19
WHAT‘S IN A FRAME? A CONTENT ANALYSIS OF MEDIA FRAMING STUDIES IN THE WORLD’S LEADING COMMUNICATION JOURNALS, 1990-2005 By Jorg Matthes This paper provides a systematic analysis of media framing studies in the world’s leading communication journals. A quantitative content analysis of 131 studies published in fifteen international journals demonstrates how frames are conceptualized and measured. Current problems in framing research include lack of operational precision, the descriptive focus of many analyses, neglect of visuals, and insuficient reporting of reliability. As stated in Journalism & Mass Communication Quarterly, ”One of the most fertile areas of current research in journalism and mass commu- nication involves the concept of ‘framing.”” Fueled by Entman’s seminal paper,2frame analysis has become a lively and important methodology. In essence, frame analysis examines the selection and salience of certain aspects of an issue by exploring images, stereotypes, metaphors, actors, and messages. Following DAngelo, framing is more a research program than a unified paradigm? As such, framing combines ”strange bedfel- lows that differ in important philosophical ass~mptions.”~ The diversi- ty of theoretical perspectives-cognitive, constructivist, and critical- has been beneficial in enabling a comprehensive understanding of all facets of the framing proce~s.~ At the same time, scholars are faced with an immense variety of theoretical and operational understandings of frames. This study attempts to go beyond offering the simple observation that terms are used inconsistently, and reports a systematic content analysis of media framing studies in leading communication journals. The aim is to show precisely how news frames are conceptualized and operationalized in the literature, thus providing insights to critically reflect what framing really is and what sites of fracture exist. This endeavor bears great theo- retical importance because, as Reese has noted, framing authors “often give an obligatory nod to the literature before proceeding to do whatev- Jorg Matthes is an assistant professor for political communication and political behavior I~MC auaderly at the Institute of Mass Communication and Media Research, University of Zurich, vol.86,No.z Switzprland. An earlier version of this paper was presented at the annual ICA Conference S t ~ m ~ 2 ~ 9 May 24-28,2007, in San Francisco. The author wishes to thank two anonymous review- ~ ~ ~ ~ m m c ers and associate editor Julie Andsager for their insightful comments and suggestions. WHAT’S IN A FRAME? 349

What's in a Frame?

Embed Size (px)

DESCRIPTION

Matthes 2009

Citation preview

Page 1: What's in a Frame?

WHAT‘S IN A FRAME? A CONTENT ANALYSIS OF MEDIA FRAMING S T U D I E S IN THE WORLD’S LEADING COMMUNICATION JOURNALS, 1990-2005 By Jorg Matthes

This paper provides a systematic analysis of media framing studies in the world’s leading communication journals. A quantitative content analysis of 131 studies published in fifteen international journals demonstrates how frames are conceptualized and measured. Current problems in framing research include lack of operational precision, the descriptive focus of many analyses, neglect of visuals, and insuficient reporting of reliability.

As stated in Journalism & Mass Communication Quarterly, ”One of the most fertile areas of current research in journalism and mass commu- nication involves the concept of ‘framing.”” Fueled by Entman’s seminal paper,2 frame analysis has become a lively and important methodology. In essence, frame analysis examines the selection and salience of certain aspects of an issue by exploring images, stereotypes, metaphors, actors, and messages. Following DAngelo, framing is more a research program than a unified paradigm? As such, framing combines ”strange bedfel- lows that differ in important philosophical ass~mptions.”~ The diversi- ty of theoretical perspectives-cognitive, constructivist, and critical- has been beneficial in enabling a comprehensive understanding of all facets of the framing proce~s.~

At the same time, scholars are faced with an immense variety of theoretical and operational understandings of frames. This study attempts to go beyond offering the simple observation that terms are used inconsistently, and reports a systematic content analysis of media framing studies in leading communication journals. The aim is to show precisely how news frames are conceptualized and operationalized in the literature, thus providing insights to critically reflect what framing really is and what sites of fracture exist. This endeavor bears great theo- retical importance because, as Reese has noted, framing authors “often give an obligatory nod to the literature before proceeding to do whatev-

Jorg Matthes is an assistant professor for political communication and political behavior I ~ M C auaderly at the Institute of Mass Communication and Media Research, University of Zurich, vol.86,No.z Switzprland. An earlier version of this paper was presented at the annual ICA Conference S t ~ m ~ 2 ~ 9 May 24-28,2007, in San Francisco. The author wishes to thank two anonymous review- ~ ~ ~ ~ m m c ers and associate editor Julie Andsager for their insightful comments and suggestions.

