Content-Type: text/html

Context in print and online environmental articles


Ryan Randazzo, Reno Gazette-Journal

and Jennifer Greer, Reynolds School of Journalism
University of Nevada-Reno



Contact: Jennifer Greer
Mail Stop 310
University of Nevada-Reno
Reno, NV 89557

Office: (775) 784-4191
Home: (775) 746-3892
<a href="/cgi-bin/wa?LOGON=A3%3Dind0209C%26L%3DAEJMC%26E%3D8bit%26P%3D829525%26B%3D--%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D_2738790%253D%253D_%26T%3Dtext%252Fplain%3B%2520charset%3Diso-8859-1" >[log in to unmask]</a>



Submitted to the Science Communication Interest Group of the Association for Education in Journalism and Mass Communication for the 2002 conference, Miami, FL.
Abstract

Context in print and online environmental articles.

A content analysis of environmental articles in leading U.S. dailies revealed that few of nine types of context examined were included in the printed version of the articles. None of the newspapers were using the potential of the Internet consistently to add more context to their environmental articles online.  In fact, fewer contextual elements appeared online than in the print versions.

In recent decades, Americans have become increasingly concerned with the interaction of our society and the environment (Askari, 1995) and have looked toward the news media to keep them informed on that interaction.  About 81% of Americans depend on the media for environmental news, according to one survey (Kennett, 1993). To quench the public thirst for information, the news media have increased the amount of coverage of environmental issues (Foundation for American Communications [FACS], 1993). With the dawn of the "environmental decade" in the 1970s came the formation of environmental journalism with the purpose of conveying environmental issues to the public (Shoenfeld, 1980).  Since then, environmental journalism has established itself as a legitimate field of reporting with a large audience.  Some large newspapers now employ more than half a dozen writers to cover the environment (Chepesiuk, 1993).
Although coverage of environmental issues has been popular for more than three decades, critics say this coverage is wrought with problems. One common complaint is that the media don't put issues into proper context; quotes are taken out of context, and current work is not related to other research (Walters, 1996).  The World Wide Web gives news organizations the potential to correct this flaw (Tenenbaum, 1997).  While some news organizations use the same articles in their print and online versions, others have taken advantage of the Internet's potential to provide more information to readers.  In addition to the ability to run longer, more in-depth articles, some online versions of newspapers like The New York Times and The Christian Science Monitor link articles to archived material on the same topic (Fredin, 1997).  Some organizations use links to other information sources, such as government databases, that add greater context, depth, and texture to online stories (Pavlik, 199
7).  Other news organizations have included court transcripts, search warrant documents, wiretap transcripts of drug dealers, and other features online that would be impossible in print (Pavlik, 1997).  To compete as a news medium, newspaper journalism on the Internet must fully exploit the medium's basic properties (Fredin, 1997).
This study examines what types of context the leading U.S. daily newspapers include in their environmental coverage and whether they are using the potential of the Web to improve context in their coverage. Print and online environmental articles appearing in The New York Times, The Los Angles Times, The Washington Post, USA Today, and The Christian Science Monitor are examined.  Additionally, differences in context among newspaper titles are analyzed.
Literature Review
In 1947 the Hutchins' Commission (The Commission on Freedom of the Press) called for a new kind of journalism.  The first duty of newspapers was "to give an accounting of the day's events in a context that gives them meaning," according to recommendations the commission published (Commission for Freedom of the Press, 1947, p. 20).  More than 50 years later, critics contend that the media still fail to put issues in proper context (Griffin, 1992).  These criticisms are especially loud for environmental reporting.  Trying to report complex, controversial, scientific issues well is difficult (Cowen, 1985; Garrison, 1992), and the limited space of the print medium exacerbates these problems.  To report on the environment, journalists must educate themselves regarding controversial issues that even experts do not often completely understand (Fischoff, 1996).
A 1993 survey of journalists showed that only 3% considered current environmental coverage to be "very good," and 40% rated it as "good,"  (FACS, 1993).  Incorporating more context into stories regarding the environment is among the improvements sought by experienced environmental journalists (Cowen, 1985).  Salomone, Greenberg, Sandman, & Sachsman, 1990). Giving background information that helps readers put issues in context in relation to their lives is an important part of journalism.  If journalists fail to do so, readers are not likely to understand events and grasp their significance and potential consequences (Griffin, 1992).
The importance of context

Context, defined as coherent analysis that helps makes complex topics understandable, is necessary in newspaper reporting because of competition from other media such as television and radio, which offer little context (McCleneghan, 1997).  Newspapers must add more perspective, analysis, and interpretation to the news to be different from other media by adding more depth to the news (McCleneghan, 1997). Because of the limitations of the print medium, newspapers struggle to compete with the timeliness of television and radio.  What newspapers can offer that those mediums cannot or commonly do not is analysis and interpretation. A good analytical news story creates understanding by showing that what happens in one time or place is related to what happens in other times and places (Hart, 1994).  Context is important because research has shown that readers learn more from articles with background and context included (Griffin, 1992).  Readers can miss out on the relative importance of
 the issue if an article on a scientific subject fails to include some background information.  Articles need to contain full perspective and background of an issue or they may harm the credibility of scientists and journalists (Burkett, 1986).  Omitting information from an article may lead to distortion of the issue (Burkett, 1986), which could in turn lead to a biased representation of the issue.
Observers offer various causes for the lack of context.  First, in contrast to the deliberative nature of scientific inquiry, newsrooms are fast-paced workplaces where information is processed and passed along as quickly as possible.  Because of these differences, information passed along by the media in news stories fails to fully explain issues or at least explain them in a manner acceptable to scientists (Salomone, 1990).  A second cause for the lack of context is journalists' poor sense of scientific information (Keating, 1997).  Once reporters have been on the environmental beat for a significant amount of time, they tend to move on.  New, lesser-experienced journalists who are unable to provide context and competent reporting replace them (Keating, 1997).  And third, even experienced journalists are constrained from writing long articles with ample context by the cost of newsprint, which accounts for 20% of a newspaper's expenses and affects the profitability of newspapers (
Vivian, 2001).
How context can be included in articles

