In the Face of Critique

Systematic Searching and Collection of Qualitative Studies

While it is difficult to locate every single study related to the experiences of journalists covering health and science, we attempted to be as comprehensive as possible in our search strategy. We limited the search strategy to the peer-reviewed literature for two reasons: (a) so as to provide a clearly focused and manageable data set and (b) to make use of the “quality” screening employed by journals. We are aware that the non–peer reviewed literature (Estabrooks, Field, & Morse, 1994) may form an additional set of useful documents for future work. A third retrospective reason for not expanding beyond the peer-reviewed literature was the conceptual richness of the studies identified, which were sufficient to provide insight into our research questions. As such, the data set uncovered should be viewed as a representative sample of qualitative studies on the experiences of health and science journalists.

The collection of this representative sample involved several strategies. First, on two occasions (October 2009 and January 2010), the following core databases were searched: Communication and Mass Media Complete (CMMC), Communication Abstracts, Academic Search Premier, PubMed, and Web of Science. These core databases were chosen in consultation with a Journalism and Communication Studies Subject librarian at Concordia University as the most likely to yield results relevant to the lived experiences of science and health journalists.

Database searches used a building block method with three concepts: journalism, science, and a qualitative filter. The concepts were used to choose appropriate, tailored subject terms and keywords in the thesaurus/index of each database. For example, for CMMC, the journalism concept used subject terms, including journalis* or mass media or “reporting&reporters” or “news” or newspaper* or press or broadcast* or blog* or investigative reporting or media spillover or digital media or photojournalis*. The science concept for CMMC included the following subject terms: mass media & the environment or SU (scien* or health or medic*) not SU (political science or social science*). The qualitative filter—a detailed selection of keywords designed to filter out studies that were not qualitative in nature ¹ —was then combined with the journalism and science concepts to search the database. Search results for each core database are presented in Table 1.

OPEN IN VIEWER

Table 1.

Database Retrieval Numbers From Core Databases

Database	Initial Date Searched a	Number Retrieved	After First Exclusion (Title and Abstract) b
Communication and Mass Media Complete	October 5, 2009	390	104
Academic Search Premier	October 10, 2009	733	66
Communication Abstracts	October 10, 2009	45	17
Web of Science	October 19, 2009	504	43
PubMed	October 19, 2009	218	17
Total		1,890	247

a.

Databases were searched a second time in January 2010.

b.

Studies were added to this total through subsequent citation, footnote, author, and journal searches before a second round of exclusions was performed to give a total of 21included studies (see Method section in text).

Additional searches were performed on Google Scholar and in the online databases of key journals (e.g., Journalism, Journalism Studies, Science Communication, Public Understanding of Science, Health Communication), which were chosen by looking at the original database search results and determining which journals were cited most frequently. Finally, we performed citation, footnote, and author searches as related to identified studies (see Barroso et al., 2003, for a fuller description of these techniques).

Selection Process, Inclusion/Exclusion Criteria

In total, our systematic searches yielded 1,890 studies. The process for identifying which of these studies were relevant to our research questions is outlined in Figure 1. First, this involved examining the articles’ titles and abstracts, excluding articles not dealing with journalism or science/health. For example, studies on public service announcements and advertising campaigns were excluded, while articles on journalism, reporting, and news media related to science/health were included. At this stage, we did not exclude articles based on whether they were qualitative studies. This process considerably reduced our pool of articles to 247 (Table 1, first exclusion).

OPEN IN VIEWER

Figure 1.

Flowchart on selection process (adapted from Barroso et al., 2003)

Second, we read full articles to determine if a study involved qualitative research with journalists covering health and science. Following Sandelowski and Barroso (2003), we generously defined “qualitative research” as empirical research with human participants that used commonly identified qualitative methods within any theoretical orientation. Studies were not excluded for reasons of quality, since consensus does not exist on quality criteria for qualitative research (Sandelowski & Barroso, 2003) and, for this reason, to avoid excluding valuable data that only some would view as containing methodological mistakes. We excluded (a) studies with no human participants, (b) studies not dealing with science/health, (c) studies not dealing with journalism or journalists, (d) qualitative studies with no identifiable results, (e) mixed methods studies where qualitative results could not be differentiated from other results, (f) studies with a mixture of participants (journalists and others) that did not allow the separation of groups, (g) non–peer reviewed studies or other nonresearch accounts. This phase was conducted independently by the two researchers, with disputes on inclusion resolved through argumentation between the authors. This process resulted in a sample of 21 articles for analysis.

There is little direction provided in existing literature on how many studies are needed to produce a saturated and transferable product of synthesis. Avoiding strict rules, Sandelowski et al. (1997) and Paterson et al. (2001) suggested a minimum of 10 to 12 primary studies. We see the question of sample size as related to (a) the comparability of identified studies and (b) matters of narrow versus broad research interest. Comparability relates to asking whether studies can be matched (Bondas & Hall, 2007) through, for example, comparing research questions across studies (Sandelowski et al., 1997), which can help set sample size parameters through the removal of incompatible studies. The second element relates to a desire for saturation or depth, in that while more studies can lead to broader theorizing, a large sample can also diminish the ability to create a deep analysis (Bondas & Hall, 2007). Here, our 21 articles were comparable via their common research questions and prioritized for depth of thematic analysis.

Data Extraction and Synthesis

To extract qualitative data from the 21 included studies, we used the metastudy approach of Paterson et al. (2001), which extracts and analyses data on three levels: metamethod, metatheory and metadata. We used this approach for two main reasons: (a) the metastudy approach asks that the contexts of each included study be taken into account and (b) the approach is comprehensive and complex in nature, thus ensuring that each included article is examined in depth and that generalizations and conclusions are drawn with caution (Thorne et al., 2004).

