Trends in Newspaper Coverage of Science Over Three Decades: A Content Analytic
Marianne G. Pellechia
Department of Communication
West Lafayette, Indiana, 47907
E-mail: [log in to unmask]
Running head: Trends in newspaper coverage of science over three decades:
A content analytic study
This paper describes a content analysis of science news reporting in three
major daily newspapers during the last three decades. Though there were some
differences between articles appearing in the different time frames, in general
science news reporting has not changed significantly in terms of the
comprehensiveness of accounts. An especially significant finding is that
articles frequently omitted methodological and contextual information, features
most often mentioned as critical for a complete journalistic account of science.
Trends in Newspaper Coverage of Science Over Three Decades: A Content Analytic
Through the decades, science and scientific research has had a growing impact,
both directly and indirectly, on public life. As a result, the general public
has shown signs of a growing interest in science and science policy. Nunn
(1979) found that newspaper audiences had a high level of interest in science
news, particularly among young adults. Miller (1986) also found interest in
science high, with 40% of American adults expressing interest in science and
science policy. It is of interest to know, then, how scientific information and
the results of scientific research have been reported to the general public.
This issue is the central concern of this paper. After a brief survey of the
relevant literature, a content analysis of science articles in three major daily
newspapers over three decades will be described. The results of this study shed
light on some of the trends that have occurred in science news reporting through
Today the popular press is paying increased attention to science and
technology. There has been a movement among the mass media to serve the
public's interest in science, with a growing number of newspapers featuring
weekly science sections and/or employing full time science writers to write for
their daily editions (Bader, 1990; Scientists Institute for Public Information,
1986) as well as the emergence of 17 television shows, 15 magazines, and 18
newspaper sections devoted to science between the years 1974 and 1984
(Lewenstein, 1987). This mass media and news coverage of science has become
increasingly important, as it can influence both the public's knowledge and
attitudes about a wide variety of scientific issues. That is, today's science
journalists report on an extensive range of subjects, including the latest
advancements in medicine, energy, environmental issues, and technological
developments. This variety in subject areas should not be surprising when one
considers that the field of science has become increasingly diversified through
Many researchers have examined various aspects of science news. While studies
have been done on various media forms including magazines (Borman, 1978;
Lundberg, 1984; Rich, 1981) and the network news (Moore & Singletary, 1985 )
most studies have focused on science news reporting in newspapers. Newspapers
are an important medium to study for a number of reasons. While a study of
young adults conducted by Patterson (1982) found that people learn most of the
latest science news through newspaper accounts, more recent research (Howard,
Blick, & Quarles 1987; Scientists' Institute for Public Information 1993) has
shown that television may be replacing newspapers as the preferred source for
science news. Still, when one considers that a national survey of scientists
revealed that 76.6% of the respondents who had participated in an interview had
done so with newspapers (DiBella, Ferri, & Padderud, 1991), the importance of
newspapers as a source of science news cannot be denied. Such aspects of
science reporting as accuracy (Pulford, 1976; Ryan, 1975; Tankard & Ryan, 1974;
Tichenor, Olien, Harrison, & Donohue, 1970), comprehensiveness (Freimuth,
Greenberg, DeWitt, & Romano, 1984; Sullivan, 1985), and sensationalism (Glynn,
1985; Glynn & Tims, 1982) have all been examined as components of newspaper
Accuracy is the primary concern of scientists, but the technical nature of
science reporting may make it more prone to error than news reports about other
topics. Studies investigating the accuracy of popularized accounts of science
revealed that scientists often consider inaccuracy to be the major problem with
science reporting. Tankard and Ryan (1974) found that when evaluated by
scientists, only 8.8% of science articles were judged to be error free compared
to rates of 40 to 59% for other news stories. However a follow-up study by
Pulford (1976), using a shorter checklist of possible errors, found 29.4% of
articles were judged to contain no inaccuracies. Tichenor, Olien, Harrison, and
Donohue (1970), found that slightly more than 40% of scientists surveyed
disagreed with the statement that science news is generally accurate, whereas
Dunwoody and Scott (1982) found 51% of the scientists surveyed criticized
science reporting for inaccuracy or distortion. However, while scientists are
critical of science reporting in general, the same scientists tend to be more
favorable when it comes to evaluating the accuracy of news stories about their
own work (Dunwoody & Scott, 1982; Pulford, 1976; Tichenor, Olien, Harrison, &
One potential problem with such studies, however, is that they have focused on
accuracy as perceived by the scientists, and are therefore limited by
subjectivity. That is, according to McCall (1988) accuracy research that is
based on the perceptions of the source is limited by the subjective nature of
the task. To overcome this limitation and to investigate the common errors of
science reporting, Singer (1990) compared news reports of scientific topics from
a variety of media (i.e., newspapers, newsmagazines, and television) to the
original research reports appearing in the scientific literature. Out of the 42
articles studied, only 7.1% were found to contain no inaccuracies. Common
errors found in the accounts analyzed included omission of qualifying statements
(found in 60% of the accounts), a lack of discussion of the methodology followed
(48% of the accounts), a "change of emphasis" (45% of the accounts) and an
overstatement of the generalizability of the research results (36% of the
Inaccuracies such as those cited by scientists in accuracy studies and
uncovered by Singer (1990) are primarily subjective inaccuracies which Dunwoody
(1982) defines as errors in meaning. In contrast to objective inaccuracies,
such as incorrect statements or errors of fact, subjective inaccuracies are
dependent on an individual's interpretation and thus are not always recognized
by all parties. Indeed, numerous studies have shown that when scientists say a
study is inaccurate, they are really referring to errors of omission and
misemphasis (Dunwoody & Scott, 1982; Pulford, 1976; Tankard & Ryan, 1974;
Tichenor, Olien, Harrison, & Donohue, 1970). Of these two subjective errors,
the most commonly cited throughout the accuracy studies is omission of
information. Scientists main complaint of journalistic accounts of scientific
research is that such accounts are incomplete because they lack relevant and
important information (Borman, 1978; Dunwoody, 1982; Pulford, 1976; Rich 1981;
Tankard & Ryan, 1974), information needed by the public to make informed
judgments. That is, "the requirement of substantial completeness is satisfied
if adequate information is published or broadcast to meet the needs of an
intelligent nonspecialist who wants to evaluate the situation being reported on"
(Klaidman 1990, p. 120). Thus, this can be viewed as more of an issue of the
completeness or comprehensiveness of such accounts rather than of accuracy.