Hart (1994) outlines three ways to include context in articles.  First, journalists should make historic connections to events, placing the current issue in historical context.  Second, writers can draw other, similar examples, the same way a reviewer compares movies.  Third, reporters can add averages or statistics that describe the overall situation.  Graphics can also add context to articles.  Topics as wide ranging as space shuttles to bypass surgery are all made easier to understand with the inclusion of photographs, schematics, and illustrations (Garrison, 1992).  Repeatedly giving the contextual information of an article to the reader helps people make sense of complex issues.  Research has shown that readers learn the most from articles with repetitive contextual information interwoven in the article and also written in a separate text box accompanying the article (Griffin, 1992).  These techniques helped people understand foreign news better than if they read articles with no background information or articles with contextual information in text references or pull out graphics alone (Griffin, 1992).
The Internet and Journalism
Science, health, finance, and technology draw the most attention from people going online for news (The Pew Center, 1998a), which may mean environmental journalism has a potential to draw a large readership online.  The Internet holds the promise of allowing print journalism to break free of its traditional limitations, which in turn could lead to more context online (Pavlik, 1997).  The Internet's strengths include the immediacy and ability to break and update news as it occurs, the nearly limitless news hole, the ability to include references and links to more information, the interactivity available for readers, and the ability to archive information. Each of these is examined in turn.
Immediacy. The potential to break stories immediately sets the Web apart from print media.  So many major stories have been broken on the Internet before they have come out in print that news of such occurrences has become redundant (Fitzgerald, 1998).  Lasica urges frequent updates: "Online news should be about getting current news and information to people when they want it or need it, not when it's convenient . . .What matters to readers is staying on top of local and national events . . . adding context, background, balance, and perspective to events as they unfold" (1997, p. 48).  The Why Files, a science news site based at the University of Wisconsin-Madison, did just that in its 1996 coverage of the wildfires across the United States. A linked map of fire-burned areas changed daily as the fire-damaged areas grew (Tenenbaum, 1997).
Unlimited space.  Adding words or other information online is virtually free.  Salon.com once ran a 6,000-word article that was rejected by The New Yorker as too lengthy (Lasica, 1998).  The nearly unlimited capacity to store and print information online can be especially helpful in conveying scientific issues because graphs and diagrams make scientific subjects much easier for the public to understand (Garrison, 1992).  An example of the larger news hole being put to use is the addition of the full text of court depositions and other official documents online to enhance news articles (Fitzgerald, 1996).  And some outlets, such as CNN, are offering up to 10% original content not available anywhere else in print or through television (Pavlik, 1998). A 1999 survey of online editors found that some used the limitless space of the Web to give expanded news coverage. But by a 3 to 2 margin, online editors showed preference for limiting the amount of information provided online compared
 to the print product (Peng & Tham, 1999).
References and links.  Occasionally Web articles contain reference material and links with more information for readers.  These outside links can be to other articles on the same topic or they can be to outside Web sites that provide a perspective to the reader unavailable in a print version of the article (Pavlik, 1997).  However, providing outside links does not necessarily mean additional information in an article.  A 2000 study found that 25% of print newspaper articles appearing in major newspapers contain URLs, or Web addresses (Trumbo & Trumbo, 2000).  This study found that, rather than adding context to stories by giving URLs to information sites, most print versions of newspapers seemed to be providing URLs to commerce sites rather than government or educational sites aimed at providing more background material (Trumbo & Trumbo, 2000).  Peng and Tham (1999) found that nearly all online newspapers provide links to outside sources, yet fewer than half linked to other news sites.
Interactivity.  While traditional newspapers can offer a means of projecting different opinions through letters to the editor, the Internet offers an instant forum for communication.  The interactive capabilities of the Internet allow readers to communicate with one another in chat rooms immediately after reading an article, should the Web editors provide that service.  After the school shooting in Littleton, Colorado, journalist Jonathan Dube spent hours chatting online with more than 1,600 readers.  He was able to convey facts that hadn't fit into his articles and, in effect, created a new article that the readers helped shape with their questions (Dube, 1999).
Audio and Video.  Online newspapers do not have to rely on the traditional format of newspaper articles because they are able to use new technologies like audio and video.  Real-time audio and video technologies have been available since 1997 when RealNetworks introduced "streaming" technology (Williams, 2000). Online newspapers have made limited use of these technologies.  Gubman and Greer (1997) found that only 14.5% of online newspapers used audio, video or animation in their news sections.  The Washington Post used these capabilities when reporting on Hurricane Bonnie (Featherly, 1998), and The Los Angles Times has used video and audio technologies to enhance articles on other issues (Phipps, 1998).  Editors at USA Today and other large newspapers have created content-exchange deals with television stations owned by the same company to have access to video footage for their Web sites (Williams, 2000).
Archiving.  Another way in which the Internet may help provide context to readers is through archiving information.  Online editors can link all of the articles the newspaper has written on a particular subject from the most current article on that subject. A 1998 survey found that people go online for news for exactly this reason, to search for news about a particular topic.  This was the third most common reason people looked online for news behind finding information unavailable elsewhere and convenience (The Pew Center, 1998b).  Archiving articles may be a good idea for online editors.  If consumers can't find the information they want from local newspapers online, they will go to other news sites like MSNBC or CNN (Lasica, 1997).  A survey by the Pew Research Center (1998b) found that readers were already beginning to turn to other news sites, with the national broadcast station Web sites being more popular than national newspaper Web sites.
Research questions
With all of the potential, it is still unclear if and how newspapers are using the Web to convey more scientific information to readers.  Factors such as time and cost-benefits may prohibit print media from using the Web to add more context to environmental issues.  Singer (1997) found that coverage of the 1996 presidential election by Denver's two daily newspapers available online actually was less than that provided in print in an analysis of two major newspapers.  One of the newspapers only provided 6% of the election coverage online that it did in print, while the other published about 15% of the coverage on its Web site (Singer, 1997).  One explanation given by managers at the newspapers for this is that the election was national news available many places on the Internet and that their niche was local news.  The study also found that the articles that did appear online were rarely different than the print version.  Thus, despite the opportunity to run a longer version of the
 articles, they had already been cut by the editors and went into the online version as they appeared in print (Singer, 1997).
This study asks if the same patterns are found in environmental coverage. Are newspapers running longer articles, adding reference material, adding graphics and images, updating material, immediately breaking news, adding external links, and using interactivity to improve context in environmental news coverage?  The following research questions are posed to shed light on the issue:
* RQ1: How much and what type of context do leading U.S. daily newspapers provide in their print versions of environmental articles?
* RQ2: Does context in print articles vary by newspaper title?
* RQ3: Are leading U.S. daily newspapers taking advantage of the Internet's capabilities to provide more context in online versions of environmental articles than in the print versions?
* RQ4: Does the difference between print and online context vary by newspaper title?
* RQ5: How does presence of an adversarial frame relate to the amount of context provided?