Essentially, metastudy involves the comparative analysis of processed data from each study and therefore involves synthesizing constructions of constructions. Processed data here refer to text, from one word to full paragraphs, contained in each study on the experiences of journalists covering health and science. Processed data were initially documented with reading grids inspired by Paterson et al.’s (2001) “primary research appraisal tool” (pp. 135-139). The reading grids recorded details for discussion, such as the studies’ research questions, the participants, the major findings, and the conclusions drawn. This extraction of data was descriptive and not evaluative. These structured summaries were used to discuss what was to be considered a “unit of data” and to ensure that we remained faithful to the meaning and context of each primary study. Claims as to what was a finding by the primary researcher was subject to our interpretation, with each report read thoroughly before textual elements were included.

Subsequently, data were formally extracted into spreadsheets along the three metastudy divisions: (a) metamethod, or where the findings came from and how they were collected; (b) metatheory, or the underlying research design strategies that drove the study; (c) metadata, or the processed, reported findings of the study. We were careful to retain the meaning and context of the data during this extraction. This provided the ability for explicit analytic comparison between divisions during the synthesis phase. The extraction produced 40 pages of processed metadata alone, supporting the premise that the selected studies contained enough data richness to amplify the results of each study and to provide the first available synthesis on the topic.

The last step in the process was to reconstruct and synthesize the extracted data. This involved moving from individual studies to a “terrain” perspective. This was first accomplished with the use of metasummaries to provide a thematic summary of reoccurring themes. This involved reviewing the metadata and summarizing them according to the themes identified in the data themselves (see Sandelowski & Barroso, 2003, and references therein for a fuller description of this method). This thematic analysis was then subjected to several synthesis devices, including investigation of theme frequency to examine effect sizes, the creation of a taxonomy of findings to provide a conceptual organization of the identified themes, and use of sustained comparisons (Sandelowski & Barroso, 2003) and data reintegration (Paterson et al., 2001) to deliberately probe deeper into the extracted data.

Thematic Metasummaries of the Findings

The 21 studies provide a body of knowledge on the experiences of science and health journalists ranging from 1994 to 2010. In this section, we describe a thematic analysis of the studies along metamethod, metatheory, and metadata divisions.

Metamethod

Analysis of the qualitative studies along the metamethod division is summarized in Table 2. The first theme to arise from this metasummary is that the majority of studies involve North Americans. Six of the 21 studies reviewed have Canadian origins, with a further 6 conducted in the United States. Just under a quarter of the remaining studies were from a European perspective, followed by 1 study from Australia and 1 from New Zealand. Only 2 studies made an attempt at research on a multinational level: one using data from Europe, Canada, and Australia, while the second loosely referred to its research site as “international.”

OPEN IN VIEWER

Table 2.

Summary of Metamethod

Author(s), Year	Country	Method(s)	Sample Size	Theoretical Orientation
Van Trigt, Jong-van den Berg, Haaijer-Ruskamp, Willems, J and Tromp (1994)	The Netherlands	Interviews (+content analysis)	7	Descriptive
Hansen (1994)	Britain	Interviews (+questionnaire)	31	Social construction
Saari, Gibson, and Osler (1998)	Canada	Interviews (+census, content analysis)	43	Descriptive
Conrad (1999)	United States	Interviews	15	Social construction
Reed (2001)	Australia	Focus groups, interviews	9	Social construction
Trumbo, Spencer, Dumlao, Yun, and Duke (2001)	United States	Interviews (+surveys)	5	Descriptive; social construction
Treise and Weigold (2002)	United States	Open-ended qualitative surveys	497	Descriptive
Hijmans, Pleijter, and Wester (2003)	Netherlands	Interviews (+content analysis)	10	Descriptive
Larsson, Oxman, Carling, and Herrin (2003)	Europe, Canada, Australia	Focus groups, interviews (+ surveys)	20, 10	Descriptive
Geller, Bernhardt, Gardner, Rodgers, and Holtzman (2005)	United States	Interviews	22	Descriptive
Waddell, Lomas, Lavis, Abelson, and Shepherd (2005)	Canada	Interviews	12	Descriptive; social change
Chew, Mandelbaum-Schmid, and Gao (2006)	United States	Interviews (+content analysis)	8	Descriptive
Gasher et al. (2007)	Canada	Interviews	12	Social construction; social change
Henderson and Kitzinger (2007)	Britain	Interviews (+content analysis)	6	Social construction
Hodgetts, Chamberlain, Scammell, Karapu, and Waimarie Nikora (2007)	New Zealand	Interviews	8	Social construction; social change
Roy, Faulkner, and Finlay (2007)	Canada	Interviews (+content analysis, surveys)	4	Social construction
Wilkinson, Allan, Anderson, and Peters (2007)	Britain	Interviews (+content analysis)	5	Descriptive
Balasegaram, Balasegaram, Malvy, and Millet (2008)	International	Interviews (+content analysis)	9	Social change
Ward and Jandciu (2008)	Canada	Interviews	25	Descriptive
Hinnant and Len-Rios(2009)	United States	Interviews (+surveys)	20	Descriptive
Maillé, Saint-Charles, and Lucotte (2010)	Canada (Quebec)	Participatory observation, interviews (+content analysis)	10	Descriptive

In total, 788 journalists were involved in the 21 studies. However, one study had a notably high sample size, using a qualitative survey to collect data from 497 participants (Treise & Weigold, 2002). It is not possible to tell if the studies made use of identical participants because of the common practice of not identifying the names of participants. Irrespective, the majority of the studies had participant sample sizes of between 10 and 43, with 11 studies falling into this range. Nine studies had sample sizes below 10 participants. Lower sample sizes correlated with the use of mixed methods (quantitative/qualitative studies) that generally used interviews and focus groups to complement content analyses and surveys, while the 10- to 43-participant category correlated mostly with studies using a single qualitative method to collect data, generally interviews with participants.

