|
Funding for Mass Communication Research:
Trends and Causes from 1954 to 1993
Jian-Hua Zhu
Department of Communication Sciences
The University of Connecticut
Storrs, CT 06269
Mark Swiencicki
Department of Sociology
The University of Connecticut
Storrs, CT 06269
Submitted to the 1995 Annual Conference of the Association
for Education in Journalism and Mass Communication
Division of Communication Theory and Methodology
April 1, 1995
____________________________________
The authors would like to thank Day Boswell, Jacquie Cartwright-Mills,
Manoj Fenelon, Mats Georgson, Tatyana Gurikova, David Horrigan,
Brett
Monroe, Krishnan Subramanian and Xiaolan Sun for assistance in data
collection, and David Weaver for suggestions and comments.
Abstract
In a content analysis of studies published in two journals in mass
communication and
three journals in psychology, sociology and political science
between 1954 and 1978,
Weaver and Gray found that mass communication research was
underfunded as compared to the
three neighboring disciplines. This study replicates and
extends the Weaver and Gray
study by including 12 journals in 1983-93 and adding several
explanatory factors. The
results show that funding for mass communication has been on the
decline and that the gap
between mass communication and the adjacent social sciences
widening. Possible causes of
the trend include the reduction of funding from governmental
and private sources. Within
the mass communication field, quantitative studies with a focus
on electronic media along
or in conjunction with print media are more likely to be
funded. Two structural factors,
the author's academic rank and institutional affiliation,
appear to have only limited
effects on the fundability.
Funding for Mass Communication Research:
Trends and Causes from 1954 to 1993
It would almost be trite to point out that funding is of central
importance to mass
communication research. Everything else being equal, more funding
produces a better
quality and a higher quantity of research. Funding is also vital
to the individual's
obtaining tenure and promotion, and to the department's quest for
status and a greater
share of university resources. However, very few studies have
investigated this important
issue which almost all mass communication scholars must face. One
exception is Weaver
and Gray's (1980) study which found that, between 1954 and
1978, mass communication
research was under-funded as compared to several adjacent social
sciences. Our study
updates and extends Weaver and Gray's study.
Funding in 1954-78
Based on a content analysis of Journalism Quarterly (JQ) and Public
Opinion Quarterly (
POQ),[1] Weaver and Gray (1980) found that one-fourth (26%) of the mass
communication studies
published in the two journals between 1954-78 received funding from either
an intramural
or extramural source. This rate of funding was quite low as
compared to three neighboring
social science disciplines: political science, sociology and psychology.
Weaver and
Gray examined the American Political Science Review (APSR), the
American Sociological
Review (ASR) and the Journal of Personality and Social Psychology
(JPSP) as the
representative of each respective discipline during the same period,
and found that the
average funding rate (55%)[2] for these three disciplines more
than doubled that of mass
communication research.
Weaver and Gray also found that funding for mass communication
research came quite evenly
from the three major sources: 7% from the university, 8% from the
government, and 9%
from private foundations. While they believed that the balance
of funding sources
minimized the reliance on any particular source and was therefore
healthy, the university
and the government appeared to be less willing to support mass
communication research than
the other social sciences. For example, 14% of the studies in political
science,
sociology and psychology received funding from the university, or
twice as much as mass
communication research funded by the university. Moreover, the
proportion of studies in
these three disciplines funded by the government (31%) was
about four times higher than
that in mass communication research. Only in funding from the
private sector did mass
communication research exceed its neighboring disciplines (9%
vs. 4%).
As Weaver and Gray reported, those mass communication research studies
employing a
quantitative methodology were more likely to be funded than those
using a non-quantitative
methodology. For example, 32% of the quantitative studies were funded
whereas only 11%
of the non-quantitative studies received any funding. In
addition, certain types of data
collection (e.g., survey and experimental studies) had a higher
funding rate than other
types of data collection (e.g., content analysis and
historical-philosophical analysis).
Structural Factors
While Weaver and Gray's study provides a baseline to assess the status
of funding in mass
communication research, we are also interested in the impact of certain
structural
factors - such as the characteristics of the individual authors and
their institutions -
on the funding rate in mass communication research. We call
these variables "structural
factors" because, unlike the methodology used or the topic
investigated, the job rank and
prestige of an author's institution are extraneous to the study
being considered for
funding. Therefore, these factors would represent some form of
social control over
academic research if proven to be related to the rate of funding.
In general, few empirical mass communication and other social science
studies have
investigated these structural factors, but two are notable. One
was carried out by
political sociologist Michael Useem in 1976. Interested in
investigating whether federal
grant agencies favored research which either legitimated the
state or provided it with
directly consumable policy relevant research, Useem surveyed
1079 social scientists from
four fields (including anthropology, economics, political
science and psychology) who had
submitted funding proposals. The study tried to predict
funding from the following seven
author-relevant variables: citation rate, policy relevance,
political perspective,
government confidence, departmental professional status,
individual's professional status,
and publication rate. While Useem found little to no relationship between
the presence
of federal funds and the authors' political perspective or
government confidence, he did
find an approximately linear relationship between the other
five variables and the
presence of funding. Nevertheless, only citation rate and policy
relevance produced
consistently strong coefficients with the presence of funding
(average beta's were .22 and
.23 respectively). However, the measure of policy relevance is not
reliable since Useem
relied on the self-report of the original authors to decide
whether their research was
relevant or not.[3] Thus, the more objective variables of
departmental and individual status
each explained an average of only about 7-8% of the variance, while
publication rate had
a strong beta in only one of the four fields (economics). In
sum, then, the three most
reliable predictors of funding in Useem's study were (in
descending order of magnitude)
citation rate, departmental status, and individual professional
status.