Method
        A content analysis was conducted to get a picture of what is typical in both print and Web coverage of the environment in leading U.S. dailies.
Population and sample
As of early 2001, more than 3,400 U.S. newspapers had online versions of their newspapers (Newslink, 2000).  Most of the articles on the Web sites are exactly the same as those that appeared in the print version, (Fredin, 1997; Houston, 1999) a publishing method known as "shovelware" (Thalhimer, 1994).  Large national newspapers are more likely to experiment with different content on their Web version than the print version (Fredin, 1997).  Gubman and Greer (1997) found that only 15.7% of newspapers with an online version used linked boxes and non-linear storytelling methods on the articles appearing on the Web.  They also found that only 14.5% used multimedia such as audio or video in their news articles, although multimedia was being used on their sites in advertising.  This study found that large newspapers were more likely than small newspapers to use these features.
Singer, Tharp, and Haruta (1998) showed that larger newspapers were more likely to have separate staff members for online operations.  Cole (1975) selected from the nation's largest newspapers to draw a sample of science articles under the presumption that the larger newspapers have the capability to hire science writers.  In addition, Trumbo (1995) used a sample of leading newspapers because their articles are likely to be reprinted in many other smaller newspapers owned by the same company, increasing the influence of the large, prestigious newspapers.  For these reasons, this study used leading national newspapers to determine if differences exist in articles between print and online versions, and to examine differences among newspapers.
The four opinion-leading newspapers in the country are The New York Times, The Wall Street Journal, The Washington Post, and The Los Angles Times (Krantz, 1994).  USA Today has a larger circulation than any of these (Vivian, 2001).  Because The Wall Street Journal is a business newspaper rather than a general interest national newspaper like the others (Vivian, 2001), USA Today was used in this study in place of The Wall Street Journal.  The influence of USA Today is strong because of its size and because of the distribution of its stories in other Gannett newspapers (Vivian, 2001).  The Christian Science Monitor, another national newspaper, also was chosen because of its depth in science and environmental coverage.
The most efficient way to get a representative stratified sample of the content of a daily newspaper for a given year is to gather two constructed weeks of data (Riffe, Lacy, & Fico, 1998).  Cole (1975) used three constructed weeks from each of three years--1951, 1961, and 1971--to measure trends in science news coverage.  However, McMillan (2000) found that most Web content analyses emphasize a particular time frame because the changing site content necessitates rapid data collection. The fleeting nature of the Web and the lack of uniformity in how newspapers archive their content, necessitated a fixed time period of data collection for this study, which was the three week period from February 22 to March 14, 2001.
Lacy, et al. (1991) used the type of purposive sample this study will use to compare conflict coverage among 12 large circulation newspapers in a study similar to this one.  This method was appropriate because, although not random, it was intended to compare certain context in coverage among newspapers and between print and online editions.  Lacy, et al. (1991) compared conflict coverage among the 21 newspapers with a mean of 16 articles from each newspaper.  This study attempted to collect at least 40 articles from the print version of each newspaper, enough from each title to perform meaningful statistical analyses.
Unit of analysis
Each print version of an environmental article collected during the sample period was one unit of analysis and the Web version of the article was another.  Each version of an article got its own code sheet.  The entire article was analyzed because of the way variables were measured. All environmental articles identified in the print version of the newspaper were included in the study.  If the article was subsequently identified on the newspaper's Web site, that version was included in the study as well.
Any environmental article with a byline in the main news section, business section, local news section, lifestyle section or science section of the newspaper was used.  Articles in the advertising sections of the newspapers were not used.  As with other newspaper research, opinion/editorial pieces were not included (Li, 1999).  The study analyzed only articles produced by the newspaper's staff and not those by a wire service used in the newspapers.  This decision was based on the assumption that online editors would be more likely to invest resources in improving articles produced by their own staff and not those from a service.  Only articles with bylines were used, as briefs and other short articles are frequently taken from wire services.  Even if the staff writes the briefs, the articles are unlikely to be enhanced for the Web.