WHAT’S IN A FRAME? 349

Page 2: What's in a Frame?

er they were going to do in the first place."6 Therefore, this study tries to show empirically what kinds of frames are analyzed and to what extent they share a common understanding. Furthermore, by identifying how framing scholars define and code frames in content analysis, we may see strengths and weaknesses in terms of reliability and validity. Thus, this article aims to provide a basis for a critical self-reflection on framing research.

Literature Review

Four aspects of conceptualizing and coding frames can be addressed: (1) definitions and how they are used for operationalization, (2) the type of frames, (3) use of theory, and (4) the methods of frame analysis.

Definitions. There are two basic genres of definiti~n.~ General defi- nitions describe the term "frame" without clear guidelines for opera- tionalization. For instance, Gitlin described frames as "principles of selec- tion, emphasis, and presentation composed of little tacit theories about what exists, what happens and what matters."8 Such general definitions, while useful, leave the explicit operational understanding of the frame concept open. Other definitions specify what frames generally do, such as defining problems, making moral judgments, and supporting remedies? Such definitions provide precise operational guidelines, enabling "infer- ences that distinguish framing from themes, arguments, assertions and other under-theorized concepts."1° Still, operational definitions can be used in different ways: They can be translated to frame indicators or cited to ground the reader in framing literature. The use of frame definitions is, therefore, central to frame validity, i.e., whether scholars really do meas- ure what they intend to measure.

Frame Type. Scholars have conceptualized news frames in different ways at differing discourse units (units of analysis). Some studies use news items or articles as discourse units," some use the proposition,12 yet others focus on visual features.I3 We can also distinguish at least three roles for visual elements. First, text is coded and visuals are ignored.14 Second, visuals are directly coded as a constituent of a frame (as a dis- course unit).I5 Third, visual elements are not the main constituent of a frame but are discussed when interpreting the frame.16 For several units of analysis, scholars have identified single or multiple frames. For instance, Carpenter;17 Kerbel, Apee, and Ross;'~ and Dimitrova and Stromba~k'~ used the news article as discourse unit, but identified multi- ple frames per unit. Thus, for each story, researchers coded "whether each of the [...I frames was present or absent and, if present, the degree to which it was featured in the story."20 Others extract one frame per unit.2l

Frames have been conceptualized at various levels of abstraction; e.g., as issue-specific or generic." Issue-specific frames mean every issue can have different issue-specific frames.= Generic frames transcend the- matic limitations as they can be identified across different issues. Iyengar's thematic and episodic frames are prime examples.24 Semetko and Valkenburg postulated five generic frames: conflict, human interest, economic consequences, morality, and responsibility.25

350 JOURNALISM & MASS COMMUNKATION QUARTERLY

Page 3: What's in a Frame?

Use of Theory. Use of theory is a central aspect when judging content analytical framing research. Some studies derive hypotheses about the structure and nature of Others formulate research questions with an interest in the description of news content.27 Scholars have argued “[mlost of the research on media framing is still fairly descriptive and relatively atheoretical.”28 This assessment may be justi- fied, but without a more systematic analysis, it remains a subjective view.

It is also necessary to examine the ways frames are connected to antecedents and consequences. In fact, one strength of framing research is its ability to bridge several research areas such as the production, con- tent, and effects of n e ~ s . 2 ~ Several connections to antecedents and con- sequences exist in the literature. Some studies focus on content analysis of frames, but at least discuss how frames evolve or what effects they might have.jO Other studies combine analysis of frames with other data sources such as interviews with journalists or other content produ~ers ,~~ factual or public opinion s~rveys.3~

Methods of Frame Analysis. Because methodological approaches are hard to systematize, single methodological steps in frame analysis are distinguished here. More specifically, it is crucial to know (a) whether the analysis is text-based or number-based, (b) whether frames are determined inductively or (c) whether coding is man- ual or computer-assisted, and (d) whether data-reduction techniques are used to reveal frames or whole frames are coded as such. These method- ological features can overlap and result in different approaches.