Another study relevant to our purpose is by Cole, Rubin and Cole
(1978) who examined the
characteristics of 1,200 National Science Foundation (NSF)
grant applicants. Although the
study used a self-selected sample, and NSF applicants generally have
better track records
than the average academicians, their results shed some light on who is
allotted NSF aid.
Since Cole et al. broke their sample into 10 separate fields (120 apiece)
it is possible
to analyze the figures for social science alone (including
anthropology and economics).
According to their 1978 report, the seven author-specific
variables demonstrated the
following degrees of explanatory power: 11% of the variance in
the NSF's funding decision
was explained by the number of citation in the previous 10 years, 9% by
the rank of
current department, 8% by the rank of Ph.D. department, 2% by the
author's age and the
type of current institution (Ph.D. or not), 1% by the academic
rank, and 0% by the number
of publications.[4] Based on the results, the authors
concluded that NSF grant giving (in
1977) was not an "old boy network" favoring scholars from the
most privileged schools, and
was only moderately determined by the prestige and characteristics of the
author. At the
NSF level, at least, reviewers evidently gave "much greater weight to the
merits of the
research proposed than to the applicant's previous scientific
achievement" (Cole, et al.,
ix).
While the above two studies are insightful, especially since both
found citation rate and
institution rank to be the two most important determinants of fundability,
their samples
are far too narrow to generalize to the overall social science
research community. In
particular, Useem's 1976 study excluded non-federal funded
studies such as state or local
government, the university, and private foundations, which
jointly outweighed federal
funding in both mass communication and other social sciences.
While Cole et al.'s 1978
study was considerably more extensive than Useem's in terms of
the number of independent
variables examined, their sample was confined to only those who
received NSF funding.
Surprisingly, then, no follow-up studies along the lines of
Useem or Cole et al. have
appeared (to our knowledge) since the late 1970s. Equally
important, no one has examined
the impact of these structural factors on funding for mass
communication research.
Therefore, the present study not only updates Weaver and Gray
(1980) but also extends
Useem (1976) and Cole et al. (1978) to mass communication
research.
Research Questions
Based on the literature reviewed, this study will focus on the
following six research
questions:
1. What is the current funding rate for mass communication research, as
compared to what
Weaver and Gray found in 15 years ago?
2. Has the gap in funding between mass communication and the three adjacent
social
sciences been closed over the last 15 years?
3. How much change has there been in the make-up of funding sources for
mass communication
research?
4. Within mass communication research, what kind of studies have been more
likely to
receive funding?
5. Does the author's academic rank affect funding chances in mass
communication research?
6. Does the author's institutional reputation affect funding likelihood in
mass
communication research?
Methods
Sampling
Initially, we precisely followed Weaver and Gray's sampling procedure,
using JQ and POQ
to represent mass communication and APSR, ASR and JPSP to
present the three adjacent
social sciences. Weaver and Gray examined one issue of each of
these journals each fifth
year. To obtain more stable statistics, we included all issues
of these journals in the
given years (1983, 1988, and 1993, respectively). Also
following Weaver and Gray, we
examined only major research articles published in these
journals, excluding research
briefs, commentaries, presidential addresses, and book reviews.
For POQ, only major
research articles dealing with a mass communication topic were
included.
A new question arises concerning the extent to which JQ and POQ
represents mass
communication research today as opposed to their ability to do so
30-50 years ago. To
minimize selection bias, we extended the sampling frame to four
additional journals:
Journal of Communication (JOC), Journal of Broadcasting and
Electronic Media (JOBEM),
Critical Studies in Mass Communication (CSMC), and Communication
Research (CR). The first
three journals in this expanded list are the official publications on mass
communication
by three major professional societies (for International
Communication Association,
Association for Broadcasting Educators, and Speech Communication
Association,
respectively), and the last one, CR, is probably the most influential
journal among the
non-affiliated publications on mass communication.[5]
As stated above, we selected every issue of these four additional
journals in 1983, 1988,
and 1993.[6] Like POQ, two of the additional journals, JOC and CR, are
not exclusively
devoted to mass communication. Therefore, we included only those
articles dealing with
mass communication in JOC and CR. As it turned out, JOC and CR
carried a lot more mass
communication articles than POQ did during 1983-93. Mass
communication research accounts
for 79% of the articles in JOC and 65% in CR, as compared to
only 11% in POQ.
With the above modifications of Weaver and Gray's procedure, we are
more confident of the
quality of our sample, both in terms of its size and its
representativeness. Our final
sample includes 539 articles from the six mass communication
journals,[7] and 457 from the
neighboring disciplines between 1983 and 1993. The mass
communication subsample
represents official publications of five major communication
associations (AAPOR, AEJMC,
BEA, ICA, and SCA).
Measurement
Similar to the sampling procedure, our measurement follows closely,
whenever necessary,
Weaver and Gray's coding scheme to ensure comparability over
time. Every article in the
sampled journals was coded in terms of its funding status
(funded or not), and funding
source (university, government, private foundation, or some
combination). Both funding
status and funding source were coded based upon the
acknowledgment made by the author(s)
at the beginning or the end of the article. For mass
communication articles, additional
information was collected on the methodology used (quantitative
vs. non-quantitative),[8] the
type of data collection (e.g., survey, experiment, content analysis, and
historical-philosophical method),[9] and the type of medium
under study (print, electronic,
mix of print and electronic media, and non-media).[10]
To test the impact of structural variables on funding status, as
discussed earlier, we
added two variables to the content analysis that were not used
by Weaver and Gray but
examined by Useem (1976) and Cole et al. (1978). The first is
the academic rank of the
author (or first author under multiple authorship),[11] which
includes full professor,
associate professor, assistant professor, lecturer/instructor,
graduate student, and
non-academician (e.g., government employee or industry analyst).