Three studies collected data from sources other than journalists; however, data originating from these nonjournalist sources were excluded for the purposes of this metasynthesis. We also excluded quantitative data, which was common with 13 of the studies using a mixed-methods approach and often arguing that this would lead to greater understanding about science and health journalism. The remaining eight studies analyzed used only qualitative methods (largely structured, semistructured, or open-ended interviews) to collect data.

Metadata

In analyzing the qualitative data extracted from the 21 included studies, 14 distinct themes emerged that could be further divided into 49 subthemes (see Table 3 and effect sizes). Below we describe the two metathemes that cut across multiple studies (major themes) as well as briefly describe the common and minor themes. A metatheme was considered “major” if it appeared in over 50% of the studies, “common” if it appeared in over 10% of the studies, and “minor” if it was below 10%. These characterizations were checked against the amount of extracted data for each metatheme so as to avoid, for example, an important theme being called minor simply because it only appeared in one study. The reader is reminded that these metathemes emerged from the processed experiences of health and science journalists as “narrated” to various researchers and reported in their publications (our metadata). Such distance from lived experience, and the issues of representation this raises between “lived and narrated lives” (Sandelowski, 2006), is a complexity that should temper the interpretation of this section.

OPEN IN VIEWER

Table 3.

Themes Emerging From the Metadata, Their Effect Size, and Their Data Source

Metathemes and Subthemes	Effect Size (N = 21)	Data Source
Major themes
1. Sourcing practices	15 out of 21	Hinnant and Len-Rios (2009); Balasegaram et al. (2008); Ward and Jandciu (2008); Gasher et al. (2007); Roy et al. (2007); Chew et al. (2006); Geller et al. (2005); Waddell et al. (2005); Larsson et al. (2003); Trumbo et al. (2001); Reed (2001); Conrad (1999); Saari et al. (1998); Hansen (1994); van Trigt et al. (1994)
1a. Expert sources
1b. Scientist as source
1c. Journal as source
1d. PR as source
1e. Pharmaceutical industry as source
1f. Internet as source
1g. Source selection
1h. Balancing sources
2. Story selection	11 out of 21	Balasegaram et al. (2008); Ward and Jandciu (2008); Gasher et al. (2007); Henderson and Kitzinger (2007); Hodgetts et al. (2007); Roy et al. (2007); Geller et al. (2005); Waddell et al. (2005); Hijmans et al. (2003); Larsson et al. (2003); Hansen (1994)
2a. News values in science journalism
2b. Value of research
2c. Journalist’s personal interests
2d. Editorial priorities
2e. News frames
2f. Social determinants of health
Common themes
3. Journalistic tools and routines	7 out of 21	Maillé et al. (2010); Hinnant and Len-Rios (2009); Balasegaram et al. (2008); Hodgetts et al. (2007); Roy et al. (2007); Reed (2001); Saari et al. (1998)
3a. Human/real- people element
3b. Visuals, graphics, and statistics
3c. Balance
3d. Wire copy
3e. Editor as “antiscientist” test
4. Journalistic constraints	5 out of 21	Maillé et al. (2010); Ward and Jandciu (2008); Larsson et al. (2003); Reed (2001); Saari et al. (1998)
4a. Time
4b. Lack of space
4c. Finding/ researching reliable sources
4d. Industry realities (e.g., budget and staff cuts)
4e. Competition and commercialization
5. Audience	5 out of 21	Hinnant and Len-Rios (2009); Hodgetts et al. (2007); Chew et al. (2006); Reed (2001); Hansen (1994)
5a. Image of audience
5b. Audiences’ perceived interests
5c. Feedback
5d. Responsibility to audiences
5e. Civic journalism
6. Autonomy	5 out of 21	Ward and Jandciu (2008); Chew et al. (2006); Geller et al. (2005); Saari et al. (1998); Hansen (1994)
6a. Relationships with editors
6b. Relationships with journalists
6c. Pressure from advertisers
6d. Pressure from interest groups
6e. Pressure from audience
7. Journalistic training/ education	4 out of 21	Ward and Jandciu (2008); Treise et al. (2002); Saari et al. (1998); Hansen (1994)
7a. Lack of education as obstacle
7b. Lack of education as benefit
7c. Professional training
8. Science literacy	4 out of 21	Hinnant and Len-Rios (2009); Wilkinson et al. (2007); Treise et al. (2002); Reed (2001)
8a. Definition/ measure of literacy
8b. Audiences and science learning
8c. How to communicate risk
9. Journalist/journalism’s role	3 out of 21	Geller et al. (2005); Waddell et al. (2005); Reed (2001)
9a. Professional identity
9b. Informers versus Educators
9c. Social responsibility
9d. Independence
10. Specialist versus generalist	3 out of 21	Henderson and Kitzinger (2007); Conrad et al. (1999); Saari et al. (1998)
Minor themes
11. Media format	2 out of 21	Henderson and Kitzinger (2007); Treise et al. (2002)
11a. TV versus Print
11b. Internet
12. Language	2 out of 21	Hinnant and Len-Rios (2009); Henderson and Kitzinger (2007)
12a. Choice of language
12b. Scientific jargon/ translation
12c. Use of metaphors
13. Conflicts of interest	1 out of 21	Geller et al. (2005)
14. Omitting methodology in articles	1 out of 21	Maillé et al. (2010)

Major theme: Sourcing practices

The most prominent theme addressed in the analyzed data was the sourcing practices of journalists (present in 15 of the 21 studies). We defined this theme as issues related to who and what journalists use as sources for their stories; how they go about finding, selecting, and dealing with them; and how they integrate information gained from sources into their stories.