A number of studies try to identify frames with a text-based, non- quantitative analysis.% Rooted in the qualitative paradigm, these stud- ies are based on relatively small samples that should mirror discourse. Typically, frames are described in-depth, with detailed quotes, but with- out quantification. Most of those studies extract frames inductively,% but others work ded~ctively.3~

Other studies code frames as variables in a quantitative content analysis, both inductively and deductively. In inductive studies, frames are derived from an initial exploratory analysis of a sample, then defined in a codebook and coded in quantitative content analysis?* An example of deductive, quantitative measurement is Iyengar ’s episodic and thematic Yet other quantitative studies do not code frames as such, but reduce several variables to frames with cluster and factor analysis. Van Gorp coded framing devices in a manual content analysis, such as problem source, responsibility, metaphors, lexical choices, or visuals.4o These variables were then cluster analyzed and the two clus- ters treated as frames. Matthes and Kohring proposed a method that codes Entman’s single frame elements in a standard content analysis. A cluster analysis of these elements then reveals the frame.4l

Yet other quantitative studies have applied computer-assisted frame analysis, e.g., Miller and colleagues suggested so-called frame

Assuming frames are manifested in specific words, the authors identified frames by examining words that tend to occur togeth- er with the help of clustering techniques (without manual coding).

WHAT‘S IN A FRAME? 351

Page 4: What's in a Frame?

For example, Miller, Andsager, and Riechert analyzed candidate frames in the 1996 primary elections.43 Some studies have advanced computer- assisted content analysis beyond merely grouping words.& For instance, Shah and colleagues created syntactic rules that capture the meaning of sentences.45

The rather unsystematic review of studies reported above demon- strates the huge variety of approaches. Both text-based and number-based studies extract frames inductively or deductively. In addition, number- based, quantitative studies can vary in regard to coding whole frames vs. using data-reduction techniques or manual or computer-assisted content analysis. From this review, it is hard to tell which procedures dominate the framing literature and how these relate to underlying theoretical assumptions. Thus, we need a more systematic analysis that answers these questions.

Research Questions

As should be obvious from this review, many things can be “in” a frame. Central to our understanding of how scholars conceptualize frames are the questions of which definitions are cited and how they are operationalized. This leads to the first research question:

RQ1: Which definitions are used for the operationaliza- tion of media frames, and how are they used?

It is then crucial to know what types of frames are identified, at which levels of abstraction and units of analysis. That raises the question of whether single dominant or several frames are identified within a news item. We also need to know if and how visual material constitutes frames. Thus:

RQ2: What types of frames are identified in content ana- lytical literature?

RQ3: To what extent are visual elements of news items considered for the identification of frames?

Closely related to the types of frames is the use of theory in framing research. The more advanced a theory is, the more deductively it will be pursued. Therefore, we need to know to what extent framing research is theory-driven. One basic indicator of that is the use of hypotheses in con- trast to research questions. Moreover, the discussion of antecedents and consequences of news framing is relevant to framing theory.“ By examin- ing the origins and effects of frames, scholars could move beyond a mere description of media content, thus ultimately advancing the understand- ing of frames.

RQ4: To what extent is media framing research theory- driven, and to what extent are consequences and antecedents addressed?

352 JOURNALISM 6. MASS COMMUNICATION QUARTERLY

Page 5: What's in a Frame?

Methods of frame analysis and their reliability are important, as method needs to be analytically distinguished from frame type. Different types of frames can be extracted with varying methods. The frame type describes what kinds of frames are explored. In contrast, the method describes how researchers extract these frames from the materi- al. This ”how” of frame extraction is important to the establishment of reliability. In order to contextualize the conceptualization of frames, we can ask how different methodological approaches relate to the theoreti- cal understanding of frames, the type of frames, and the question of intercoder reliability. This leads to the final research question:

RQ5: What are the most prevalent methods of frame analysis, and how do they relate to the theoretical conceptu- alization of frames and reporting of intercoder reliability?