One of the most common omissions is the omission of important details of
methodology (Dunwoody, 1986; Goldstein, 1986). In the study by Tankard and Ryan
(1974), the scientists surveyed ranked "relevant information about the method of
study omitted" as the most serious problem of newspaper accounts of science.
The more objective findings of Singer (1990) support this view, with 48% of the
42 articles studied having no mention of research methods at all. Furthermore,
of the studies that did mention research methods, 15 or 35% gave what was
considered inadequate information, while 3 or 7.1% gave erroneous information.
Similar results were found in a study comparing science reporting in both the
prestige and national tabloid presses (Evans, Krippendorf, Yoon, Posluszny, &
Thomas, 1990). Of the newspaper accounts analyzed, only 36.2% of the articles
in the New York Times and Philadelphia Inquirer and 19.8% of the tabloid
articles contained a "more than minimal" discussion of the research design,
leading the authors to conclude "that at neither the prestige nor the national
tabloid newspapers is it common practice to include methodological details."
Molitor (1994) found that five major national newspapers gave an incomplete
description of what turned out to be a highly unrepresentative sample used in an
aspirin and heart attack study (reported in the New England Journal of
Medicine), information that was necessary to interpret the generalizability of
the study's findings to the general public.
Other common omissions include criticism of the research and theory by other
scientists, qualifying statements and/or other information that would limit the
findings or conclusions of the research (Dunwoody, 1986; Goldstein, 1986).
However, it is important to note that though science journalists increasingly
recognize the need to use qualifying or cautionary language in their copy
(Burkett, 1986; Gorchow, 1990, Perlman, 1974), and thus may write using such
cautionary phrases as "preliminary" or "early results, "many readers do not
recognize the implications of these words (Gorchow, 1990).
Another commonly recognized problem with science news reporting is that
previous research is often ignored. This omission can lead to misemphasis
within science stories and sensationalized accounts of new developments, with
research findings presented as discrete events or "breakthroughs" and tentative
findings treated as scientific fact (Freimuth, Greenburg, DeWitt, & Romano,
1984; Goodell, 1985; Greenburg, Freimuth, & Bratic, 1979; Molitor, 1994;
Trachtman, 1982). Indeed, the scientists in the Tankard & Ryan (1974) study
cited "continuity of research with earlier work ignored" as one of the 9 major
problems of science reporting. Thus, it has been asserted that science should
be treated as "more of a motion picture instead of a snapshot" (Bodde, 1982) as
there is as much a need for the general public to understand the internal
processes by which scientific knowledge is generated and validated as there is
the content of the specific areas of science (Millar & Wynne, 1988). LaFollette
(1995) echoes this view, asserting that
Effective modern citizenship demands a higher level of 'knowing about'
science...Acknowledgment that all sciences have histories is part of this type
of knowledge. Research knowledge about any one subject at any one time
represents an accumulated body of information, as science continually
reconstructs itself. To 'know about' science than is to know about that
reconstruction and reconstitution (p.235).
Though opinions are often expressed about the adequacy of various kinds of
coverage by facets of the media, this presents a problem when casual
observation, rather than careful documentation, becomes a basis for serious
suggestions for major changes in the mass media (Stempel, 1981). The issue of
the quality of newspaper coverage of science can be looked at in this framework.
While scientists' impressions indicate a lack of comprehensiveness in many
respects, formal documentation is needed. In addition, differences of opinion
exist on whether the rise of the professional science writer or journalist,
which began in the 1920s-1930s, has been associated with an improvement in the
quality of science reporting (Burkett, 1986; Burnham, 1987). Though it has been
suggested that because these journalists are trained in science many may adopt
the values of scientists and lose their ability to be critical (Cole, 1975;
Nelkin, 1987), others associate the rise of the professional science writer with
improvements in the comprehensiveness of science reporting (Burkett, 1986;
Focus of Present Study
According to Elliott and Rosenberg (1987), many models have been utilized to
study the communication of scientific information. Some of these approaches,
including coorientation, knowledge gap, and agenda setting "have focused on
specific media or process effects" of science communication (Elliott & Rosenberg
1987, p. 168). For example, Mazur (1981) found that with increased media
coverage of a controversial science and/or technology issue public opinion
tended to indicate that people not only were inclined to recognize the
controversy but also that there tended to be increased opposition to the
technologies in question as well. This then provided support of an agenda
setting effect of the media in terms of science news. However theories or
models such as agenda setting do not appear to enlighten the issue of the
content of science articles or to make specific predictions about the content
and quality of such communications and how these may have changed through time.