Data under this major theme suggest that a main job component of health and science journalists is finding appropriate sources. There is the sense that they do not rely on “regular” sources (Conrad, 1999, p. 288) but instead use so-called expert sources to clarify, shape, and illustrate their stories, as well as to lend credibility to their work. The data highlight factors at play in determining who is quoted in a final story. With journalists working under strict time pressure, accessibility is one of the most pressing issues documented in the data. Simply put, those experts who are easily reachable or call journalists back are more likely to be used in stories. Another major factor in source selection depends on how the expert sources “speak” to the press. Sources who help journalists put research into context and can explain complex scientific concepts in everyday language are valued much higher than those who apparently lack such abilities. For example, one study quoted a journalist as saying, “Researchers who can talk like human beings are like gold!” (Waddell et al., 2005, p. 133). Journalists also noted that an expert source’s reputation plays a significant role in their use. This is based on a number of features, such as their status in their field, their notoriety or even “celebrity” status, their reputation among journalists, the individual journalist’s past experiences with them, their affiliation, and level of independence.

Several distinct subthemes did emerge from the sourcing practices theme (Table 3). For example, one subtheme dealt with the “scientist as source.” While the reported data demonstrate the interconnectedness of these two groups—with journalists relying on scientists for information for their stories and, as discussed by the journalists, the scientists’ reliance on them to popularize their research—the relationship between the two was narrated as unstable. The reported experiences suggest that scientists are currently more accessible than in the past, with many taking steps to build relationships with journalists. However, levels of distrust were apparent between the two groups, with, for example, reporters finding it difficult to make “reliable contacts in the scientific community” (Saari et al., 1998, p. 76) and often encountering scientists who do not understand embargo systems (Ward & Jandciu, 2008). The “scientist as source” subtheme emphasized the different values of the two groups, as related to time, use of language, the notion of balance, and prospective audiences (Waddell et al., 2005, p. 134).

A related subtheme was “journal as source,” which revealed health and science journalists’ heavy reliance on scientific journals. Journals were cited as the main source for story ideas. Press releases and press conferences, as well as public relations officers, were also cited (“PR as source”), however, the data suggest that journalists have differing opinions on their value. While PR could lead to story ideas, sometimes even making the story-writing process easier for journalists, many journalists remained skeptical of organizations trying to push their own agendas. PR was discussed differently depending on its origin. For example, press releases from universities were the best received, while those from the pharmaceutical industry were met with the most cynicism. A small part of the data suggested that journalists would like to see the scientific community improve its PR techniques, specifically in Canada (Ward & Jandciu, 2008).

Journalists narrated a wealth of other information under this theme. This included subcategories such a: (a) “Pharmaceutical industry as source,” where journalists viewed the pharmaceutical industry as the most aggressive in terms of PR and greeted this with the most suspicion, while also explaining that a pharmaceutical industry PR officer can provide patients in order to give their stories a “human element” (Gasher et al., 2007,); (b) “Internet as source,” where journalists related how the Internet has changed their jobs by giving access to virtually unlimited sources of information (Trumbo et al., 2001); and (c) “balancing sources,” where the data suggest that journalists do not always feel a need to balance sources, arguing that the nature of scientific research makes this superfluous, but they were also confronted with the common pressure of needing “two sides of the story” (Conrad, 1999).

Major theme: Story selection

Story selection was the second most prominent theme covered in the literature analyzed (11 of the 21 studies; Table 3). We defined this theme as the processes and criteria used by journalists covering health and science to select which stories make it into print or on air. The journalists cited in the data discussed six subcategories under this theme.

First, health and science journalism generally makes use of the same news values as other forms of journalism: Stories need to be timely, novel, and relevant to their audience. They need to grab the readers’ attention by being understandable and interesting. In this vein, scientific “breakthroughs” or “breaking news” were mentioned as being particularly eye-catching stories. Many journalists suggested that elements of controversy (e.g., emphasizing a “battle” between scientists or divisive policy issues) or human interest (often referred to as the human face angle) are needed to make good stories. For example, Henderson and Kitzinger (2007) reported that “conflict” and “real people” were seen as “more interesting than consensus and ‘impersonal organisations’” (p. 76). In short, as Geller et al. (2005) summarize, health and science journalists generally believe that newsworthy stories are “novel, applicable, controversial, credible and entertaining” (p. 200).

A second factor related to journalists’ perceptions of the value of the research covered, such as whether the research was sound, whether the source was reputable, and whether the results had fairly direct implications for local audiences. More specific story selection criteria, however, related to individual journalist’s scientific background or level of experience/education and were applied differently in individual cases. In fact, a number of journalists reportedly decided for themselves what stories to cover, relatively unrestricted from editorial pressure. This discussion of personal interests, and the implicit assumption of associated freedom, formed the third subtheme. As one journalist related, “It’s very important that you be interested in the subject and convey that interest” (Geller et al., 2005, pp. 201-202).

Of course, health and science journalists are affected by the editorial priorities of their news outlet; as such, a fourth subtheme was the role that editors play in story selection. This was discussed in terms of how an editor’s personal interests can overrule the journalist’s story selections, even if the editor has less experience or is less knowledgeable about health and science issues. The following excerpt from Henderson and Kitzinger (2007) illustrates this power relation:

[The editor] had read some piece in the New Yorker which had probably taken six months to write [. . .] it obviously came as a complete surprise to (him) who hadn’t read anything I had written about it . . . [s]uddenly it’s a big story and they are all over you and they want thousands of words with something you have been struggling to get into the paper for five years. (p. 75)

Fifth, health and science journalists discussed use of recurring news frames to make the research covered relevant to audiences, such as the “lifestyle” approach, which emphasizes personal responsibility to health (i.e., smoking, eating well, exercising; Hodgetts et al., 2007). This linked to the suggestion that the reliance on such news frames is one reason why health and science journalism does not often cover topics like social determinants of health (a unique focus of some of the social change studies included in our sample), as these types of stories do not fit into common frames. Here, journalists reported difficulties in putting a human face on such stories, not having enough familiarity with the subject, or a perceived disinterest within the general audience that prevented them from wanting to cover such stories.

