units long. . . . (A half-dozen stimuli were given in a different ran-
dom order for each observer. The observer continued to estimate their length.)
Once the interviewer was satisfied that the observer was drawing lines in
approximately correct proportion to numbers, he proceeded with the following
instructions:
Now notice what you have done. You have drawn all your subsequent lines propor-
tional to your first line. In other words, you have drawn your 20-unit line to be twice
as long as your 10-unit line, and so on. We will use this principle in subsequent estima-
tions.
You probably have noted that some of the professors in this department are looked
up to and some are looked down on-they have a different amount of prestige, approval,
respect, esteem, or status. I would like you to indicate the local status of each, here in
the department, using lines as you did before. Start with Professor ...... ..... This line
will represent his status. (He was a professor of medium status and the line was ten
centimeters long.) If another professor has three times his status in your estimation,
draw a line three times as long; a fourth as much status, draw a line one-fourth as long.
In other words, draw a line (or lines) for each professor so that the total length is pro-
portional to his local status.
Although the same professor was used to start each series of estimations of
local status, the others were presented in a different random order for each
observer. For the other dimensions also, the professors were presented to each
observer in a different random order so as to cancel out any order-of-presenta-
tion effect.
After measuring local status, the observer was asked to estimate the teach-
ing ability of each of the professors, their professional status, professorial
appearance, cordiality, and the merit of their publications. Continuing, each
observer was asked to estimate the extent to which he liked each of the
professors and also the amount of personal knowledge he had of each pro-
fessor and his work. The observers were instructed that if, in their estimation,
a professor had no local status, no teaching ability, etc., they were to indicate
it by a zero instead of a line.
In the first investigation, the measurement for the merit of publications
was obtained indirectly; i.e., the observers were asked to estimate the status
accruing from different numbers of publications. The numbers presented as
stimuli corresponded to the number of publications of each professor. In this,
apparently we made the same error that some deans make. However, all of
our observers except one (who said, "It's the quality, not the number, that
counts.") were willing to make such judgments.
Once the measurements were taken, the median estimate for each variable
was calculated for each professor. Assuming that these medians were the
best collective representation for the status phenomena under investigation,
they and not the individual scores were analyzed.
RESULTS
To test our hypotheses, the medians for each of the ten variables for each
professor were transformed to logarithms and analyzed using linear multiple
regression correlation analyses which condensed the equations to statistically
significant relationships." Thus we tested the logarithmic form of our multi-
variate hypothesis for local status, and then for professional status.
11 Some might criticize the present investigation because only eleven observers were
used. Actually, in social science, a large N is ordinarily employed to improve the accuracy
and the statistical significance of the results. Note that in Tables 1 and 2 the statistical
TABLE 1
The Multiple Regression Results for Local Status
Study I Study II
Regression t Regression t
Variables Coefficient Value Coefficient Value
NOTE: Only the independent variables whose beta weights are significantly different
from zero are shown; the others were dropped by the condensation procedure. R2 was
.97 for both investigations. A t- value of 2.57 is significant at the .05 level for 5 degrees of
freedom, two-tailed test. Without the logarithmic transformations, R2 was .87 and .84,
respectively, for the first and second investigations.
LOCAL STATUS. The multiple regression results for local status are given
in Table 1. Note that the results are similar for both investigations. Local
status of professors given by the graduate students is apparently a power
function of teaching ability and of professorial appearance. The regression
coefficients for all of the other variables were not significantly different from
zero. This includes merit of publication (measured either way), personal
liking, and knowledge of the professors. Furthermore, the R2 is .97 in both
investigations, indicating that almost all the variation in local status is
accounted for by the variations in teaching ability and professorial appear-
ance. Finally, note that the regression coefficients, the exponents in the multi-
variate power function, vary some from the first to second investigation. Ex-
ponents will so vary because of measurement error, but probably not this
much. Increasing the professors from eight in the first investigation to thirteen
in the second may account for the major portion of the differences. If so, then
the second set of results are more accurate; apparently then, the power
exponents for teaching ability and professorial appearance are circa .84 and
.54, respectively.