We may, however, have some pretheoretic conceptions about newspaper coverage of
science and some expectations about how this coverage may have changed through
time based on prior research reviewed earlier. Content analysis is useful and
valuable in analyzing trends and changes in content over time (Ogles, 1985) and
thus it is a useful technique for examining changes in newspaper coverage of
science. In this study, content analysis was used to compare and uncover trends
in science coverage in three major daily newspapers over the three time periods
of 1966-1970, 1976-1980, and 1986-1990. While this study attempts to address a
number of issues surrounding science news reporting, the primary focus of this
content analysis is on the omission of important and relevant information from
journalistic accounts of science. This area was emphasized because many of the
discrepancies both cited and observed in science reporting in past studies
(Singer 1990; Tankard and Ryan 1974) involve omissions of such critical
information as qualifying statements (for example indicating limitations in the
research), methodological details, or significant findings. The importance and
implications of the omission of such critical information cannot be overlooked.
That is, in the words of Singer (1990)
Whether such omissions and alterations should be regarded as inaccurate
reporting depends on how we define accuracy. If readers and viewers are not
made aware of [any] contingencies, if mass media accounts do not reflect
limitations in the data or research methods used, and if conflicting findings
are presented without interpretation or evaluation, then flaws exist in the
communication process, whether we call these flaws 'inaccuracies' or not (p.
With this in mind, this study attempted to answer three questions:
1) Has overall coverage of science by newspapers become more frequent and does
it cover a broader range of topics? Because public interest in science has
increased and the scientific community has become more diversified, one might
expect to find broader and more regular coverage during the latter periods.
2) Have articles become more complete and comprehensive as time has progressed?
In this study comprehensiveness concerns the inclusion of pertinent factual
information. Since more reporters covering science are professionally trained
science writers, it may be expected that more comprehensive articles appear in
the latter time frames.
3) Do "methodological citations" (Caudill & Ashdown, 1989) or descriptions of
the research design employed, appear more frequently? Are these citations more
thorough in articles from particular time frames? As lack of "relevant
information about the method of study omitted" was ranked as the most serious
problem by the scientists surveyed in the Tankard and Ryan study (1974), this
seems to be an especially significant area to analyze. Such statements of
methodology would influence opinion about the merits of the study and provide
readers with enough information to assess the validity of the research for
themselves. It may be expected that such methodological citations appear more
frequently in articles in the latter time periods.
The three time periods examined in this study were chosen because they reflect
periods marked by a growing interest in science and technology in the general
public, as well as an increased effort by the mass media in general and
newspapers in particular to meet those interests. It was during the period of
the 1960's that an increase in the public interest in science and technology can
be readily seen. For example, Swinehart and McLeod (1960, cited in Elliott &
Rosenberg, 1987) found increased interest in science and technology following
the launch of the first space satellite, Sputnik I, in 1957. This interest
continued to grow throughout the 1960's and the so called space race era. The
period of the middle to late 1970's then saw the breakthrough of popular science
into the media including the development of specialized science sections in a
number of major daily newspapers (Lewenstein 1987). This, according to
Lewenstein can be attributed in part to maturation of the World War II baby boom
generation which grew up "with Sputnik, the environmental movement, the war on
cancer, the Space Program and the energy crisis" and thus had "an explosive
hunger for more information on science and technology" (Lewenstein, 1987, p.
30). Finally, the period of the middle to late 1980's still saw a strengthening
of weekly science sections in newspapers, with a threefold increase in the
number of such sections from 1984-1986 (Lewenstein, 1987).
According to Stempel (1981) and Ogles (1985), analysts should employ coding
systems already developed and previously used by other researchers, as some
measure of their usefulness will be known and this was the approach used in this
study. A coding scheme to answer all three questions was developed using
schemes from four previously published studies (Caudill & Ashdown, 1989; Cole,
1975; Evans, Krippendorf, Yoon, Posluszny, & Thomas, 1990; Hinkle & Elliott,
1989) all of which used content analysis to investigate various aspects of
science reporting in newspapers.
Cole's (1975) definition of science news was used in this study to identify
those stories to be analyzed. Using this definition, science news was defined
as all news stories that have substantial subject matter concerning the results
and interpretation of empirical research in the sciences, applied science or
development, technology, engineering, medicine, and public health. Stories
provided by staff writers, news services, and wire services were analyzed.
Stories by syndicated or local columnists, editorials, or other articles printed
on the opinion-editorial page were not considered to be science news.
To further clarify what constitutes scientific research, the scheme of Evans,
Krippendorf, Yoon, Posluszny, and Thomas (1990) was applied. Thus, an article
was considered to deal primarily with scientific research "if its major focus
was on specific findings of a scientific research endeavor." According to this
scheme then, while an article may merely cite a scientist, or may incidentally
report on the planning of scientific projects, it was not coded as science news
unless it discussed specific research findings. As in the Evans, Krippendorf,
Yoon, Posluszny, and Thomas (1990) article, only articles that focus on the
results of completed scientific studies were analyzed. Stories identified as
science news were divided into three categories:
1) Medicine and Health: Stories in this category include news of research in
health related areas. Examples of items in this category include research on
new drugs, surgical procedures, and diseases, as well as research in health,
nutrition and fitness. Articles on psychology were also included if they were
the result of scientific research conducted by a researcher affiliated with a
hospital or university.
2) Technology: This category included stories that focused on developments in
the fields of engineering and applied sciences. Stories falling in this category
included news of the space program, computer technology, and superconductor
3) Natural and Physical Science: Stories included theoretically oriented
stories from the areas of astronomy, physics, chemistry, biology, zoology,
marine biology, and environmental science.