Brief summary of common themes

Eight themes were identified with effect sizes ranging from 3 to 7 of out 21 studies (Table 3). We refer to this effect size as “common,” since the themes appeared in multiple studies that included varied samples of journalists who independently discussed the topic. For example, while the audience theme—which encompasses issues related to how the journalists perceive and treat audiences, and what role audiences play in journalists’ work and stories (Table 3)—was much less common across the data set, it was nevertheless discussed in 5 studies that involved a diverse sample of 68 journalists.

Of the common themes, journalistic tools and routines were discussed most frequently across the studies. This theme discussed how journalists covering health and science use a number of tools and routines—beyond researching stories, conducting interviews, and actually producing the stories—to illustrate their stories and make them understandable for audiences. The human element was one tool cited on a number of occasions by journalists to get their audiences to relate to health and science stories and add context. A typical comment was the following from Henderson and Kitzinger (2007):

You’ve got some breakthrough and you need to put a human face to [it] so the job was to go out and not only get all the facts but also find a case study [. . .] [i]t was very hard because sometimes you have to find these people, very sick people and explain to them why they are not being used for entertainment of some kind. (p. 78).

This human element was accompanied with the use of visuals and statistics, concepts of how to balance voices to make stories interesting and understandable (Conrad, 1999), and the use of editors to ensure stories were not overly scientific (“antiscientist” test).

It should come as no surprise that journalistic constraints were also commonly discussed (Table 3). This common theme came in the form of constraints originating (a) directly from within the newsroom and journalistic processes themselves or (b) from external sources such as the realities existing in the media industry. Time was one of the most often cited constraints. Journalists spoke of a deadline-driven occupation, characterized as a hard “rhythm” to get into (Saari et al., 1998, p. 72), a pressure to perform, a need to deliver material throughout the day, and difficulties in judging which sources are most reliable or unbiased. Compounding this was the familiar remark that journalists and scientists operate under different conceptions of time (Conrad, 1999; Maillé et al., 2010). Next to time, space limitations (Ward & Jandciu, 2008) showcased feelings of not having enough room for “good” journalism. There was also a clear sense that shrinking media markets, lower budgets, and a lack of dedicated health and science specialist reporters were on journalists’ minds (e.g., Larsson et al., 2003).

The audience was also an important theme addressed. It is well known that journalists’ perception of their audience has an impact on how they produce their stories. When addressing questions about who journalists see themselves writing for, the analyzed literature points to a notion of the “everyday” person, sometimes referred to by journalists as “Joe Bloggs” (Hodgetts et al., 2007, p. 50) or the journalist’s grandmother (Hinnant & Len-Rios, 2009, pp. 99-100). It was clear that some journalists see themselves as part of the audience, or at least a good measure of the general audience member, basing story selection and coverage on their own interests. Thus, the interests of the “imagined audience” (Reed, 2001, p. 285), it seems, are generally not based on any kinds of readership surveys but rather on what individual journalists consider important and interesting and their own assumptions of the demographics of their readers.

Three studies did independently, and specifically, discuss the role of the journalist (Table 3). These discussions pointed to included journalists generally believing that producing news articles is solely their responsibility and that it is up to them to decide what is covered and how the stories are framed. One journalist explained it this way in Maillé et al. (2010): “[The journalist] is responsible for the inclusion of science in social debates. He [sic] is thus the bearer of society’s questioning. He is not there to represent scientists’ interests” (p. 75). Journalists expressed a perception of having a lower “social status” (Reed, 2001, p. 289) than scientists in society and relate attempts—often unsuccessful—to equalize the playing field. They also saw their roles as information disseminators whose job it is to inform the public of important health and science stories, rather than educating the public and raising science literacy. These sentiments included comments such as the following from Geller et al. (2005):

Whenever people talk about the media’s role in educating the public, I get a little worried. . . It seems to me that [our] role is to report the news . . . beyond that, I’m not sure it’s our job” (p. 200)

Here, there was a sense of social responsibility: “The only model that works is to serve our readers first” (Waddell et al., 2005, p. 131).

Brief summary of minor themes

As a final example of the content that arose out of our thematic metasummaries, several studies did have more unique themes that appeared in only one or two studies. These were dominant enough in a study to warrant their inclusion as a metatheme arising from the data set, so as to work to remain true to the content of each particular study. One example of a minor theme was what we termed media format, which, for example, indicated that some included journalists (Henderson & Kitzinger, 2007; Treise & Weigold, 2002) saw print as generally more appropriate for covering health and science. Television was perceived to work faster than print and allow less time and space for health and science stories. How language was used, discussions of conflicts of interest, and how much scientific methodology to include in a story were also minor themes.

Taxonomy of the Metadata Findings

Overall, the 14 distinct metadata themes provide a “map” of the qualitative data contained in the included 21 studies. This map serves as an empirical foundation for further interpretive tasks and, in particular, as a clear summary from which further linkages can emerge (Thorne et al., 2004). As one purpose of this metasynthesis was to provide an assessment of the state of qualitative study to inform future theory development, we proceeded to create a taxonomy of the findings to transform the metadata themes into a conceptual form (Sandelowski & Barroso, 2003) that would aid the synthesis of the results (a further synthesis of this conceptual form with metatheory and metamethods results is described in the next section).