For the this study, environmental articles were those that regarded any of the following:  air pollution, water pollution, solid waste, deforestation/logging, endangered species, population growth, government departments such as the Environmental Protection Agency or the Department of Interior including state-level agencies, natural resources, urban sprawl, animal rights, invasive species, biotechnology, genetically engineered food, ecology, mining, global warming, wildlife, wilderness areas, national parks, national forests, environmental racism, eco-feminism or any other topic germane to the environment.  To be used in the study, the article had to have a major focus, which revealed itself in the headline or first five paragraphs, (Boyle, 1995) on one of the topics listed above. Articles that only mentioned briefly environmental issues were not included.
During the collection period, two coders tracked the print newspapers for any environmental articles. Next, coders looked for articles on the respective newspaper's Web site.  Only environmental articles on the Web site that had a print counterpart were coded.  During the collection period, only two articles were found online and not in print.  One was an advertorial; the other included a page number for the print edition, suggesting that it was inadvertently left out of the print version. Reliability between the two coders on 10% of the stories (N = 41) ranged from 87% (N = 2) to 100% (N = 16), with an average of 96%.
Context
The context errors that scientists note the most in reporting of their work are quotes taken out of context, lack of reference to other research, and a lack of perspective of the issue (Walters, 1996).  Because interviews with quoted individuals would have been required to determine whether quotes in the articles were used in proper context, that criteria of context was not measured in this study. Past research has conceptualized and measured context in several other ways.  This study borrows nine of these, each coded as present (1) or not present (0). These measures are summarized below.
* Reference to research the article is based on.  One study found that 40 percent of science articles are significantly different than the research on which they are based (Walters, 1996).  A reference to research appearing anywhere in the article including photograph captions and cut-lines was coded as present.  Links to the original research in the Web version also were coded as present.
* Historical context.  Hart (1994) suggested other ways to add context to news, including adding historical information.
* Comparison using similar examples.  Hart (1994) suggests writers draw other, similar examples to issues to add context, the same way a reviewer compares movies.
* Reference to other research.  This study also coded for reference to other research that pertained to the topic in each article.
* Sources of more information.  Whereas print newspapers are limited in the amount of information they provide, the Web allows more information to be included and linked to. The presence of sources where readers could find more information, such as telephone numbers or Web addresses, was recorded.
* Three types of visual elements.  Readers can understand science better with the aid of photos, diagrams, maps, and other visual aids (Tenenbaum, 1997; Griffin, 1992; Garrison, 1992).  Three variables were examined in this category: 1) maps, 2) photographs, 3) diagrams or illustrations. In addition, the total number of visual elements accompanying a story was recorded.
* Audio and video.  Current technology allows for visitors to watch videos or listen to audio clips provided by Web sites (Williams, 2000).  These formats may enhance environmental issues in the same manner as still visuals.  Any audio or visuals included in a story were recorded. Because audio and video are not possible in print editions, this variable was automatically coded as not present in print articles.
Three other variables also were examined.  First, to compare overall context, the nine variables above were summed to produce a context score.  Second, article length was measured in number of paragraphs. If a difference was found between the print and online editions, the articles were examined word-by-word to determine whether actual differences existed in the story length.  Finally, the presence of an adversarial frame was noted.  An adversarial frame is dichotomy or duality in an article.  The most obvious feature of this is representation of issues in terms of two distinct, mutually exclusive camps (Karlberg, 1997).  Cole (1975) coded science newspaper articles for the presence or absence of conflict.  If there was a mention of at least one antagonist or contradictory, opposing sides of an issue or a reason for conflict, then the article was coded as having conflict.  This study coded for any presence of an adversarial frame in the same manner, either present or not present.
Results