Question 1, which focuses on the amount of science coverage in the different
time frames and the range of topics covered, was investigated by answering the
a) Are there differences in the types of science covered by the newspapers
during the periods analyzed? Differences were determined by comparing the
number of stories falling under each category.
b) Do newspapers in the different time periods differ in the amount of science
covered? Differences in amount of coverage were indicated by differences in the
number of science stories per issue.
c) Do the newspapers from the different time periods differ in the emphasis
they place on science stories? Following the coding scheme of Hinkle and
Elliott (1989), "emphasis" was operationally defined as the percentage of
science stories as a percentage of all news stories.
To determine this percentage, all news articles in each of the newspaper issues
analyzed were counted. To be counted as an article, a news story had to be at
least 75 words in length. The following items were not counted as news
articles: obituaries, social and business announcements (i.e., engagements,
weddings, promotions, etc.), pieces on the opinions and editorial pages,
stockmarket listings, calendars of events (or other announcements of such things
as meetings, recitals, lectures, etc.) any weekly or otherwise regular column
and any reviews.
To answer question 2, concerning the comprehensiveness of accounts, all science
news articles contained in the newspapers analyzed were coded for a variety of
content. Many of these features have been identified as important for media
coverage of science and include
- depth of coverage. This is operationalized as the number of lines and number
of words per science story (Hinkle & Elliott, 1989).
- identification of the original forum or source of the research, such as
journal article, conference or symposium paper, book, etc. (Evans, Krippendorf,
Yoon, Posluszny, & Thomas, 1990);
- identification of the researcher(s) by name (Borman 1978; Lundberg 1984);
- identification of the researcher(s)' institutional affiliation(s) (e.g.,
university, government, hospital, private firm, etc.);
- frequency of comments from the researcher(s) who conducted the study(ies);
- frequency of comments from other scientists, in support of the findings
- frequency of comments from scientists with opposing viewpoints (Perlman,
- contextual factors. According to Evans, Krippendorf, Yoon, Posluszny, and
Thomas (1990) these include comments that place the research in context with
prior research, as well as comments regarding the limitations or
generalizability of the research findings.
According to a 1986 survey by the Scientists' Institute for Public Information,
while the New York Times was the only newspaper that had a weekly science
section in 1978, by 1986 66 daily newspapers had added weekly science sections,
and more than 80 started shorter science pages (Scientists' Institute for Public
Information, 1986). The introduction of such sections has been found to affect
newspaper science coverage, by increasing both the number of science stories and
length of such stories elsewhere in the newspaper as well (Bader, 1990).
Therefore, individual articles were also coded for the following:
- news source of article (i.e., wire, staff writer, etc.)
- appearance in weekly science section or in other section.
Question 3 investigates differences in the appearance of methodological
citation, or a statement of the research design, used in the study being
discussed. Therefore the unit of analysis was "statement of methodology."
Using the scheme developed by Caudill and Ashdown (1989), statements of
methodology were analyzed by counting the number of words used to explain the
research method. As operationalized by those authors, an explanation of less
than 30 words or less was considered brief, and thus inadequate. In addition,
methodological citations were also be coded for the manner in which the
statement characterizes the research, for example "method specific" (referring
to a specific research methodology such as experiment, survey, etc.) or method
nonspecific (for example phrases such as "study," "research," or "report").
Three major daily newspapers, The New York Times, The Chicago Tribune, and The
Washington Post were analyzed in this study. These newspapers were chosen
because they are widely distributed on a national level. A constructed week for
each of the three time periods being examined was generated using a computer
program written in MicroSoft Quick Basic. In addition to this constructed week,
a random Tuesday from each of the time periods was also generated using the same
program. This was done to increase the number of science news articles to be
analyzed, as Tuesday is the day that science sections appear in many newspapers.
Thus, a one-week plus one day period for each of the three time periods was
analyzed for each of the three newspapers. This resulted in a sample of 72
newspapers and 107 individual articles to be coded.
Each newspaper edition in the drawn sample was coded by a primary coder.
Furthermore, a randomly selected subset representing ten percent of the
newspaper editions and of the science articles was coded by a trained coder to
determine intercoder reliability.
Research Question One: Amount and Range of Coverage
The results show that although science makes up a small percentage of articles
in each period, there was an increase in emphasis on science issues in the later
decades. As can be seen in Table 1, the percentage of science articles
increased the later the time period. That is, in
Insert Table 1 about here
1966-70, science articles made up 0.42% of the week's articles. This increased
to 1.42% in 1976-1980, and finally 2.04 % in 1986-1990.
To determine if there are differences in the types of stories covered during
the different periods, categories of science articles were analyzed (Table 2).
In terms of the diversity of
Insert Table 2 about here
coverage, there was little difference in the range of topics covered in each of
the three periods. Rather, coverage was very similar in each of the time frames
studied, with an emphasis on medical and health related issues. More than 70%
of the articles in each period were classified as dealing with medicine and
health (72.22% in 1966-70, 75.76% in 1976-80 and 71.43% in 1986-90). Articles
on natural and physical science were the next most frequent article type,
accounting for 16.67% of the articles in 1966-70, 24.24% of those in 1976-80,
and 25% of the articles from 1986-90. Articles dealing with technological
issues and developments were the least frequent in all three time periods
(making up 11.11% of the articles in 1966-70, versus 0% of those in 1976-80 and
3.57% of those in 1986-90). There was no significant difference between the
three time periods in terms of the of the subject matter covered in the articles
analyzed ((2 [4, N=107] = 4.42, p>.05). Scott's pi for the measure of story
type was 0.80.