This study analyzed all articles (including research notes) that have reported a content analysis of media frames published in fifteen journals. Selection of journals was guided by three principles: they must have a focus on communication; they should include different episte- mological perspectives; and the study of framing research should include an international perspective. Following these principles, fifteen journals were selected. The major communication journals were Journalism & Mass Communication Quarterly (N = 20 articles), Journal of Communication (9), Communication Research (4), Political Communication (23), Mass Communication & Society (7), Haward International Journal of Press/Politics (lo), and American Behavioral Scientist (4). To integrate jour- nals focusing on critical perspectives, the journals Critical Studies in Media Communication (7), Journal of Communication Inquiry (5), and Journalism: Theory, Practice b Criticism (6) were included. The major European journals European Journal of Communication (5), International Communication Gazette (13), and Communications: The European Journal of Communication Research (3) were analyzed. Finally, the Canadian Journal of Communication (8) and the Asian Journal of Communication (7) were selected as representatives for Canada and Asia. Because framing research evolved in the 1990s, journals were analyzed from January 1990 to December 2005.

For each journal, articles that identified, named, and extracted media frames in a content analysis were selected. Articles merely using the frame metaphor were excluded. Keywords, abstracts, and parts of articles were read in order to identify relevant articles. If available, search engines of the journals were searched for ”frame” and “framing.” This procedure resulted in 131 articles.

Coding Instrument. Descriptive variables included media sources analyzed and time frame of coverage. Two phases (1990-1999: n = 36; 2000-2005: n = 95) of studies were distinguished.

Conceptual variables included the type of frame (issue-specific or generic), unit of analysis, the number and names of extracted frames, citation of as many as three definitions of ”frames,” and coding of mul-

Method

WHAT’S IN A FRAME? 353

Page 6: What's in a Frame?

tiple or single frames per discourse unit. Coding determined whether a definition was explicitly translated to frame extraction or if the definition was cited just to ground the 1~ader.4~ For instance, if the conceptual defi- nition of frames included attributions but none were included for opera- tionalization, then the definition was used for grounding. In contrast, when clear coding instructions were derived from the definition, then the definition did undergird frame extraction. For example, Lawrence identi- fied the game frame from the literature and coded it as such; Kensicki operationalized frames explicitly following Entman’s definition.@ The part of the article in which the definitions were cited was coded.

For use of visuals, three levels were coded: (a) text- or word-based coding (no visuals), (b) explicit coding of visual material as frames, and (c) use of visual material to contextualize and interpret the frame. These codes were derived from an iterative analysis of all articles examining visuals (n = 22).

Theory variables included the use of hypotheses and hypothesis tests rather than research questions and descriptive re~ults.4~ Antecedents or consequences of framing were addressed. Antecedents were coded as: (a) merely discussed (without presenting other data), (b) interviews with journalists presented, (c) factual data presented, or (d) press releases ana- lyzed. Similarly, consequences of framing could merely be discussed, or survey and experimental data could be presented.

Method variables included sampling, intercoder reliability, and method of frame extraction. For the latter, broad typologies like quantita- tive vs. qualitative methods50 or the typology proposed by Matthes and Kohring51 were not applied because of their heuristic nature. Instead, frame derivation was coded as performed based on text or on numbers; inductive or deductive extraction; whether data-reduction techniques were used to reveal the frame; and whether coding was performed man- ually or computer-assisted (word frequency lists, keyword-in-context lists, or syntactical word relationships).

The author performed the Krippendorff’s alphaB was computed, yielding sufficient reliability scores.” Because a full census of articles was conducted, no tests of statistical significance were performed.

Results Basic Descriptive Results. Fifty-three percent of studies examined newspaper coverage, 13% analyzed TV, 10% TV and newspaper, 8% mag- azines, 5% newspapers and magazines, 2% online news, 2% press releas- es and newspaper coverage, 1% videos, and the remaining 6% a mixture of media. The number of analyzed articles ranged from 1 to 42,695. Most studies used a purposive sample (79%), 5% worked with random samples, 2% with a constructed week, 3% with other samples, and for 12% no infor- mation about sampling was provided.