If an administrative
title (dean, chair, or director) is used along or in conjunction
with an academic rank, we
coded the person as an administrator, distinguished from any academic
ranks, based on the
assumption that an administrative position might enhance funding chances.
Another addition of ours is the institutional affiliation of the
author/first author (or
the primary recipient of the funding if different from the
author/first author). As
suggested by the literature reviewed above, the reputation of the
institution may carry
some weight in determining the likelihood of funding. In
operationalizing the
institutional prestige, however, we had to choose between coding the
rank of the
university or the rank of the program (department, school, or college
within the
university). These two ranks are not necessarily consistent. For
example, a significant
number of top-notch universities had only an M.A. or even a
B.A. program in
journalism/communication, while a Ph.D. program did exist in
several second-tier
universities for many years. We decided to measure the rank of the
university, rather
than that of the program, assuming that: (1) more prestigious
universities carry more
weight at the extramural funding agency, and (2) better
universities presumably have more
intramural funding for mass communication scholars. The
university rank is based upon the
1987-94 Carnegie Classifications of post-secondary academic
institutions.[12]
Several problems arose in measuring mass communication scholars'
citation rate, a
variable that both Useem and Cole et al. found to be an important
predictor of federal
funding in other disciplines. First, a number of mass
communication journals have not
been referenced by the Social Science Citation Index (SSCI),
which would result in a
biased sample. Second, the SSCI does not spell out the author's
first and middle name so
that it would take countless hours to verify the authorship
involved. Finally, as a
number of researchers have pointed out, the pattern of citation by
mass communication
scholars is noticeably different from other disciplines since the
former tended to be
self-citing and reliant on older literature (Beniger, 1988;
Tankard, Chang & Tsang, 1984).
Therefore, citation rate may not be as good a measure for mass
communication research as
it is for other disciplines. Based upon these considerations,
we decided not to measure
this variable for the study.
The authors of this paper coded the articles in the nine journals with
help from nine
graduate students enrolled in a mass communication theory
seminar. Inter-coder
reliability was assessed for about 5% of the total articles coded, with
mutual agreement
ranging from 85% to 98% across all variables. The high
agreement was expected because of
the straightforward coding task involved.
Analysis
The six research questions formulated above call for a series of
descriptive and causal
analyses. For example, the first three research questions
involve aggregate-level
comparison between mass communication and the neighboring social
sciences over time,
including the data of 1954-78 as reported in Weaver and Gray
(1980), while the last three
research questions require individual-level, causal analysis of
the data within mass
communication. To validate our study, we also compare, whenever
appropriate, our results
with relevant content analysis (e.g., Copper, Potter & Pupagne,
1994) and survey studies
(Weaver & Wilhoit, 1988).
Because our data involve mostly categorical variables, statistical
significance is tested
based on a series of log-linear models. Therefore, since coefficients
estimated by
log-linear models are well-known for their difficulty in
interpretation, we report
descriptive statistics (mostly in percentage of funding rate)
whenever necessary.
Findings
Funding in Mass Communication Over Time
Of the 539 articles published in the six mass communication journals
between 1983 and
1993, 22% acknowledged receipt of financial support from an
intramural or extramural
source. This funding rate is comparable to a 1987 survey of 893
journalism and mass
communication educators by Weaver and Wilhoit (1988) in which 28%
of the respondents
reported receiving research grants within the last 12 months.[13]
Compared to the funding rate of 26% between 1954 and 1978 which was
found by Weaver and
Gray, the funding situation for mass communication research
appears to have deteriorated,
although the decline over time is not statistically
significant, based on a 2-way,
saturated log-linear model in which Funding (yes vs. no) is treated
as the dependent
variable and Time (1954-78 vs. 1983-93) as the independent
variable (g = -.053, s.e. =
.057, p < .60). However, it is instructive to note the
monotonic decline of the funding
rate, which dropped from 36% in 1954, to 28% in 1963, to 25% in
1983, and 21% in 1993
(Figure 1).
----------------------------
Figure 1 about here
----------------------------
Inter-disciplinary Comparison
The steady decline in funding for mass communication is puzzling
because there has been
no sign of reduction of funding during the same period in the
neighboring disciplines such
as political science, psychology and sociology. Between 1983 and 1993,
for example, 56%
of the 457 articles published in APSR, ASR, and JPSP were funded. In
fact, this funding
rate of 56% represents a slight increase over the previous
level (55%) between 1954 and
1978, although the increase is not statistically significant (g
= .012, s.e. = .044, p <
.80).
To test the differential funding rate between mass communication and
the three adjacent
disciplines over time, we fitted a series of 3-way log-linear
models using Funding as the
dependent variable and Time (1954-78 vs. 1983-93) and
Discipline (mass communication vs.
non-mass communication) as the independent variables. These
models differ in their
specifications of the presence or absence of the main-effects and
the interaction term.
The resulting best-fitted model is the one containing only one
main-effects variable,
Discipline (L2 = 2.481, df = 6, p < .85). No additional term to
the model will produce
any significant improvement given the reduction in the degrees
of freedom. The test of
the individual parameters (i.e., Time, Discipline, and Time x
Discipline) in the saturated
3-way log-linear model confirms that Discipline is the only significant
factor (g =
-.334, s.e. = .036, p < .000) that contributes to the observed
difference in funding rate
between mass communication and the neighboring social sciences.