During the sample period, 215 environmental articles were coded from the print versions of the five newspapers.  Of the articles, 96% (N  = 206) appeared on the Web.  Only 4% (N  = 9) of the articles appeared in print only, adding to a combined total of 421 coded articles from print and the Web.  The Los Angles Times had the most articles during the period and The Christian Science Monitor had the least.  The Christian Science Monitor and The Washington Post transferred 100% of the articles in the sample to the Web; USA Today transferred the smallest percentage (see Table 1).

Table 1.  Newspaper articles online and in print.
Newspaper
Print
Online (percent transferred online)
Total
USA Today
Los Angles Times
New York Times
Christian Science Monitor
Washington Post
Total
44
58
50
19
44
215
39 (89%)
56  (96%)
48  (96%)
19  (100%)
44  (100%)
206  (96%)
83
114
98
38
88
421

RQ 1: How much and what type of context do leading U.S. daily newspapers provide in their print versions of environmental articles?
Of the eight variables related to context that could be included in a print article, historical information was the most frequently occurring in the sample.  The least likely context variable to be included was maps, closely followed by sources for more information (see Table 2).  Audio and video, of course, were not present in any of the articles, but this variable included for later comparison to online articles.
Table 2.  Types of Context included in print environmental articles.
Context variable.
Percent of articles containing this form of context.
Historical information
Photos
Similar examples
Reference to research
Reference to other research
Diagrams/illustrations
Sources of more information
Maps
74% (N = 159)
49% (N = 106$„Ž E CN#p8„`_0‘ ¤Ÿr``l_PŸA_ai>0Ÿ@j_, ._D   %… _E"_DPDø_lHp!ý"p____ ZX@<_aD_2AOUO_HO\Eš_X]Y>š_XXU\_XUH_HUUXs[s[_EHY[U_Hs[_HššUX>HUU_^_\sXX>\E_\U[_[_]\_XUK_HOUš_HUY[[š_]XO[H_[_UH_>UH YX[s[_EH\_XUHUU_Z[_Y>U_HU_HUU_^_\sXX>\EEKYX[s[_E]UU_Z[_Y[H[_]Y[E[_sY[EO_s[Y\UHUU_sX_]YH_\>EE\EOUš_H>š_[s[_\_XU\EYX[s[_E^HUY[U>H_[_UH[_OUš_H_>UH E_HYX[_ontext score for all of the print articles was 2.73 (median = 3, mode = 2).
The number of visual elements and paragraphs in the articles also was recorded.  The mean number of paragraphs was 22 (median = 20, mode = 24).  The mean number of photographs per article was 0.83 (median = 0, mode = 0).  The mean number of maps per article was 0.19 (median = 0, mode = 0).  The mean number of diagrams or illustrations per article was 0.42 (median = 0, mode = 0).
RQ 2: Does context in print articles vary by newspaper title?
A one-way analysis of variance (ANOVA) demonstrated a significant difference between newspapers for the number of diagrams and illustrations included in print articles (F (4, 211) = 3.53, p < .01) and for number of paragraphs in print articles (F (4, 212) = 3.58, p < .01).  A slight, but not significant difference was found for the total context score of print articles (F (4,211) = 2.03, p < .10).
Post hoc analyses using Tukey's HSD demonstrated that, for the number of diagrams and illustrations, USA Today (M = 0.95) had significantly more present in articles than the Los Angles Times (M = 0.31, p < .05).  USA Today also had significantly more diagrams and illustrations than The New York Times (M = 0.32, p < .05) and The Washington Post (M = 0.27, p < .023), and slightly more than The Christian Science Monitor (M = 0.16, p < .06).
For story length, post hoc analyses also demonstrated that The Los Angles Times had significantly more paragraphs (M = 27) than USA Today (M = 17, p < .003).  The rest of the newspapers' mean number of paragraphs fell between those scores.
Differences in the total context score were approaching significance between USA Today (M = 3.09) and The Washington Post (M = 2.2, p < .06) with the other three newspapers falling between those scores.
Other differences in the context provided in the newspapers appear when comparing each discrete type of context.  Crosstab analysis displayed a significant difference in references to research included in the print articles of the newspapers ((2(4) = 23.65, p < .001).  USA Today was twice as likely as expected to include a reference to research the article was based on, while the rest of the newspapers were less likely than expected to include this information.
Crosstabs analysis also displayed a significant difference for presence of historical information ((2(4) = 12.21, p < .05).  USA Today was less likely to include this information than expected, while The New York Times and The Washington Post were more likely than expected to include historical information.  The Los Angles Times and The Christian Science Monitor provided this information almost exactly as much as their expected values indicated.
A significant difference in which newspapers included photographs with their print articles also was found ((2(4) = 14.96, p < .01).  USA Today was less likely than expected to include photographs with articles, while The New York Times and The Christian Science Monitor were more likely to include photographs than expected.
Significant differences also emerged for sources for more information ((2(4) = 13.41, p < .01).  USA Today was twice as likely as expected to provide these sources, while The Washington Post was six and a half times less likely than expected to do so.  The other three newspapers provided sources of more information in proportion to what was expected.
RQ3: How does presence of an adversarial frame relate to the amount of context provided?
An independent samples t-test found that the average structural context score for articles in the entire sample written in an adversarial frame (M = 1.84) was significantly lower than for articles not written in an adversarial frame (M = 2.97, t (419) = 6.79, p < .001). Articles with an adversarial frame were significantly less likely ((2(1) = 16.82, p < .001) than those without an adversarial frame to contain a reference to research the article was based on.  Articles with an adversarial frame were also significantly less likely than those without an adversarial frame to have a similar example of the issue used as a comparison ((2(1) = 7.33, p < .01), to have a reference to other pertinent research ((2(1) = 8.78, p < .01), to have diagrams or illustrations ((2(1) = 7.04, p < .01), or to provide sources of more information ((2(1) = 5.67, p < .05).  No significant difference was found between adversarial and non-adversarial articles for historical information, photographs, or maps.
Adversarial and non-adversarial articles did not differ significantly on the number of photographs, number of maps, or number of paragraphs included.  But the number of diagrams and illustrations in adversarial articles (M = .14) was significantly lower than the number appearing in non-adversarial articles (M = .53, t (213) = 2.31, p < .023).
RQ4:  Are leading U.S. daily newspapers taking advantage of the Internet's capabilities to provide more context in online versions of environmental articles than in print versions?
Independent samples t-tests compared the mean number of photographs, maps, diagrams and illustrations, audio and visual elements, and total context scores.  Print articles had a higher average number of photographs and maps than online articles. No other significant differences were found (see Table 3).
Table 3.  Average number of context elements appearing online v. print.
Variable
Mean
Print           Online
Number of photos?
 .83              .511
Number of Maps?
 .19             .072
Number of diagrams?
.42             .46
Number of audio/video?
.00             .05
Number of paragraphs?
21.82             21.85
Total context score
2.73             2.62
                                                1 t (419)  = 3.03, p < .01
                                                2 t (419) = 1.98, p < .05