To do this, we clustered themes from Table 3 that were similar or went together (LeCompte, 2000) by constantly moving between the metadata and an evolving taxonomy that compared themes and subthemes. Through taxonomic analysis, as described by Spradley (1979), we grouped together themes according to their “semantic relations” (pp. 117-118): specifically “X is part of Y” (X being the themes that emerged from the metadata and Y being the taxonomic names evolving out of our interpretative groupings). This process resulted in the emergence of four taxonomic categories that grouped the metathemes in Table 3 according to their similarities and differences, thus providing interpretive categories that describe a higher level meta-order for the data that we extracted from the studies (Table 4). Importantly, this taxonomy shows a relational landscape of what has been studied and reported on in the qualitative literature. We specifically use the word relational to indicate that each taxonomic category can be related to the others (i.e., they are all ultimately related to the profession of science journalism) but that interpretative divisions are apparent.

OPEN IN VIEWER

Table 4.

Taxonomic Groupings of Metathemes

Taxonomic names (Y Items)	Grouped Themes (X Items)
Y1. Story construction elements	X1. Sourcing practices
	X2. Story selection
	X3. Journalistic tools and routines
	X11. Media format
	X12. Language
	X13. Conflicts of interest
	X14. Omitting methodology
Y2. External factors	X4. Journalistic constraints
	X5. Audience
	X7. Autonomy
Y3. Journalist identity	X6. Journalist training/education
	X9. Journalist/journalism’s role
	X10. Specialist versus Generalist
Y4. Science literacy	X8. Science literacy

The first taxonomic category involved story construction elements (Table 4, Y1). This was by far the largest category, containing 7 of the 14 themes present in Table 3. These 7 themes clustered together because they all represented factors that are part of journalistic story production and were described as under the journalist’s control. These factors may be tangible items, such as press releases or technology; processes, such as source location/selection and interviewing; or elements of “what makes a story a story,” such as the journalist’s personal interest or editorial priorities, and the decision making that goes with this. Excluded from this category were themes that did not deal directly with story production or, if they did relate to story production (e.g., journalistic constraints), were related as outside the control of the journalist. The fact that 7 themes could be clustered in this way show the attention paid to story construction in researching the experiences of health and science journalists. This is evident when we consider that “sourcing practices” and “story selection” were the only two major themes to appear in more than 50% of included studies (Table 3). Hence, it is sources that often drive stories in health and science journalism (further discussed in the next section).

The second taxonomic category was external factors (Table 4, Y2). This category encompassed 3 of the 14 themes, which clustered because of their reference to factors and influences outside of the newsroom that have effects on journalism quality and the external processes related to how journalism is done/created, such as industry realities and pressures from outside sources to cover a certain issue. The distinguishing feature of the category was therefore loss of journalistic control. Interestingly, this taxonomic categorization highlights that the factor journalistic constraints (Table 3), which obviously affect story construction, was discussed in the studies as outside a journalist’s control, causing it to be placed in this category. This clustering also brings the daily constraints that journalists report in the 21 studies (lack of time, lack of space, finding/reaching reliable sources, industry realities, competition and commercialization; see Table 3) into linkage with their audience and autonomy. As such, the taxonomy underscores that the autonomy of health and science journalists is not reported as an internal struggle for independence, as it might be if it was more strongly related to the third taxonomic category below. Instead, the included journalists expressed their autonomy in terms of pressures and relationships (see the autonomy subthemes in Table 3), something at the whim of external constraints and the need to respond to audience demands. As one journalist related in Gasher et al. (2007), when asked about what she considers when making story selections: “I’m very sensitive to the readership” (p. 564).

In contrast, the third taxonomic category contained internal factors related to the journalists’ own perceived role or how they feel about their journalistic identity, such as their responsibility to audiences, to journalism, and to science. This category was termed journalist identity (Table 4, Y3). Such identity factors were often described with terms such as informers, advocates, and translators of health and science information. This category clustered the journalists’ role—expressed as a socially responsible, independent informer or educator of their audiences (e.g., Geller et al., 2005; Reed, 2001; Waddell et al., 2005)—to their training/education and the place of specialist versus generalist reporters. In linking these themes, the included studies made it clear that there is no agreement on whether having a background in science is beneficial or not for health and science journalists (see Saari et al., 1998; Ward & Jandciu, 2008) but that specialization (in terms of years of experience) is beneficial to fulfilling their expressed role.

Last, the fourth taxonomic category involved issues that are part of science literacy and science learning in a social context (Table, Y4). This final category contained only the science literacy theme from the metadata (Table 3), which did not group with the others because the metadata for this theme did not broach identity issues or external constraints. While this category did touch on relationships with audiences, journalists did not discuss this theme in terms of their roles or external pressures. Instead, they spoke of how to effectively measure science literacy, how audiences learn about science, and how literacy relates to the audiences’ ability to gauge communicated risks (Hinnant & Len-Rios, 2009; Reed, 2001; Treise & Weigold, 2002; Wilkinson et al., 2007). The closest taxonomic relation was arguably discussions of how readers use health and science journalism to acquire knowledge as related to story construction elements. The clearest articulation of this was in the study by Treise and Weigold (2002) where a journalist asked, “Do we know what is science news to readers? Do we know how they pick which stories to read first? [. . .] I don’t think we know; otherwise, we’d frame many of our stories to meet those expectations” (p. 317).

Such discussions were about how readers make “sense” of science journalism, not how journalists construct stories, thereby leading us to place science literacy as its own taxonomic category. Furthermore, the need for a separate science literacy category resulted from studies discussing the social aspects of science learning and measuring science literacy, which did not group with issues of audiences or roles. This choice is supported by journalists suggesting that science literacy was one of the most pressing issues for science communicators and researchers to address (Treise & Weigold, 2002). However, the choice to have science literacy as its own taxonomic category should not be interpreted as an indication that it is mutually exclusive from the other three categories. Even though the studies examined here did not link science literacy to identity issues or external constraints, science literacy is closely tied to how journalists’ create stories and view their audiences and roles. This raises the important point that if one moves away from the metadata themselves, it is possible to theorize on how the taxonomic categories can be related to each others. Indeed, the goal of the presented taxonomy was to transform the metadata themes into a conceptual form that future analysis could probe for such insights.