Professorial appearance was one of the variables suggested to us by Ellis,12
significance indicated by the t-values is quite high (beyond the .01 level, in most cases).
Thus, with the present rather accurate measurement procedures, a small N such as used
here may be quite adequate. Ordinarily however, we prefer, perhaps illogically, to use
20 or better 30 observers in investigations such as these. This is one reason why we
replicated the investigation.
12 Ellis, op. cit. Some people have reacted negatively to stimulus-response theory be-
cause it seems to paint an automatic, unthinking picture of social behavior. This is
because they learned from Watson not Mead. Mead, a behaviorist and a thorough-going
stimulus-response theorist, waged a polemic against Watson on this very point. He pointed
out time and again that the social act has many parts, and that it may be drawn out over
a long period in some cases. Via an internal conversation of gestures (stimuli) the indi-
vidual sometimes uses self to test out others' responses until he is confident he knows what
TABLE 2
Study I Study II
Regression t Regression t
Variables Coefficient Value Coefficient Value
NOTE: Only the independent variables whose beta weights are significantly different
from zero are shown; the others were dropped by the condensation procedure. A t-value
of 2.31 is significant at the 5 per cent level, two-tailed test, for 8 degrees of freedom.
These results illustrate the fact that the t-test is valid just for the set of variables in-
cluded in the analysis. In the first study, where a valid measure of merit of publications
was missing, one collection of "significant" variables obtains. In the second investigation,
with merit of publications measured appropriately, two of the previously significant vari-
ables dropped out.
gestures or symbols (stimuli) will elicit the desired response from others. This is what we
suggest happens with status. The feelings are elicited rather automatically by the behavior
of others. However, the individual may choose to express these feelings, to cover them,
or to ignore them, depending upon desires, the situation, etc. Actually, the conditioned
stimulus-response theory of feelings and emotions is the only theory developed to date,
unless one wants to move to a serious consideration of instincts. Cf. G. H. Mead, On
Social Psychology, A. Strauss (ed.), Chicago: University of Chicago Press, 1964.
publication. Such an error can be fatal in a multivariate test. since the rela-
tionships that emerge in a multiple regression analysis can be distorted by
the absence of one or more significant variables. Thus we choose to ignore
the results of the first investigation, even though there is some overlap with
the results of the second, since the measurement of merit of publication was
so poor that in effect it was not measured at all. We will interpret just the
results of the second investigation.
Professional status here turned out to be a multivariate power function
of four variables: merit of publication and teaching ability, and less impor-
tantly, cordiality (which is negatively related), and professional age. R2 is
.99 which suggests that practically all of the variance in professional status
is explained by these four variables.
DISCUSSION
We did not set out to prove that graduate students in the department
in question actually exchange status for valued attributes and behavior of
professors; rather, we assumed that they did. Are the results consistent with
that assumption?
In the local status context, status was significantly and independently
associated with meritorious teaching and with what was interpreted as pro-
fessorial demeanor. Most skeptics would concede that after sitting through
17 to 19 years of classes, a student would be conditioned to value and thus
approve, respect, or esteem good teaching, but what about professorial de-
meanor? Is that something students would value? In the modern university
setting, professors often take other roles, hence their demeanor is often more
that of a research entrepreneur, a research administrator, an author, a high-
priced public lecturer, a clinician, and so on. It seems not unlikely that
graduate students would approve of a professor to the extent he took his
professorial role seriously outside of the classroom and thus spent time and
effort advising them, answering questions, giving suggestions and criticisms
to dissertation proposals, etc. Classroom teaching is only part of a graduate
professor's role; and it seems likely that students would value and thus ap-
prove of a good job outside of the classroom.