Although there is no significant interaction effect between Time and
Discipline (g =
-.032, s.e. = .036, p < .60), which means that the difference in
funding rate between mass
communication and the neighboring disciplines is constant throughout the
period of
1954-93, a visual inspection of the long-term trend in Figure 1
shows a gradually widening
"funding-gap" between the disciplines over time. For example, the
observed difference in
funding rate between mass communication and the neighboring social
sciences is 28% in
1954-78, but 29% in 1983, 34% in 1988 and 38% in 1993.
Changes in Funding Source
One plausible explanation for the chronicle under-funding for mass
communication research
is the lack of support from the government. Between 1954 and 1978, Weaver
and Gray found
that mass communication research received only a quarter of
what political science,
sociology and psychology received from government agencies. Our
data show that the gap
has increased during 1983-93. As Figure 2 shows, while
government funding for the
adjacent disciplines remained about the same level (from 31% in
1954-78 to 29% in
1983-93),[14] there was a more substantial decline in government
funding for mass
communication (from 8% in 1954-78 to 5% in 1993).[15]
-----------------------------
Figures 2 about here
-----------------------------
Figure 2 also suggests another explanation for the decline in
financial support for mass
communication: the shrinkage of funding from the private
sector during the last decade.[16]
In 1954-78, mass communication actually enjoyed a substantial
edge over the neighboring
social sciences in attracting funding from private foundations
(9% vs. 4%). However, mass
communication has lost much of its advantage between 1983 and 1993 as
private funding
dropped from 9% to 6% for mass communication, while it remained
at 4% for the other social
sciences at the same time.
Funding from mixed sources (i.e., some combination of government,
private and university
sources) has remained unchanged for mass communication (2% in
1954-78 vs. 3% in 1983-93),
but increased noticeably for non-mass communication (5% in
1954-78 vs. 11% in 1983-93).
This adds to the widening funding-gap between mass
communication and the adjacent social
sciences. The only bright spot for mass communication research
is the intramural source.
From 1954-78 to 1983-93, there has been a slight increase in funding from
the university
for mass communication (from 7% to 9%) while the adjacent
disciplines have experienced a
slight decline in intramural funding (from 14% to 12%).
Similar to our test of inter-disciplinary difference over time as
reported above, we
examined the change in the composition of funding sources between
mass communication and
the neighboring disciplines over time by fitting a series of
log-linear models, using
Source (with five categories including university, government,
private, mixed sources, and
non-funded) as the dependent variable, and Time and Discipline as the
independent
variables. The models vary in the inclusion of main-effects and
interaction terms. The
results show that the best-fitted model contains only one
main-effects variable,
Discipline (L2 = 10.373, df = 8, p < .20). The inclusion of Time and
the interaction
between Discipline and Time does not provide any significant
improvement over and above
this model against the reduction in the degrees of freedom. As
compared to the
neighboring disciplines throughout the 40-year period, mass
communication has a
significantly lower rate in obtaining funding from the government (g =
-.576, s.e. = .094,
p < .000) or the mixed sources (g = -.304, s.e. = .142, p < .05), but a
significantly
higher rate in receiving funding from the private sector (g =
.482, s.e. = .119, p <
.001). On the other hand, no significant difference appears
across the disciplines in
getting funded within the university (g = -.058, s.e. = 1.008, p
< .60), with mass
communication being less funded despite the above improvement.
Characteristics of the Study
----------------------------
Table 1 about here
----------------------------
Within the field of mass communication, what kind of research is most
likely to attract
funding? One possible factor is the methodology used. As
Weaver and Gray found,
quantitative studies had a higher probability of receiving funding
than non-quantitative
studies during 1954-78 (col. 1 of the methodology panel in
Table 1). Our data of 1983-93
show that this pattern still holds (col. 2 in Table 1). Of the
539 articles we analyzed,
70% are considered quantitative studies, which include those
with a combination of
quantitative and qualitative methods, and 30% non-quantitative.[17]
The quantitative studies
appear to have a much better chance than the non-quantitative
studies in getting funded
(27% vs. 12%). However, while it has increased slightly (from
11% in 1954-78 to 12% in
1983-93) for the non-quantitative studies, the funding rate for
the quantitative studies
has in fact declined from 32% to 27% during the same period.
The test based on a 3-way
log-linear model (Funding by Time and Method) shows that the
difference in funding rate
between quantitative and non-quantitative studies is
significant throughout the time
period under study (g = .276, s.e. = .077, p < .001).
The type of data collection can also affect the funding rate. As the
data collection
panel of Table 1 shows, survey and experimental studies[18] have
continued to be more likely
to receive funding than content analysis and
historical-philosophical studies throughout
the entire 40-year period. A 3-way log-linear (Funding by Time
and Data Collection) shows
a significant difference between survey and content analysis (g = .556,
s.e. = .200, p <
.01), survey and historical-philosophical studies (g = .454,
s.e. = .142, p < .01),
experiment and content analysis (g = .634, s.e. = .239, p < .01)
and experiment and
historical-philosophical studies (g = .532, s.e. = .194, p < .01).
No significant
difference is found either between survey and experiment, or between
content analysis and
historical-philosophical studies. While mixed methods of data
collection were rare in
1954-78 (only three cases, or 2% of the total, with none being
funded), the
multiple-method approach has become increasingly popular (33
cases, or 6% of the total,
with 27% of the studies using mixed data collection methods
funded).[19]
The fundability of a study also depends on the type of medium studied.
Perloff (1976)
found that 56% of the articles published in JQ between 1955-74
were exclusively about
print media, while only 5% dealt exclusively with electronic
media. We found the
percentage of print-media studies to have dropped to 25% during
1983-93, while electronic
media studies rose to 43%. However, there has been a decline
in the proportion of studies
involving both print and electronic media (21% in 1955-74 vs. 11% in
1983-93). The
percentage for non-media studies remained about the same (18% in
1955-74 vs. 21% in
1983-93). As the last panel of Table 1 shows, research on print
media is the least likely
to get funding, with only 14% of the studies about print media being
funded. Electronic
media has a better chance (26%) of receiving funding. The
studies most likely to be
funded are the combination of both print and electronic (38%).