Sourcing practices: A Synthesis of Metamethod, Metatheory, and Metadata

In the final Results section, we probe deeper into the metasynthesis data by providing a synthesis of one identified story construction element—sourcing practices. We end with this example because of the heavy critique that health and science journalists have faced in term of content production and presenting expert opinions (Bubela et al., 2009; Holtzman et al., 2005). Journalists have faced this critique while often having a common goal: to produce good stories that matter. But the definition of “good” in this context is a recalcitrant debate (Secko & Smith, 2010) that has yet to yield to a clear consensus. There is, therefore, a strong rationale for learning from the perspectives of journalists about why health or science stories are constructed as they are. Furthermore, with the dominant role sourcing practices play in the studies examined (see Table 3) it seems appropriate to end by asking, “What have we learned about sources?” And importantly, how have we learned it? What follows is an interpretative synthesis that considers sourcing practices as a whole (Sandelowski & Barroso, 2003). It pays particular attention to combining our taxonomy with its associated metamethod and metatheory results.

Comparing the Metasynthesis With the Wider Literature

To summarize, systematic searches for qualitative research identified a representative sample majorly skewed to the study of Canadian and American journalists. From this sample, we identified 14 distinct metathemes and 49 subthemes (Table 3) that could be further organized into four taxonomic categories (Table 4). An important result of the metasynthesis showed that this sample of qualitative research intensely focused on story construction elements, despite arguments calling for a shift from assessing only content production (e.g., text; see Roy et al., 2007) to more widely understanding health and science journalism through the perspective of journalists. Here it was sourcing practices that played a particularly dominant role in the identified studies. Furthermore, while only specifically addressed in three studies, the journalist’s perceived role (Table 3) was strongly identified as information provider over educator, with journalists seeing themselves as solely responsible for the content and frame of their work. In completing this work, the “human element” was the key journalistic tool commonly mentioned and appeared to be tied to perceptions of audience needs. Revealing a tension in this perception, the literature addressing science literacy (Hinnant & Len-Rios, 2009; Reed, 2001; Treise & Weigold, 2002; Wilkinson et al., 2007) suggests, however, that health and science journalists are ultimately unsure how their audiences make use of journalism to understand health and science.

In comparison, some studies have reviewed science communication literature from a non–methodologically specific orientation, identifying a number of major themes. For example, Logan (2001) distinguished three major areas in science communication research, namely, sourcing practices, news messages, and news channels. Weigold’s (2001) review of the literature points to the importance of story selection, specialist versus generalist journalists, and journalist relationships with audiences, editors, and scientists. These themes overlap with our results and offer suggestions for improved science journalism through an increased attention to audience needs and a critical examination of source relationships. Furthermore, the Logan (2001) and Weigold (2001) reviews closely mirror our results on content production and journalistic tools. This raises the intriguing point that for these themes the specific use of qualitative methods, when viewed on a metalevel, only reinforces the wider literature as opposed to adding a unique comparative contribution (we return to this point in the next section).

However, there are also differences. For example, while Weigold (2001) suggests specialist science reporters are more educated in science and thus hold different values than their generalist peers, our analysis suggests it is largely journalistic values that come first. The notion of an “inner circle” (Weigold, 2001) of senior health and science journalists sharing resources for their stories was also evoked and attributed as a cause of homogenous news coverage. While our metasynthesis results did suggest that health and science journalists may share resources at times, homogenous news coverage was alternatively attributed to cyclical sourcing practices (e.g., journalists consistently using the same sources) because of press releases and other PR efforts, as well as deadline-driven time constraints diminishing the opportunities for more in-depth coverage. As a final example, wider reviews suggest that health and science journalists have confidence in scientists. Our results imply that although journalists trust the information coming from scientists, this confidence is hampered by the fact that journalists largely view the scientific community as difficult to deal with or as attention seekers trying to push their own agendas. However, the main point—that the relationship between journalists and scientists is a fragile one—is reflected in the reviews and our metasynthesis results.

Recent research has begun to address overlooked areas identified in the literature reviews and our metasynthesis. For example, Stocking and Holstein (2010) investigated how journalists’ understanding of science influenced their core beliefs about their roles and perceptions of the audience, while Jensen’s (2010) research on the suggested failure of the “fourth estate” in science journalism cites social conditions (e.g., intermedia agenda setting, organizational constraints on independent judgment, pro-science bias, and pursuit of prominent placement for professional gain) as barriers to journalistic autonomy. Notably, much current research is focusing on the impact of the Internet on science journalism. Blogs seem to be a major point of interest, with research addressing issues such as new forms of connectivity and relationships to audiences (Allan, 2009), as well as bloggers’ sourcing practices, nonlinear (e.g., hyperlinks) story constructions, and related challenges to conventional journalistic norms (Walejko & Ksiazek, 2010).

Christians and Carey’s Promise of Qualitative Methods

The comparison of our results with the wider literature raises the issue of whether our corpus illustrates the promise of qualitative methods to provide a unique understanding of the experiences of health and science journalists, as described at the outset by drawing on Christians and Carey (1989). Adapted to our context, Christians and Carey’s argument suggests that a major task for qualitative researchers would be to systematically clarify what science and health journalists already know—to interpret the interpretations of journalists to better understand what guides their work—by giving naturalistic observations that are contextualized, widely compared, and crystallized to suggest areas to advance our thinking.