In the wider professional context, status was significantly and independ-
ently associated with (1) merit of publications, (2) merit of teaching, (3)
length of experience or service in the discipline, and (4) negative cordiality
-that is, maintaining an appropriate social distance. In this context, the
students could be reflecting the values of the profession. Certainly, who
would argue that publications of merit are not valued in academia, or that
meritorious teaching is not valued? Clear, insightful, exciting presentations
are essential to the first rate academic. Perhaps the value of good teaching
is over-estimated by graduate students but then students are often less cynical
than older heads. What bright aspiring young professor has not chafed against
the value of length of service? Other things equal, even the gifted are usually
valued more over time as their experience accumulates. Negative cordiality
may not be a value in the usual sense, but it may be an unconscious value,
for professional status has an element of heroism, and people generally react
as if they value a hero more at a distance.
Thus, the results appear to be consistent with the assumption that attri-
butes and behavior of the professors which evidently earn them status from
graduate students are valued by the graduate students.
However, the main focus here is on the precise nature of the output-input
relationships involved in the assumed status exchange. On the assumption
that status feelings are non-voluntary responses, we predicted that status
(both local and professional) would be a multivariate power function of a
subset of the independent variables which were assumed to be values.'3 The
prediction was confirmed. R2 was circa .98 which was expected if we were
fortunate enough to include all of the significant independent variables. This
is the important finding of the present investigation which any alternative
theory of the results must account for in detail.'4
The linear logarithmic version of the stimulus-response hypothesis was
used in the regression analysis because it fits the linear assumptions involved.
However, this is a technical detail. The theoretical meaning of the results
are given by the algebraic version. Let us explore these meanings in a brief
way.
Consider the simplified multivariate equation Y=X1 X2 (which by con-
vention is the simplified version of Y=1.0 Xj11 .X210), and consider i
linear cousin Y=X1 +X2. In the linear equation, the effects of X1 and X2
are literally independent. If X1 is increased 2 units, Y will increase 2 units
regardless of the value of X2. However, in the multivariate power function,
the effects of X1 and X2 are quite interdependent. If X1 and X2 are both 1,
then Y is 1, and increasing X1 to 5 will increase Y from 1 to 5. However, if
X1 is 1 and X2 is 100, then Y= 100 and increasing X1 from 1 to 5 will
increase Y from 100 to 500. Thus a 4-unit increase of X1 will increase Y by
4 or 400 units depending upon the value of X2. This is known as the multiplier
effect.
Now let us submit that knowledge of such a multiplier effect is crucial for
an adequate theory of status. Translate X1 and X2 into merit of teaching and
professorial demeanor, respectively, and Y into local status. Our results say
that the local status of a professor in the department in question is a multi-
plier effect of merit teaching and professorial demeanor. Would one really
understand the dynamics of local status among graduate students in the
department in question without knowing about such a multiplier effect? Of
course, this case is not as simple as in the above example-where X1 and X2
were equally weighted by 1.0 exponents. Merit of teaching is more important
by a factor of about 1.5 as indicated by the respective exponents. However,
this again is a detail. Can one theoretically understand multivariate phe-
nomena without a rather precise idea of the nature of the relationships in-
volved? Is it enough to assume implicitly that a multivariate relationship is
additive, as it is often done in modern sociological theories? Would not
the essence of what happens be lost?
What we are suggesting, is that to develop an adequate theory, a scientist
must solve two theoretical problems: (1) He must accurately isolate and con-
ceptualize the independent variables which produce variations in the de-
pendent variable in question, and (2) he must correctly isolate and analyze
the nature of the relationships among the variables in question. Many present
social theorists appear to be fixated on the first question and almost oblivious
to the second. Yet while isolating and conceptualizing variables is important,
it is not more so than understanding the relationships among significant vari-
ables. Failure to answer either point adequately will result in a weak theory,
or, more technically, in unexplained variance. Without correctly isolating
and conceptualizing the crucial variables, the scientist will not know what to
15 Homans, op. cit. See also, G. C. Homans, "Bringing Men Back In," American
Sociological Review, 29 (December, 1964), pp. 809-818.