Based on a 2-way
log-linear model (Funding by Medium),[20] the studies of electronic
media or of combined media
have a significantly higher funding rate than the studies on the print
media (g = .381,
s.e. = .143, p < .01 for contrasting electronic and print
media, or g = .634, s.e. = .180,
p < .01 for contrasting between combined media and print-only). Studies
of combined
media also have a significantly higher rate of funding than
studies of non-media (g =
.479, s.e. = .180, p < .01), although the difference between
electronic media and combined
media, or between print and non-media, is not significant.
Characteristics of the Author
There appears to be some association between funding status and the
author's academic
rank, although the relationship is by no means linear. Our data
show that while full
professors have the highest rate of funding (30%), the next most
likely to be funded are
graduate students (27%) whose funding rate is higher than
instructors (23%), associate
professor (22%), assistant professors (21%), or academic
administrators (e.g., school
deans, department chairs and center/institute directors, 21%).
The least likely to
receive funding are those working outside the academic world. In
fact, none of the 19
non-academicians who published studies in the six mass
communication journals between 1983
and 1993 acknowledged receipt of funding.[21]
However, most of the inter-rank differences are not statistically
significant. A 2-way
log-linear test (Funding by Rank) reveals that the only group
significantly different from
the norm is the non-academicians. The group differs at the .05 level from
professors (g
= 1.099, s.e. = .431), graduate students (g = 1.025, s.e. =
.447), associate professors (g
= .892, s.e. = .430) and assistant professors (g = .870, s.e. = .428); and
it differs at
the .10 level from both instructors (g = .973 s.e. = .519) and
academic administrators (g
= .873, s.e. = .466). The other marginally significant
difference is between professors
and assistant professors (g = .228, s.e. = .142, p < .15).
Characteristics of the Institution
Institutional affiliation also seems to have some impact on the
fundability of a study.
But the relationship is by no means linear. Our data show that
28% of the authors at the
Research-I universities were funded, followed by 27% at the
Doctorate-II, 23% at the
B.A.-grant universities, 21% at the Research-II, and 19% at the
Doctorate-I. The least
likely to be funded are those at M.A.-grant universities (6%).
The average funding rate
for the miscellaneous group (including nonacademic, foreign and
unspecified institutions)
is 16%.
Similar to the test of the relationship between Funding and Rank, a
2-way log-linear
model (Funding by Institution) shows that the only major
significant gap exists between
the M.A.-grant universities and the rest. The difference
reaches the .05 level and beyond
between the M.A. schools and the Research-I (g = .886, s.e. = .284), the
Research II (g =
.697, s.e. = .319), the Doctorate-I (g = .654, s.e. = .336), Doctorate-II
(g = .860,
s.e. = .335), and the B.A. schools (g = .805, s.e. = .414). The
M.A. universities do not
differ significantly from the miscellaneous institutions,
whereas the Research-I
universities do have a significantly higher rate of funding than the
miscellaneous
institutions (g = .354, s.e. = .175, p < .05).
All Things Considered
So far, we have examined the possible impact of five factors
(methodology, study design,
medium studied, academic rank, and institutional affiliation)
on funding separately. It
is obvious that some of the relationships reported above may be
spurious because of the
possible inter-correlations between them. For example, while
we have reported that
content analysis and print media are among the least fundable
studies, these two are in
fact closely related. It is also possible that the impact of
academic rank on funding may
be confounded by methodology used because our data suggest that all ranks
of professors
are more likely to use quantitative methodology than academic
administrators, instructors
and graduate students. Therefore, it is necessary to conduct a
multivariate analysis
using all these five factors as independent variables and the
funding status as the
dependent variable.
As in the previous tests, we performed the multivariate test by
fitting a series of
log-linear models to the data, and calculated the difference in
the likelihood-ratio
between two hierarchically-nested models to determine the
significant, net contribution
made by each of the five independent variables. The first
model in Table 2 is a "baseline
model" because it contains all second-order interactions among Funding and
the five
independent variables.[22] Each of the next five models contains
the same 15 interaction
terms among the independent variables, but only four
interactions Funding and the
independent variables. The difference in the goodness-of-fit between
each of these
partial models and the baseline model (shown in col. 2 of Table 2)
is stringent test of
the net impact of that absent independent variable on Funding
is the difference between
the partial model and the full model.
----------------------------
Table 2 about here
----------------------------
As Table 2 shows, all but one of the independent variables survive
this multivariate
test. Relatively speaking, Medium is the strongest predictor of
funding (p < .01) while
Methodology, Author Rank and Institution Rank are all
significant at the .05 level. The
only non-significant variable is Data Collection, whose impact
on Funding is largely
overlapped with Methodology. Therefore, Model 3 (without the
interaction between Funding
and Data Collection) presents the most parsimonious model of the data.
Table 4 reports
the estimated parameters of this final model.
----------------------------
Table 3 about here
----------------------------
An inspection of the parameter estimates suggest that most of the
patterns uncovered in
the bivariate analyses reported earlier still hold in the
multivariate test. For example,
after controlling for all other factors, quantitative methodology still
has a
significantly higher rate of funding than non-quantitative methodology;
both mixed- and
electronic-medium studies are still more likely to be funded
than the print-only studies;
non-academic authors are still less funded than all ranks of
academic authors; and M.A.
universities are still significantly less successful in
attracting funding than either
Research-I or Doctoral-II universities. Nevertheless, several
other comparisons that were
significant in the bivariate tests have become non-significant. For
example, the
difference between professors and assistant professors (the least
funded academic authors)
is no longer significant; nor is the comparison of M.A. with Research-II,
Doctoral-I, or
B.A. institutions.