Considering the growing diversity of the scientific community, such similarity
in coverage across the three time periods is contrary to what might be expected.
Rather, based on these results, science reporting over the decades has not
reflected this increased diversification. However, the emphasis on medicine and
health issues should not be surprising, considering that biomedical issues have
been a dominant theme in the newspaper coverage of science (Meadows, 1986).
These findings then, are in agreement with those of Hinkle and Elliott (1989)
who found an emphasis on medicine and health related issues in their analysis of
three newspapers and three tabloids.
Research Question Two: Comprehensiveness of Science Articles
Articles were coded for a variety of content that reflect the comprehensiveness
of the coverage. These measures reflect content that has been identified in
past research as necessary features of a complete, comprehensive, and accurate
journalistic account of science. Table 3 contains the information on these
various article measures for each of the three time periods.
Insert Table 3 about here
To get an indication of the depth of coverage, articles in each time frame were
coded for both the number of lines and the number of words contained in the
story. Articles in each of the three time periods contained a similar number of
lines per article. In 1966-70, the average number of lines per article was
108.83, versus 100.58 lines in 1976-80 and 117.70 lines in 1986-90. A one-way
ANOVA revealed no significant difference among these means (F [2, 104] = 0.38,
p>.05). However, the number of words in articles increased the later the time
period. In 1966-70, the mean number of words was 487.09. This increased to
561.75 words per article in 1976-80 and 607.50 words in 1986-90. Though the
number of words increased, a one-way ANOVA revealed no significant difference
among these means (F [2, 104] = 0.49, p >.05). The percentage agreement for
number of lines per article was 99% and for words per article was 98%.
Articles were additionally coded for a variety of other variables. These
variables reflect what have been referred to by Evans, Krippendorf, Yoon,
Posluszny, and Thomas (1990) as "minimal details" that should be included in
science news accounts. In terms of identification of the original research
source (i.e., a journal article, conference presentation, etc.), there was no
significant difference among the three time periods for inclusion of such
information ((2 [2, N=107] = 1.28, p>.05). That is, in all three time periods,
the percentages of articles that identified the research source were similar
(72.22% in 1966-70, 60.60% in 1976-80, and 69.64% in 1986-90). Scott's pi for
this measure was 0.80.
In another measure of comprehensiveness of reporting, articles were coded for
whether the researchers who conducted the study were clearly identified in the
article by name (Scott's pi=1.00). There was a significant difference among the
three time periods in the tendency to include this information ((2 [2, N=107]
=7.41, p<.05) For the period of 1966-70, 94.44% of the articles identified the
researcher(s) whereas in 1976-80 only 72.73% of the articles included this
information. This percentage rose to 89.29% during 1986-90. Scientists have
cited the omission of researchers' names as a relatively common characteristic
of print coverage of scientific research (Borman, 1978; Tankard & Ryan, 1974).
Indeed, Evans, Krippendorf, Yoon, Posluszny, and Thomas (1990) found that only
70.3% of articles from the New York Times and Philadelphia Inquirer analyzed
identified any researchers by name. While the majority of articles in this
study tended to include this information, there was a decline in this practice
during the period of 1976-80 in the articles analyzed.
Additionally, chi square yielded a significant ((2 [2, N=107] =29.41, p<.05)
difference between the time periods for identification of the researcher(s)'
institutional affiliation (Scott's pi=0.80). In the period of 1966-70, 100% of
the articles contained such information, compared to only 69.70% of the articles
in the period 1976-80. In 1986-90, 94.64% of the articles contained this
information. Once again there was a decline in the percentage of articles
including this information during the period of 1976-1980, and a subsequent rise
during 1986-90. The reasons behind such a pattern of results remain unclear.
Articles were also coded for the presence of comments made by the researcher(s)
who conducted the study. A one-way ANOVA yielded no significant difference in
the number of comments from the researchers contained in the article (F [2, 104]
= 0.58, p>.05). The mean numbers of researcher(s)' comments in articles from
1966-70 were 4.11, compared to a mean of 3.73 for those articles in 1976-80 and
4.05 for those in 1986-90. Percentage agreement for this measure was 80%.
In addition to comments of the researchers, it has been asserted that one of
the ways science writers can exercise "clinical judgment" in the stories they
cover is to include comments from other scientists who are unconnected to the
reported research (Perlman, 1974). Thus, any complete and comprehensive
journalistic account of science should also contain comments from other
scientists who support the research findings, as well as from those who are
critical of the findings. Such unfavorable comments are especially important,
since they act as disclaimers and thus can often serve to qualify the findings.
In light of this then, articles were coded for the presence of both favorable
and unfavorable comments from other scientists, in regard to the research. A
one-way ANOVA revealed no significant difference between the number of favorable
comments from other scientists included in articles from the three time periods
(F [2,104] = 1.29, p>.05). However, in terms of including comments from
scientists critical of the research being discussed, there was a significant
difference (F [2, 104] = 3.62, p>.05) between the time frames. That is, in
1966-70, the mean number of comments was 0, versus 0.15 comments in 1976-80 and
0.71 comments in 1986-90. Thus, in terms of including such important
"qualifying" comments, coverage has improved over the three decades. Percentage
agreement for these two measures was 67%1 and 86% respectively.
One important characteristic of scholarly science writing is that research
results are reported in terms of whether they are in accordance with or deviate
from, previous findings. In addition, suggestions for future research endeavors
are often proposed, based on the obtained results. One consequence of such an
approach is that scientific inquiry is acknowledged to be an ongoing process
where results are continually scrutinized and validated, rather than one where
isolated findings immediately become "facts." Because it has been asserted that
the tendency in the media to treat tentative findings as "breakthroughs" results
in sensationalized accounts of new developments, the inclusion of such
contextual information becomes a critical feature of science news articles.