Crosstabs analyses were run on the discrete context variables to compare each for print and online versions.  Far fewer online articles contained photographs than expected ((2 (1) = 10.82, p < .01).  Online articles also were significantly less likely than expected to include maps ((2 (1) = 9.3, p < .01) and diagrams and illustrations ((2(1) = 5.17, p < .05). Online articles had more sources for more information than expected ((2(1) = 15.65, p < .001).  No other differences were found between print and online versions.
RQ5: Does context online compared to print vary by newspaper title?
Crosstabs analyses were conducted to determine if articles in the five newspapers were more likely to contain any of the nine context variables in their print or online versions.  Independent samples t-tests were conducted between the online and print versions of each newspaper to determine if there was a significant difference in the number of photographs, number of maps, number of diagrams or illustrations, or total context score.
Print articles in USA Today were more likely than online versions to include photographs ((2(1) = 8.81, p < .01).  When USA Today articles had photographs, the number accompanying print articles (M = 0.75) was significantly greater than the number accompanying online articles (M = 0.15, t (81) = 2.4, p < .05).
Print articles in The Los Angles Times were more likely than online articles to contain maps ((2(1) = 8.85, p < .01). Articles in The Los Angles Times had a significantly greater number of maps when they appeared in print (M = 0.79) than when they appeared online (M = 0.64, (112) = 3.07, p < .01).  The print articles in that newspaper also had a significantly greater number of diagrams and illustrations (M = 0.31) than the online articles (M = 0.13, t (112) = 2.03, p < .05).
Online The Washington Post articles included photographs significantly less often than expected ((2(1) = 5.73, p < .05).  Articles in this newspaper had a significantly greater number of photographs when they appeared in print (M = 0.61) than when they appeared online (M = 0.18, t (86) = 2.87, p < .01).  Online articles from The Washington Post were significantly more likely than expected ((2(1) = 14.97, p < .001) to contain sources for more information.
Online The Christian Science Monitor articles were more likely than expected to have sources for more information ((2(1) = 5.73, p < .05).  Online articles in The New York Times also were significantly more likely than expected to have sources for more information ((2(1) = 3.98, p < .05). No other differences were found between print and online for these newspapers.
Summary of findings
Nearly all of the environmental artiales xhat appd`red in privt wbe |ransferred to the Web, witp two newspapers, The Christian Science Monitor and The Washington Post, transferring all of their environmental articles to the Web.
All of the eight possible types of context were found in at least a small percentage of the print articles. On average, articles included less than three of the story components that can add context.  The most common component was historical information, followed by photographs; maps were the least common type of context included.  The mean number of paragraphs per article was 22, and some had more than 90 paragraphs, fairly long for newspaper articles.
The individual newspaper titles varied in the amount of context they provided in their print environmental articles.  Despite having the shortest articles, USA Today had the highest context score and was the most likely to include a reference to research the article was based on. USA Today had more diagrams and illustrations than the other newspapers. While only 15% of the articles provided sources for more information, USA Today was far more likely to do so than the others.  The Washington Post was far less likely to do so. The Los Angeles Times had significantly more paragraphs in environmental articles than USA Today but not the other newspapers.  The New York Times and The Washington Post were more likely than the other newspapers to include historical information.  The study revealed that stories presented with an adversarial frame included less context overall than non-adversarial articles.
None of the newspapers appear to be using the potential of the Internet consistently to add more context to their environmental articles.  In fact, fewer contextual elements appeared online than in the print versions.  This is true only for visual elements, such as photographs and maps, and not for text elements like references to research and historical context.  Because in most cases the text appearing online was identical to what appeared in print, some elements of context appeared online in the same abundance as in print. The only way that more context was included with articles appearing online was if sources of more information not found in the print version were included.  Often, these were links to more information and, in many cases, were links to previous articles published on the site.  Overall, the online versions of articles were more likely to have sources for more information.  Among newspaper titles, The Christian Science Monitor, The New York Times, and The Washin
gton Post articles were significantly more likely to provide sources of information online than in print.
Implications
Because there is no baseline to compare it to, it is difficult to make a statement about the amount of context provided in environmental articles in the leading U.S. daily newspapers.  The average context score of 2.73 for print articles appears low considering that a total score of 8 was possible (with audio and video excluded).  It is hard to imagine an article containing two references to research, a similar comparison, historical information, a map, a diagram, a photograph and a place for readers to find more information without taking up a considerable amount of space in the newspaper and considerable resources from the staff.  To look at this result in the best light is to say that the leading newspapers provide some degree of context in their environmental articles based on what the critics say is missing.  They also provide all types of context but not all in each article.  To look at it another way, it is likely the newspapers could work to include more of these context v
ariables in each article, as the average article is lacking many of the attributes of context.
Frequently included forms of context.  The newspapers are doing a good job of providing context in the form of historical references, photographs, similar examples used as a comparison, and diagrams and illustrations.  The presence of historical references outnumbering other types of context is not surprising.  Reporters may be familiar with an issue because they have written about it before, making it easy to include historical information.  Photography is of course a major part of journalism, so the large percentage of environmental articles that contained photographs is not surprising.  Almost half of the articles compared the issue featured to a similar issue. Nearly a quarter of the environmental articles included a diagram or illustration, most likely a larger percentage than for newspaper articles in general.  As stated above, observers have argued that science-based issues are often best explained with diagrams and illustrations. Editors at leading U.S. daily newspapers se