Summary and Discussion
We summarize the findings of this study by revisiting our six research
questions.
1. What is the current status of funding for mass communication
research? The funding
rate has continuously been declining, from an average of 26% in
the earlier period to 22%
in the last 15 years. Although the 4% difference is not
statistically significant,
largely due to the small sample size in the baseline study (Weaver
and Gray, 1980), the
decline has been a monotonic trend throughout the entire
40-year period. Furthermore, the
funding rate for mass communication research based on our content analysis
results
corroborates the findings from a survey of mass communication
scholars in the same period
(Weaver and Wilhoit, 1988).
2. Has the funding gap between mass communication and neighboring
social sciences been
closed? The answer is an unfortunate "No." The gap has
actually enlarged slightly, as
the ratio in the funding rate between mass communication and
the three adjacent
disciplines (psychology, sociology and political science) has increased
from 1:2.1 in
1954-78 to 1:2.5 in 1983-93.
3. Has there been any major change in the make-up of funding sources?
We have found a
mixed picture. Funding from both government and private
sources has shrunk from 8% and 9%
of the published studies in 1954-78 to 5% and 4% in 1983-93, respectively.
On the other
hand, mass communication scholars now fare better in getting
intramural funding (from 7%
to 9%). However, the support from universities for mass
communication research is still
at the level enjoyed by the neighboring social sciences (12%).
4. Looking inside the mass communication field, what kind of studies
have been most
likely to be funded? As in 1954-78, quantitative studies continue
to enjoy some advantage
over non-quantitative studies in getting funded. Independent of that,
studies involving
either electronic media or combining electronic and print media
fare better than
print-only studies.
5. Does the author's academic rank matter? Our data suggest that
academic rank has a
very limited effect on the fundability of a study when all other
factors are equal. The
only substantial difference found in the study is that the
non-academic authors have
received significantly less funding than all types of academic
authors ranging from full
professors to instructors and graduate students.
6. Does institutional affiliation matter? Again, it matters to a
limited degree. The
major difference occurs between M.A.-grant schools on the one
end, and Research-I,
Doctoral-II and miscellaneous institutions on the other end.
Research-II, Doctoral-I and
B.A.-grant universities lie in between but do not differ
significantly from either ends.
Among these findings, the decline in governmental support for mass
communication, from an
already low starting point, is probably the most important. This is
especially alarming
in light of the fact that the governmental funding for
psychology, sociology and political
science has not been reduced during the same period. One explanation is
that many
government funding agencies, such as the NSF, have not recognized
communication as an
academic discipline. Another possible reason for the continuous
decline in government
funding is that the 1980s and beyond have not seen any major
federally-funded programs,
such as "Surgeon General's Report" and several others during
the 1960-70s. Mass
communication scholars have a long way to go before being well
recognized and fully
supported by various funding sources within the government.
The decline in private money may have resulted from a set of quite
different reasons. As
our data show, as an applied field, mass communication still has a small
edge over the
neighboring disciplines in getting private funding. However,
this advantage is fading.
Without systematic investigation, we speculate that the finding
does not necessarily mean
that the private sector has lost interest in mass
communication. More likely, the money
has been shifted from research to education, assuming a
relatively stable amount of
funding has been available. A case in point is the proliferation
of privately-endowed
professorships in many journalism and mass communication schools
across the country.
While this is certainly a welcome trend, efforts need to be made
to convince the private
sector that research is equally important and equally worthy of
investment.
If there is any good new coming out this study, it is that almost
everyone within the
academic world, from full professor and administrators to
graduate students and
instructors, has equal footing in getting funding. Therefore, the
structural constraint
of academic rank is not at work. As noted in the methodology
section, the impact of
author's reputation could have also been measured by the citation
rate, as was done in the
two previous studies reviewed. Practical considerations prevented us from
including this
potentially influential variable but future research needs to overcome
this limitation.
Another structural constraint (institutional reputation) also appears
to have a limited
effect. For example, there is virtually no difference in
funding among Research-II,
Doctoral-I and B.A.-grant universities. Expectedly, those sitting
on the top of the
academic ivory power (i.e., Research-I institutions) have received
more funding than those
at lower ties. However, it is puzzling why the M.A.-grant schools,
instead of the
B.A.-grant schools, rank at the bottom of the list. This remains
to be further
investigated.
References
Beniger, James R. (1988). Information and Communication: The New
Convergence.
Communication Research, 15, 198-218.
Cole, S., Rubin, L., & Cole, J. R. (1978). Peer Review in the National
Science Foundation:
Phase One of a Study. Washington, D.C.: National Academy of Sciences,
1978.
Copper, R., Potter, W. J., & Dupagne, M. (1994). A Status Report on Methods
Used in Mass
Communication Research. Journalism Educator, 48(4), 54-61.
Danielson, W. A., & Wilhoit, G. C. (1967). A Computerized Bibliography of
Mass
Communication Research, 1955-1964. New York: Magazine Publishers
Association.
Heffner, A. G. (1981). Funded Research, Multiple Authorship, and
Subauthorship
Collaboration in Four Disciplines. Scientometrics, ?, 5-12.
Perloff, R. M. (1976). Journalism Research: A 20-Year Perspective.
Journalism Quarterly,
53, 123-126.
Tankard, James W., Chang, Tsan-Kuo, & Tsang, Kuo-Jen (1984). Citation
Networks as
Indicators of Journalism Research Activity. Journalism Quarterly,
61, 89-96.