Among the articles in the three time periods, there was no significant
difference in the tendency to include such contextual information (Scott's pi
=0.80). Of the articles analyzed, 55.55% of those from 1966-70, 51.52% of those
from 1976-80, and 44.64% of those from 1986-90 had any mention of prior and/or
future research studies ((2 [2, N=107]) = 0.44, p>.05). Additionally, there was
no significant difference among the articles in the three time periods in terms
of including comments regarding the limitations or generalizability of the
research findings ((2 [2, N=107] = 0.81, p>.05). In the period of 1966-70,
38.89% of the articles contained such comments, compared with 48.48% of those
articles from 1976-80 and 41.07% of the articles in 1986-90. Thus in all three
time frames, less than 50% of the articles analyzed contained any statements
that express any limitations to the research being reported and the
generalizability of the results to the public. Scott's pi for this measure was
The importance of such information cannot be underestimated, as it is the
omission of this kind of information that may lead to the misemphasis and
sensationalism in science stories so often complained of, as well as to a
misunderstanding of any implications the results may have for the general public
(Freimuth, Greenberg, DeWitt, & Romano, 1984; Greenberg, Freimuth, & Bratic,
1979; Molitor, 1994; Trachtman, 1982). However, these results suggest that this
important feature of science stories, has failed to become more prominent during
the last three decades, at least in the three newspapers analyzed.
Research Question Three: The Presence of Methodological Citations
A one-way ANOVA was performed to determine if differences exist in the adequacy
of methodological citations, as a function of the number of words employed,
appearing in science news articles.. Results indicate that there is no
significant difference among the articles in the three time periods in the
number of words utilized in such methodological descriptions (F [2, 104] = 0.86,
p>0.05). The mean was 30.50 words for those articles from 1966-70, compared to
19.09 words for articles from 1976-80 and 22.25 words for those articles from
1986-90. It is interesting to note that only the mean from the earliest time
period analyzed meets the 30 word operationalization employed as an "adequate"
description (Caudill & Ashdown, 1989). The means from the two later periods
fell well below this standard. Percentage agreement for the number of words in
the methodological descriptions was 97%.2
Furthermore, there was no significant difference ((2 [2, N=107] =2.00, p>0.05)
among articles in the three time periods in terms of including such "adequate"
research descriptions. Of the articles analyzed, only 44.44% of those from
1966-70 contained descriptions of 30 or more words, compared with 33.33% of
those in 1976-80 and 26.79% in 1986-90.
Methodological descriptions were also coded to see if method specific terms
(i.e., experiment, survey, etc.) or method-non-specific terms (i.e., study,
research, report, etc.) were used to characterize the research endeavor. There
was no significant difference ((2 (4, N=107) =7.63, p>0.05) among the articles
from three time periods in their utilization of such terms. That is, the
majority of articles in all three time frames characterized research designs
primarily using method nonspecific terms instead of method specific terms (50%
versus 38.89% for 1966-70; 75.75% versus 18.18% for 1976-80; 75.00% versus
12.50% for 1986-90). Scott's pi for this measure was 0.80.
Overall, the results show that, at least in the three major daily newspapers
analyzed, newspaper coverage of science over the last three decades does not
differ substantially in terms of the range of topics covered, as well as
information that has been both included and omitted from science news accounts.
Although science articles represent only a small percentage of the total number
of articles in the newspaper, this percentage has steadily increased with each
time period.. This increased coverage by newspapers might be in response to, as
well as reflect, the public's growing interest in science (Miller, 1986; Nunn,
1979). In terms of the diversity of coverage, however, there was little
difference in the range of topics covered in each of the three periods. Rather,
the range of topics covered was very similar in each of the time frames studied,
with a large percentage (i.e., 72-75%) of the articles covering medicine and
health related issues. Natural and physical science accounted for 17-25% of
articles in each of the time frames. Technology was clearly the least covered
category, in all three time periods.
In terms of the comprehensiveness of science articles appearing in the three
newspapers analyzed, it appears that coverage in general has not become more
rigorous with the passage of time. While a number of variables were coded for
in this study, it seems especially significant that two features recognized as
crucial to any complete journalistic account of science, that is contextual
factors and methodological details, are still frequently omitted.
The fact that there is no significant difference between articles in the three
time periods in terms of the inclusion of these two types of vital information
can be viewed as problematic. Indeed, Burnham (1987) has argued that the mass
media tends towards a "bits-and-pieces" approach to popularizing science,
presenting uncontextualized facts and paying little attention to the process of
scientific research. This apparently has not changed significantly over the
three time periods analyzed, as these results suggest that newspapers still tend
to report primarily on the findings of scientific research without mention of
relevant contextual factors. This may be cause for concern considering it has
been asserted that a "contextual approach to science communication has never
seemed more important for developing public support for and intelligent use of
science" (LaFollette, 1995 p. 236).
It is also noteworthy that in over three decades of coverage, at least in the
three newspapers analyzed, there has been little change in the inclusion of
methodological details. This seems especially significant when one considers
that a survey of scientists conducted over twenty years ago (Tankard & Ryan,
1974) listed the omission of information about the research method as the most
serious problem with newspaper accounts of science. Based on the results of
this study, this problem has not been remedied.