em to realize this.  The frequent inclusion of diagrams and illustrations may also relate to environmental articles being written as feature articles in special weekly sections.  Some of the articles in the sample were breaking news written on a short deadline.  Others, such as those longer than 90 paragraphs, obviously had a longer deadline allowing more time for visual elements to be prepared.
Infrequently included forms of context.  The newspapers are not often providing context in the form of references to research, sources for more information, or maps.  Maps were found least often but were not absent from environmental articles.  Maps may have been included less often than other visual elements of context because for local environmental issues, the readership is familiar with the area.  The newspapers in the sample are national newspapers though, and from observation, most of the articles examined did not appear to be local issues.  While maps were found less often than any of the other elements, the editors may have special guidelines for including them when appropriate.  If not, the infrequent inclusion of them certainly warrants attention and the creation of such guidelines.
The infrequent mention of sources of more information on the issues raises questions. Are these not included because of space constraints or because editors think readers wouldn't be interested?  In either case, the lack of sources for more information leaves the readers on their own without any guidance as to where to look to learn more about the issues.
References to additional research also are rare, perhaps indicating that many environmental issues reported in leading U.S. daily newspapers are not based on or about scientific research.  This makes some sense, because an environmental article can be about a debate over environmental regulations, in which case other story elements are more pertinent than research.  At some level though, most if not all, environmental issues are based on research.
Differences among newspapers. It was not surprising that USA Today articles contained diagrams and illustrations more often than the other newspapers, or that it had significantly fewer paragraphs in articles than The Los Angles Times. After all, the print version of USA Today is known for these traits.  Brief articles may also explain why USA Today included sources for more information more often than expected; readers are sent to other sources that couldn't be included in the short articles.  What was unexpected was that the total context score for USA Today articles was higher, although not significantly, than all of the other newspapers and that USA Today had more references to the research on which the articles were based.  This result can be viewed at least two different ways.  First, articles in USA Today are more often research-based or second, that the writing in USA Today is less enterprising as it is often based on press releases from research institutions without furth
er interpretation.  The question of which way to view the findings can't be answered without further research, but from observation of the sample articles the reason for USA Today including more references to research appears to be a mix of the two.
Finding that The New York Times and The Washington Post were more likely than expected to include historical information is not surprising, as they appear to only be including more than USA Today.  That The New York Times includes more photographs with environmental articles than the other newspapers may be due to the special weekly science section of that newspaper where most of the environmental articles were found.  The preparation of that special section may allow for more photographs than the other newspapers.
Curiously, articles containing conflict and written in an adversarial frame have a lower amount of context. However, this is likely the reason why The Washington Post had a nearly significant lower context score than USA Today.  The Washington Post had a greater percentage of adversarial articles, and its lower overall context score can likely be attributed to that.
Context online vs. print. The drop in the context score as the articles were transferred to the Web was surprising. Also unexpected was that only 5 of the 206 articles appearing online had audio or video and the online verere already beginning to turn to other news sites, with the national broadcast station Web sites being more popular than national newspaper Web sites.
Research questions
With all of the potential, it is still unclear if and how newspapers are using the Web to convey more scientific information to readers.  Factors such as time and cost-benefits may prohibit print media from using the Web to add more context to environmental issues.  Singer (1997) found that coverage of the 1996 presidential election by De put to use to improve the context in environmental articles to any significant extent.
Finding that The Christian Science Monitor, The New York Times, and The Washington Post are the only three adding sources for more information to their online articles in significant proportions also was unexpected, as adding links to an article is technically simple. Whatever the reason, the editors are passing up an opportunity to give their readers more information in a relatively easy way.
The missed potential for context to be enhanced online might be the result of the economic pressures facing the newspaper industry.  Some media outlets have already faced large cutbacks in their online departments (Barringer, & Kuczynski, 2000).  Additionally, some editors may be hesitant to devote resources to an unprofitable part of the operation.  Finally, the technical limitations also may work against enhancing context.  Adding images, audio or video to a Web page decreases the speed at which the page loads.  The utility of su using interactivity to improve context in environmental news coverage?  The following research questions are posed to shed light on the issue:
* RQ1: How much and what type of context do leading U.S. daily newspapers provide in their print versions of environmental articles?
* RQ2: Does context in print articles vary by newspaper title?
* RQ3: Are leading U.S. daily newspapers taking advantage of the Internet's capabilities to provide more context in online versions of environmental articles than in the print versions?
* RQ4: Does the difference between print and online context vary by newspaper title?
* RQ5: How does presence of an adversarial frame relate to the amount of context provided?