Useem, M. (1976). Patterns in Government Financing of Academic Social
Research, American
Sociological Review, ?, 613-629.
Weaver, D. H. (1993). Communication Research in the 1990s: New Directions
and New Agendas?
In P. Gaunt (Ed.), Beyond Agendas: New Directions in Communication
Research (pp.
199-220). Westport, CT: Greenwood.
Weaver, D. H., & Gray, R. G. (1980). Journalism and Mass Communication
Research in the
United States: Past, Present and Future. In G. C. Wilhoit & H.
de Bock (Eds.), Mass
Communication Review Yearbook, Vol. 1 (pp. 124-151). Beverly
Hills, CA: Sage.
Weaver, D. H., & Wilhoit, G. C. (1988). A Profile of Journalism and Mass
Communication
Educators: Traits, Attitudes and Values. Journalism Educator,
Vol. 43, No. 2 (Summer
1988), pp. 4-41.
Wilhoit, G. C. (1981). Introduction. In G. C. Wilhoit & H. De Bock (Eds.),
Mass
Communication Review Yearbook, Vol. 2 (pp. 13-33). Beverly Hills, CA:
Sage.
Table 1. Percentage of Mass Communication Studies Funded (the number of
cases based on
which the percentage is calculated in parentheses)
1954-78a
1983-93
Change
1. Overall Rate
26%
(122)
22%
(539)
-4%
2. Source
Government
8%
(122)
5%
(539)
-3%
Private
9%
(122)
6%
(539)
-3%
University
7%
(122)
9%
(539)
+2%
Mixed
2%
(122)
3%
(539)
+1%
3. Methodology
Quantitative
32%
(87)
27%
(378)
-5%
Non-Quantitative
11%
(35)
12%
(161)
+1%
4. Data Collection
Survey
35%
(44)
32%
(190)
-3%
Experiment
50%
(12)
25%
(56)
-25%
Content Analysis
11%
(19)
16%
(128)
+5%
Historical-Philosophical
20%
(44)
13%
(132)
-7%
Mixed
0%
(3)
27%
(33)
+27%
5. Medium Studied
Print
n.a.
14%
(137)
Electronic
n.a.
26%
(231)
Mixed
n.a.
18%
(111)
Non-media
n.a.
38%
(60)
a The 1954-78 data are based on Weaver and Gray (1980).
Table 2. Goodness-of-Fit of Log-Linear Models of Funding
Model
L2
df
Difference in L2 from the Full Model
Difference in df from the Full Model
1. Full (with 5 predictors)
789.3
3,727
--
--
2. Methodology absent
793.9
3,728
4.7*
1
3. Data Collection absent
795.7
3,731
6.4
4
4. Medium absent
801.3
3,730
12.0**
3
5. Rank absent
804.2
3,733
14.9*
6
6. Institution absent
803.9
3,733
14.6*
6
* p < .05; ** p < .01.
Table 3. Log-Linear Estimates of Predictors of Funding
Parameter
-
s.e.
Methodology (non-
quantitative=0)
.499
.147
Medium (print=0)
Mixed-media
1.316**
.384
Electronic
.940**
.302
Non-media
.568
.371
Rank (nonacademic=0)
Graduate student
2.583*
1.109
Instructor
2.534*
1.266
Assistant prof.
2.442*
1.074
Associate prof.
2.502*
1.075
Full professor
2.992**
1.080
Administrator
2.602*
1.142
Institution (misc.=0)
Research-Ia
.153
.396
Research-II
-.363
.502
Doctoral-I
-.355
.552
Doctoral-IIb
.054
.551
M.A.a,b
-1.576*
.704
B.A.
.148
.776
* p < .05; ** p < .01; *** p < .001.
a, b indicates the pair differs at .05 level or beyond.
Appendix. A Partial List of Extramural Funding Sources (with the frequency
in
parentheses)
Government
National Science Foundation (7)
National Institute of Health (3)
National Institute of Mental Health (3)
Social Science and Humanities Research Council of Canada (3)
Fulbright Programs (2)
National Endowment for Humanities (2)
National Heart, Lung and Blood Institute (2)
National Institute of Education (2)
Minnesota Heart Health Program (2)
Private
Gannett (including Gannett Co., Gannett Foundation, Freedom Forum) (10)
National Association of Broadcasters (5)
Dow Jones Newspaper Fund (3)
Spencer Foundation (3)
Ameritech (2)
Shell (Shell Foundation and Shell Research Foundation) (2)
Note: A total of 57 funding agencies were acknowledged. Only those
mentioned more than
once are listed here.
[1] All major research articles in JQ but only mass communication-
related articles in POQ
were included in their study.
[2] This funding rate is identic
al to the findings from a 1976 study of journal articles
in political scie
nce and psychology by Heffner (1981).
[3] The survey asked questions such as "Do you hop
e that your research ... will directly
or indirectly benefit ... the Feder
al government?" to ascertain policy (Useem, 1976, p.
617) and other simil
ar self-reported questions to measure political perspective and
governmen
t confidence.
[4] All percentages of explained variance were based on a probit model.
Citation counts
were gathered from the authors' proposal jackets; rank of current depar
tments taken from
the 1969 American Council on Education (ACE) rankings; r
ank of Ph.D. departments from the
1964 ACE; and academic rank comprised a
six point scale from researcher to full professor
(see Cole et al., 1978
, pp. 178-181).
[5] In a recent content analysis of uses of methodology in mass commun
ication research,
Cooper, Potter and Dupagne (1994) included all the journ
als listed above plus three more
-- Communication Monographs, Human Commu
nication Research, and Quarterly Journal of Speech
Communication. These are not include
d in the current study because the professional
associations these three
journals represent (ICA and SCA, respectively) are already
covered in our
sample and, more importantly, the majority of articles in these journals
deal with non-mass communication content.