According to Evans, Krippendorf, Yoon, Posluszny, and Thomas (1990), the image
of science presented in the press is one of a "disembodied enterprise," with
accounts containing very little discussion of research procedures, reported
findings rarely linked to other findings, and limitations of those findings
rarely noted. This study suggests that this has been the case during the last
three decades of science news reporting. While there has been a movement among
newspapers to include science and health related information as part of regular
news coverage, it is surprising to see so little change in the way science has
been and continues to be reported.
Obviously it is hard to generalize about trends in all newspaper coverage of
science based on this limited sample of articles from only three newspapers.
And while admittedly it is difficult to assert with definite certainty that the
trends uncovered in this study can be generalized to coverage in all newspapers,
the results do suggest that while there has been an increase in the amount of
science coverage, there has not been a concomitant improvement in the
completeness and comprehensiveness of this coverage.
It would be interesting if additional analysis on an expanded sample of
newspapers and articles uncover the same results as this study. Additionally,
research into the effects of specific article features on readers may have
important implications for journalistic coverage of science. That is, while
many studies stress the importance of including such pertinent information as
discussions of methodology and qualifying statements, few have looked at how the
inclusion of such components affect individuals' understanding of the
information and issues involved, as well as their perceptions of the scientific
process. Research into science news may need to not only analyze the content of
these journalistic accounts but also look at the potential effects of these
features on readers' in order to uncover what makes a truly accurate, complete,
and effective science news story.
News coverage of science has become increasingly important, as it can influence
both the public's knowledge and attitudes towards science. There is no question
that science continues to make local and national headlines, with newspapers
carrying reports on the latest developments in medical research, to news on the
environment, to the latest in computer technology. While the public has shown a
growing interest in science and science policy (Miller, 1986; Nunn, 1979), many
individuals lack a functional knowledge of these issues. Indeed Miller (1986)
asserts that any measures that can be taken to raise the level of scientific
literacy and to foster informed and intelligent participation in science policy
issues "will improve the quality of both our science and technology and our
political life" (Miller, 1986). It is here, then, that the press could play an
important role. That is, according to Nelkin (1987a)
The press should provide the information and the understanding that is necessary
if people are to think critically about decisions affecting their lives. For
most people the reality of science is what they read in the press...Good
reporting can be expected to enhance the public's ability to evaluate science
policy issues and the individual's ability to make rational choices; poor
reporting is cause for alarm (p. 2-3).
However, it has been asserted that the tendency to deliver scientific research
as "important snippets of news" (Burnham, 1987), omitting many important details
in these accounts, results in both a trivialization and misunderstanding of
specific research results, as well as science in general. While it remains to
be seen if this is actually the case, it appears that, at least over the last
three decades in these three major daily newspapers, science news reporting has
not taken the necessary steps to improve the comprehensiveness of these
1) Work is currently underway to improve the percentage agreement on this
2) Percentage agreement was calculated for the overall subsample. However, if
you examine each article separately, percentage agreement for this measure is
considerably lower on average (only 45%). Work is currently being done to
improve the reliability of this measure.
Acknowledgment: The author wishes to thank a number of people who provided
valuable assistance throughout the course of this project: Perry J. Pellechia,
Department of Chemistry, Purdue University for writing the computer program used
to generate the dates for the constructed week sample; Josh Schaefer for his
time, diligence, and patience in coding newspaper articles; and finally
Professors Glenn G. Sparks and Katherine E. Rowan, Department of Communication,
Purdue University, for all of their guidance, advice and support in completing
Bader, R.G. (1990). How science news sections influence newspaper science
coverage: A case study. Journalism Quarterly, 67, 88-96.
Bodde, T. (1982). Biologists and journalists: A look at science reporting.
Bioscience, 32, 173-175.
Borman, S.C. (1978). Communication accuracy in magazine science reporting.
Journalism Quarterly, 55, 345-346.
Burkett, W. (1986). News reporting: Science, medicine, and high technology.
Ames, IA: The Iowa State University Press.
Burnham, J.C. (1987). How superstition won and science lost: Popularizing
science and health in the United States. New Brunswick: Rutgers University
Caudill, E. & Ashdown, P. (1989). The New England Journal of Medicine as a news
source. Journalism Quarterly, 66, 458-462.
Cole, B.J. (1975). Trends in source and conflict coverage in four metropolitan
newspapers. Journalism Quarterly, 52, 465-471.
DiBella, S.M., Ferri, A.J., & Padderud, A.B. (1991). Scientist's reasons for
consenting to mass media interviews: A national survey. Journalism Quarterly,
Dunwoody, S. & Scott, B.T. (1982). Scientists as mass media sources.
Journalism Quarterly, 59, 52-59.
Dunwoody, S. (1982). A question of accuracy. IEEE Transactions on Professional
Communication, PC25 (4), 196-199.
Dunwoody, S. (1986). The scientist as a source. In S.M. Friedman, S. Dunwoody,
& C.L. Rogers (Eds.), Scientists and journalists: Reporting science as news (pp.
3-16). New York: The Free Press.
Elliott, W.R. & Rosenberg, W.L. (1987). Media exposure and beliefs about
science and technology. Communication Research, 14, 164-188.
Evans, W.A., Krippendorf, M., Yoon, J.H., Posluszny, P. & Thomas, S. (1990).
Science in the prestige and national tabloid presses. Social Science Quarterly,
Freimuth, V.S., Greenberg, R.H., DeWitt, J., & Romano, R.M. (1984). Covering
cancer: Newspapers and the public interest. Journal of Communication, 34,
Glynn, C.J. (1985). Science reporters and their editors judge "sensationalism."
Journalism Quarterly, 62, 69-74.