Method
        A content analysis was conducted to get a picture of what is typical in both print and Web coverage of the environment in leading U.S. dailies.
Population and sample
As of early 2001, more than 3,400 U.S. newspapers had online versions of their newspapers (Newslink, 2000).  Most of the articles on the Web sites are exactly the same as those that appeared in the print version, (Fredin, 1997; Houston, 1999) a publishing method known as "shovelware" (Thalhimer, 1994).  Large national newspapers are more likely to experiment with different content on their Web version than the print version (Fredin, 1997).  Gubman and Greer (1997) found that only 15.7% of newspapers with an online version used linked boxes and non-linear storytelling methods on the articles appearing on the Web.  They also found that only 14.5% used multimedia such as audio or video in their news articles, although multimedia was being used on their sites in advertising.  This study found that large newspapers were more likely than small newspapers to use these features.
Singer, Tharp, and Haruta (1998) showed that larger newspapers were more likely to have separate staff members for online operations.  Cole (1975) selected from the nation's largest newspapers to draw a sample of science articles under the presumption that the larger newspapers have the capability to hire science writers.  In addition, Trumbo (1995) used a sample of leading newspapers because their articles are likely to be reprinted in many other smaller newspapers owned by the same company, increasing the influence of the large, prestigious newspapers.  For these reasons, this study used leading national newspapers to determine if differences exist in articles between print and online versions, and to examine differences among newspapers.
The four opinion-leading newspapers in the country are The New York Times, The Wall Street Journal, The Washington Post, and The Los Angles Times (Krantz, 1994).  USA Today has a larger circulation than any of these (Vivian, 2001).  Because The Wall Street Journal is a business newspaper rather than a general interest national newspaper like the others (Vivian, 2001), USA Today was used in this study in place of The Wall Street Journal.  The influence of USA Today is strong because of its size and because of the distribution of its stories in other Gannett newspapers (Vivian, 2001).  The Christian Science Monitor, another national newspaper, also was chosen because of its depth in science and environmental coverage.
The most efficient way to get a representative stratified sample of the content of a daily newspaper for a given year is to gather two constructed weeks of data (Riffe, Lacy, & Fico, 1998).  Cole (1975) used three constructed weeks from each of three years--1951, 1961, and 1971--to measure trends in science news coverage.  However, McMillan (2000) found that most Web content analyses emphasize a particular time frame because the changing site content necessitates rapid data collection. The fleeting nature of the Web and the lack of uniformity in how newspapers archive their content, necessitated a fixed time period of data collection for this study, which was the three week period from February 22 to March 14, 2001.
Lacy, et al. (1991) used the type of purposive sample this study will use to compare conflict coverage among 12 large circulation newspapers in a study similar to this one.  This method was appropriate because, although not random, it was intended to compare certain context in coverage among newspapers and between print and online editions.  Lacy, et al. (1991) compared conflict coverage among the 21 newspapers with a mean of 16 articles from each newspaper.  This study attempted to collect at least 40 articles from the print version of each newspaper, enough from each title to perform meaningful statistical analyses.
Unit of analysis
Each print version of an environmental article collected during the sample period was one unit of analysis and the Web version of the article was another.  Each version of an article got its own code sheet.  The entire article was analyzed because of the way variables were measured. All environmental articles identified in the print version of the newspaper were included in the study.  If the article was subsequently identified on the newspaper's Web site, that version was included in the study as well.
Any environmental article with a byline in the main news section, business section, local news section, lifestyle section or science section of the newspaper was used.  Articles in the advertising sections of the newspapers were not used.  As with other newspaper research, opinion/editorial pieces were not included (Li, 1999).  The study analyzed only articles produced by the newspaper's staff and not those by a wire service used in the newspapers.  This decision was based on the assumption that online editors would be more likely to invest resources in improving articles produced by their own staff and not those from a service.  Only articles with bylines were used, as briefs and other short articles are frequently taken from wire services.  Even if the staff writes the briefs, the articles are unlikely to be enhanced for the Web.
For the this study, environmental articles were those that regarded any of the following:  air pollution, water pollution, solid waste, deforestation/logging, endangered species, population growth, government departments such as the Environmental Protection Agency or the Department of Interior including state-level agencies, natural resources, urban sprawl, animal rights, invasive species, biotechnology, genetically engineered food, ecology, mining, global warming, wildlife, wilderness areas, national parks, national forests, environmental racism, eco-feminism or any other topic germane to the environment.  To be used in the study, the article had to have a major focus, which revealed itself in the headline or first five paragraphs, (Boyle, 1995) on one of the topics listed above. Articles that only mentioned briefly environmental issues were not included.
During the collection period, two coders tracked the print newspapers for any environmental articles. Next, coders looked for articles on the respective newspaper's Web site.  Only environmental articles on the Web site that had a print counterpart were coded.  During the collection period, only two articles were found online and not in print.  One was an advertorial; the other included a page number for the print edition, suggesting that it was inadvertently left out of the print version. Reliability between the two coders on 10% of the stories (N = 41) ranged from 87% (N = 2) to 100% (N = 16), with an average of 96%.
Context
The context errors that scientists note the most in reporting of their work are quotes taken out of context, lack of reference to other research, and a lack of perspective of the issue (Walters, 1996).  Because interviews with quoted individuals would have been required to determine whether quotes in the articles were used in proper context, that criteria of context was not measured in this study. Past research has conceptualized and measured context in several other ways.  This study borrows nine of these, each coded as present (1) or not present (0). These measures are summarized below.
* Reference to research the article is based on.  One study found that 40 percent of science articles are significantly different than the research on which they are based (Walters, 1996).  A reference to research appearing anywhere in the article including photograph captions and cut-lines was coded as present.  Links to the original research in the Web version also were coded as present.
* Historical context.  Hart (1994) suggested other ways to add context to news, including adding historical information.
* Comparison using similar examples.  Hart (1994) suggests writers draw other, similar examples to issues to add context, the same way a reviewer compares movies.
* Reference to other research.  This study also coded for reference to other research that pertained to the topic in each article.
* Sources of more information.  Whereas print newspapers are limited in the amount of information they provide, the Web allows more information to be included and linked to. The presence of sources where readers could find