[6] By coincidence, our sample of JOC inclu
ded several special issues. For example, No.
3 of 1983 was the famous "Ferment in the F
ield" and No.'s 3 & 4 of 1993 were the
bidecennial of the Ferment. These
specials may introduce two types of biases: on the one
hand, the articles in the specia
l issues were mostly invited commentaries, and therefore,
less likely to be funded (henc
e, an underestimate of our dependent variable), on the
other hand, the au
thors were more likely to be prominent scholars in the field (thus, an
ov
erestimate of one of our independent variables). Therefore, we substituted
each of
these special issues with the same issues in the following year.
Specifically,
No. 3 of
1983 ("Ferment") is replaced by No. 3 of 1984, No. 4 of 1983
("
Books") by No. 4 of 1984,
No. 1 of 1988 ("International Telecommunication
Policy") by No. 1 of 1989, and No.'s 3 & 4
of 1993 ("Future of the Field" I and II) by
No.'s 3 & 4 of 1994, respectively. Other
theme-specific symposia in JOC
that involved referred articles of empirical studies are
not excluded bec
ause the articles in these symposia are essentially the same as regular
p
ublications.
[7] Before merging the six journals into one group, we examined the
averag
e funding rate
of JQ/POQ and that of the four additional journals, and found no signific
ant difference
between the two groups, either across the entire 1983-93 pe
riod or within each sampled
year.
[8] Thus a study combining quantitat
ive and qualitative methodologies is coded as
"quantitative."
[9] Metho
dology and data collection are two distinct concepts. For example, there
are
quantitative as well as qualitative content analyses.
Historical-philosophical data
collection can also involve both qualitative and quantitative
method (e.g.
, quantitative
analysis of historical records). See Weaver and Gray (1980
) for more details.
[10] Weaver and Gray did not examine the type of medium but cited a
baseline study
(Perloff, 1976) that involved the same publication (JQ) dur
ing a similar time period (from
1955 to 1974). The categories used in the current study
come from Perloff. The category
of non-media includes, for example, journalism educati
on or mass communication theory.
[11] However, if the acknowledgment indicates that the
author or first author of the
article was not the recipient of the funding
, as in several cases, then the rank of the
actual recipient (or the firs
t name in the recipient list) was coded. Our rationale was
that we were
interested in predicting the funding status based on, in part, the academic
rank of the recipient.
[12] The "Carnegie Classification" rankings of both 1987 and 1
994 can be found in The
Chronicle of Higher Education (April 6, 1994), and
utilizes a 10-point scale including
Research-I, Research-II, Doctoral-I,
Doctoral-II, MA-I, MA-II, BA-I, BA-II, AA (associate
of arts colleges) a
nd all other post secondary-schools. Based on the observed
frequencies,
we combined MA-I and MA-II into one category, and BA-I, BA-II, AA, and all
others into another.
[13] If we excluded from our sample graduate students, non-acad
emic and foreign authors
(who were not presented in Weaver and Wilhoit's s
urvey), the funding rate for the
published studies would increase to 24%.
The remaining 4% difference between our study an
d Weaver and Wilhoit's
survey falls into the boundaries of sampling error.
[14] Note that this 2% difference do
es not necessarily means a decline in government
funding for these discipl
ines because there has been a significant increase in funding
from mixed
sources, which include the government, for these disciplines. See the
discus
sion below.
[15] Of the government-funded studies, a diverse list of sources were ac
knowledged.
Among the most frequently cited sources of funding was the Na
tional Science Foundation (7
times), followed by the National Institute o
f Health (3), the National Institute of Mental
Health (3) and the Social Sciences and Hu
manities Research Council of Canada (3). See
Appendix 1 for a detailed l
ist of the governmental funding sources.
[16] The most frequently acknowledged private f
oundations include Gannett (10 times), the
National Association of Broadcasters (5), Dow
Jones (3) and Spencer (3). See Appendix 1
for a more detailed list of th
e private funding sources.
[17] This ratio between quantitative and non-quantitative is
very close to what Copper et
al. (1994) found in their content analysis of eight communi
cation journals. Calculated
based on their Table 2 (p. 57), the ratio bet
ween quantitative (including studies of both
quantitative and qualitative
methods) and non-quantitative for 1983-89 is 69% vs. 31%, or
1% differen
t from our finding.
[18] Although the funding rate for experimental studies has droppe
d from 50% in 1954-78
to 25% in 1983-93, we should notice the small number
of cases (12) in 1954-78, based on
which the 50% funding rate was derive
d.
[19] The estimated percentage for multiple methodology studies was 0 in
1944-64
(Danielson and Wilhoit, 1967), 3% in 1954-78 (Weaver and Gray, 1980),
5% in 197
8-80
(Wilhoit, 1981), and 9% in 1983-93 (the current study).
[20] Time,
the third variable in all previous log-linear models, was not included
because
the Perl
off study of 1955-74 did not measure funding.
[21] A somewhat different pattern was repo
rted by Weaver and Wilhoit (1988) based on
their 1987 survey of journalism
and mass communication educators. They found virtually no
difference in fundability ac
ross the three levels of professorship (32% of assistant, 30%
of associate, and 31% of f
ull professors). On the other hand, 22% of the administrators
or 14% of
the instructors in their survey reported receiving funding. No graduate
students or non-academicians were included in that survey.
[22] This is not a saturated
model, however, because only all possible second-order
interactions are sp
ecified. Third- or higher order interactions are excluded because none
are statisticall
y significant.
|