Glynn, C.J., & Tims, A.R. (1982). Sensationalism in science issues: A case
study. Journalism Quarterly, 59, 127-131.
Goldstein, J.H. (1986). Social science, journalism, and public policy. In J.H.
Goldstein (Ed.), Reporting science: The case of aggression (pp. 1-9). Hillsdale,
Goodell, R. (1985). Problems with the press: Who's responsible? Bioscience,
Gorchow, S (Summer 1990). Preliminary reports: How journalists interpret early
findings. Sciphers, 11 (1) 1-2.
Greenberg, R.H., Freimuth, V.S., & Bratic, E. (1979). A content analytic study
of daily newspaper coverage of cancer. Communication Yearbook, 3, 645-654.
Hinkle G. & Elliott, W.R. (1989). Science coverage in three newspapers and
three supermarket tabloids. Journalism Quarterly, 66, 353-358.
Howard, H., Blick, E. & Quarles, J. (1987). Media choices for specialized news.
Journalism Quarterly, 64: 620-623.
Klaidman, S. (1990). How well the media report health risk. Daedalus 119 (4),
LaFollette, M. (1995). Editorial -- wielding history like a hammer. Science
Communication 16, 235-241.
Lewenstein, B.V. (1987). Was there really a popular science "boom?" Science,
Technology, & Human Values, 12 (2), 29-41.
Lundberg, L.J. (1984). Comprehensiveness of coverage of tropical rain
deforestation. Journalism Quarterly, 61, 378-382.
Mazur, A. (1981). Media coverage and public opinion on scientific
controversies. Journal of Communication, 31, 121-127.
McCall, R. (1988). Science and the press. American Psychologist, 43, 87-94.
Meadows, J. (1986). The growth of science popularization: A historical sketch.
Impact of Science on Society, 144, 341-346.
Millar, R. & Wynne, B. (1988). Public understanding of science: From contents
to processes. International Journal of Science Education, 10, 388-398.
Miller, J.D. (1986). Reaching the attentive and interested publics for science.
In S.M. Friedman, S. Dunwoody, & C.L. Rogers (Eds.), Scientists and journalists:
Reporting science as news (pp. 55-69). New York: The Free Press.
Molitor, F. (1994). Accuracy in science news reporting by newspapers: The case
of aspirin for the prevention of heart attacks. Health Communication, 5,
Moore, B. & Singletary, M. (1985). Scientific sources' perceptions of network
news accuracy. Journalism Quarterly, 62, 816-823.
Nelkin, D. (1987a). Selling science: How the press covers science and
technology (pp. 2-3). New York: W.H. Freeman.
Nelkin, D. (1987b, Sept./Oct.). The culture of science journalism. Society,
Nunn, C.Z. (1979). Readership and coverage of science and technology in
newspapers. Journalism Quarterly, 56, 27-30.
Ogles, R.M. (1985). An overview of content analysis for communication research.
Unpublished manuscript. Department of Communication, Purdue University, West
Lafayette, IN 47907
Patterson, J. (1982). A Q study of attitudes of young adults about science and
science news. Journalism Quarterly, 59, 406-413.
Perlman, D. (1974). Science and the mass media. Daedalus, 103, 207-222.
Pulford, D.L. (1976). Follow-up study of science news accuracy. Journalism
Quarterly, 53, 119-121.
Rich, J.T. (1981). A measure of comprehensiveness in newsmagazine science
coverage. Journalism Quarterly, 58, 248-253.
Ryan, M. (1975). A factor analytic study of scientists' responses to errors.
Journalism Quarterly, 52, 333-336.
Scientists' Institute for Public Information (1986). Now in 66 dailies:
Newspaper science sections spreading nationwide. SIPIscope, 14, 1-17.
Scientists' Institute for Public Information (1993). Science news: What does
the public really want? SIPIscope, 20, 1-10.
Singer, E. (1990). A question of accuracy: How journalists and scientists
report research on hazards. Journal of Communication, 40, 102-116.
Steeple, G.H. III (1981). Content analysis. In G.H. Stemple III & B.H. Westley
(Eds.). Research methods in mass communication. Englwood Cliffs, NJ: Prentice
Sullivan, D.F. (1985). Comprehensiveness of press coverage of a food
irradiation proposal. Journalism Quarterly, 62, 832-837.
Tankard, J.W. & Ryan, M. (1974). News source perceptions of accuracy of science
coverage. Journalism Quarterly, 51, 219-225.
Tichenor, P.J., Olien, C.N., Harrison, A., & Donohue, G. (1970). Mass
communication systems and communication accuracy in science news reporting.
Journalism Quarterly, 47, 673-683.
Trachtman, L. (1983). Informing the public: Why bother? The case against. Phi
Kappa Phi 63, 6-8 [reprinted from SIPISCOPE, published by the Scientists
Institute for Public Information 10, 5-7, 1982.].
Total number of news articles, science articles, and percentage of science
Total science articles
Percentage of science articles as % of all articles
Categorization of science articles
Medicine & Health
Comprehensiveness measures for individual science article
% identifying source of research
% identifying researcher(s)
% identifying researcher(s)' institutional affiliation*
Mean comments from researcher
Mean comments from other scientists (favorable)
Mean comments from other scientists (critical)*
% articles citing prior research
% articles citing limitations in research
Methodological citation measures
Mean words in description of method
% articles having "adequate"
(i.e. >30 words) description
% articles using method non-specific terms to describe research
% articles using method specific terms to describe research
Trends in newspaper coverage of science over three decades: A content analytic
study Page: 35