Although a complex point to untangle, it is possible to assess the corpus against Christians and Carey’s (1989) foundational argument. First, naturalistic observation involves entering a situation so deeply that you can recreate the imagination of those observed. This requires a broad set of methods from participant observation, to assessing personal documents, to open-ended interviews. As the metasynthesis shows, while each study is a rich source of data, the corpus as a whole was narrowly focused on interviewing, often in the context of a mixed-method approach (Table 2). Second, considering contextualization as both tying results to the environment they were produced in and recognizing each study as conditioned by past research, it was possible (within the limitations discussed below) to assess the environment of each study. However, the studies did not extensively reference each other and did reinforce the wider literature in many ways. This suggests that the corpus has yet to live up to studying health and science journalists, “so that two visits to Moscow, though seemingly identical in every respect, are never considered the same because the second occurrence is always conditioned by the first” (Christians & Carey, 1989, p. 365). Third, taking maximized comparison as varying the composition of the group studied to improve the substance and explanatory power of qualitative results, the metasynthesis clearly shows that the field has so far focused on North America and print journalists, suggesting the future need to improve the multiplicity of moments and locations (Christians & Carey, 1989). Fourth, if the ultimate goal is to crystallize important observations to suggest areas to advance thinking, the results have much to offer as the richness of the 14 metathemes and 49 subthemes (Table 3) reveal many important aspects to follow up on.

We would like to be very clear on our interpretation of these points. While overall the use of qualitative methods as assessed through our corpus has yet to fully meet the promise of a macroscopic, transformative use, as advanced by Christians and Carey (1989), individual studies are an abundant source of useful information. It is not our intention to critique studies individually but to point to strengths and weaknesses on a metalevel. Nevertheless, it is intriguing to speculate that the identified, substantial embedding of qualitative methods in mixed methods research, which in some articles was discussed with the air of simply a supporting research tool, might be impeding its more robust use as an individualized method. To this point, it is important to keep in mind that we identified only 21 studies, suggesting that this type of inquiry is in its early stages. But such work does seem to be gaining in popularity, as indicated by just under half of the studies appearing since 2007 (Table 2). This increase in popularity is a further retrospective reason to account for, synthesize, and assess our current qualitative knowledge against the promise of qualitative methods.

Limitations of the Study

Before concluding, it is worth making a few final points about the methodology that was used. This work is based on literature searches that attempted to be comprehensive but with the realization that it is difficult to ensure the collection of every single study. In addition, databases and literatures are not static but continue to evolve, so new studies will appear. Conducting a qualitative metasynthesis involves the creation of an aggregated account of the literature in question based on our own interpretations of the primary researchers’ constructed work. As Paterson et al. (2001) state, qualitative metasynthesis is “research of research” (p. 5). We and the reader must remain aware that we are synthesizing constructions of constructions and are several steps removed from the primary data and actual lived experiences of the journalists. Moreover, the qualitative metasynthesis that is presented is an interpretative product (Sandelowski & Barroso, 2003) and cannot be thought of as the only possible synthesis that could be created from the included studies. Two additional limitations exist: (a) While we worked to remain true to each included study, a qualitative metasynthesis by its nature decontextualizes data while having to rely on the published report to clarify the context of the research, and thus we may have had only selective access to the context in which the data were created: (b) This type of metasynthesis does not have access to original data, and thus our synthesis is only as good as the articulation it received from the primary researchers.

Conclusion: Where We Are Going

By drawing connecting threads between the included studies—the pieces of the jigsaw puzzle—we worked to illuminate underlying issues in science and health journalism that may not have been directly or fully addressed by the existing research. With some humility, we conclude with the suggestion that these underlying issues point to “where we are going” in the study of science and health journalists.

The first signpost is the dominant role that sourcing practices played in the identified studies, which related a complex set of experiences on balancing scientific accuracy with usability, an orientation to authority for the sake of story credibility, and instability in such relationships that may draw disruptive energy from authorship, ownership, and status conflicts. Such findings were drawn from three types of theoretically oriented studies (labelled descriptive, social construction, and social change studies), which were principally interested in descriptive inquiry (see Sandelowski, 2000). However, the extent of our knowledge on sourcing practices can be taken as indication for the need to move from description to interpretation to, in a few limited instances, the advocacy of particular changes to practice.

If such a path is charted, a second signpost is the limited scope of qualitative work on the longstanding controversy over the proper role of the health and science journalist (Denzter, 2009; Logan, 2001), for which lack of clarity can contribute to the difficulty of dispelling often differing expectations between scientists and journalists. In three studies (Geller et al., 2005; Reed, 2001; Waddell et al., 2005), it appeared that journalists felt some possessiveness over their role in moving the discussion of science into the public arena. But it was equally clear that these journalists understood that scientists might feel fear, or at least profess fear, over journalists appropriating science by simplifying esoteric knowledge, to borrow a term from Reed (2001). This is congruent with current literature that argues there is a fundamental disconnect between how scientists and journalists interpret and describe the research process (Bubela et al., 2009). We are still a long journey away from understanding how widely the above views are held and whether a diversity of opinion also exists.

The third and final signpost we will suggest is the past dominance of story construction elements and consequently the new depths that future qualitative research could explore by asking journalists about the other three taxonomic categories: external factors, journalist identity, and science literacy, as well as the linkages between all four categories. Other results that could be built on include diversifying opinion on underrepresented themes in Table 3 and participant sampling beyond North Americans. We suggest this is all wrapped up in the human element, as the key journalistic tool commonly mentioned. The desire for human faces to create connections to complicated science, however, is a time-consuming problem that can lead to oversimplification, especially when considering that journalists suggested they are unsure how their audiences make use of their journalism. Therefore, with discussion about increased media fragmentation (Allan, 2009) and about media use varying depending on individual audience members’ “social identity and values,” a vital direction emerges from the qualitative metasynthesis over how journalists see themselves meeting the great challenge of simply reaching audiences (Bubela et al., 2009).

In conclusion, as per Christians and Carey (1989), it may well be a qualitative journey past these three signposts that will make public how health and science journalists have “guarded their sheep” (p. 373) to improve the consultable record of journalistic practice. In the face of critique, it is hoped this will help to answer increasing calls for how qualitative studies can better inform journalistic practice.