Definitions and Types of Frames. RQ1 dealt with definitions. E n t m a n ’ ~ ~ ~ definition of media frames was most influential (see Table l), followed by Gamson and Modigliani’s,56 Gitlin’s,” and Iyengar ’~ .~~ However, citing a definition does not necessarily mean that it is used as a direct guide for operationalization. Only 20.5% of definitions explicitly

354 JOURNALISM b MASS COMMUNICATION QUARTERLY

Page 7: What's in a Frame?

TABLE 1 Cited De3nitions of Media Frames

Number %

Entman (1993,2004)

Gamson & Modigliani (1987,1989); Gamson (1992)

Gitlin (1980)

51 28.7

36 20.2

21 11.8

Iyengar (1991) 11 6.2

Semetko & Valkenburg (2000); Valkenburg, Semetko & de Vreese (1999)7s 9 5.1

Goffman (1977)79 8 4.5

Cappella & Jamieson (1997)80

Tankard (1991)81

7 3.9

4 2.2

Other 31 17.4

Total 178 100

Note: As some studies reported multiple definitions, as many as three main definitions were coded per article.

undergirded classification, with the remainder used to ground the read- er (91.2% of which are mentioned in the introduction). Narrower defini- tions by Iyengar (81.8%), Cappella and Jamieson (85.7%), and Semetko and Valkenburg (100%) were mostly translated to operationalization. Thus, the translation of frame definitions to frame indicators was fre- quently left unspoken.

RQ2 asked what kinds of frames were operationalized. Most studies (42%) measured 2-3 frames, followed by more than 7 frames (18%), 4-5 frames (17%), 1 frame only (15%), and 6-7 frames (8%). Only 24% of all studies used propositional units as main discourse units, 63% worked with thematic units (i.e., the whole article), 3% analyzed visuals as main units, and for 10% the level of analysis was unclear.59 Overall, 34% extracted more than one frame per discourse unit. For instance, the unit was the article, but several frames could be coded. The majority of all studies coded only one frame per discourse unit (53%), and for 13% this was not detailed. Thus, framing scholars seem to prefer broader lev- els of analysis, and, generally, a dominant frame per news item was coded.

Textual elements were treated as the main constituents of frames, rather than visuals (RQ3). Only 5% had directly coded visuals (3% as the main discourse unit), and 83% completely neglected visuals. However, 12% discussed the use of visuals, mostly when interpreting frames. That means that, although visuals are not the main signifiers of frames, they were treated as complementary elements and used for

WHAT'S IN A FRAME? 355

Page 8: What's in a Frame?

frame interpretation. Interestingly, the dominance of textual information was also true for those studies that analyzed TV content (72% of them ignored visuals).

One of the major conceptual questions in framing research is whether scholars use issue-specific or generic frames. Of the studies, 78% were issue-specific, and 22% reported generic frames. We identified 561 different issue-specific frames and 29 different generic frames. Among generic frames, the conflict frame was most frequently reported (12 stud- ies), followed by the issueframe (9), the thematicframe (9), the attribution of responsibility frame, the economic consequences frame, the episodic frame, and the human interestframe (8 studies each), or the leadership and the morality frame (5 each). Other generic frames were operationalized in very few studies.@

This led to or suggested a useful typology of discourse format and level of frame abstraction.6l Type A studies (n = 5) worked with generic frames coded on a propositional unit of analysis. Put differently, a gen- eric frame (i.e., conflict) can be coded for every proposition. Type B stud- ies ( n = 23) coded generic frames on the article level, e.g., presence or absence of conflict is coded for every article. Likewise, type C studies (n = 26) analyzed issue-specific frames on a propositional level, meaning the presence of issue-specific frames was coded for every single proposition. Type D studies (n = 59), the largest category, coded issue-specific frames on the article level.

This typology allows an accurate classification of framing studies and enables further comparisons!* Type B studies yielded the smallest number of frames (mostly 1-3), type D studies reported an average num- ber of M = 5.44 (sd = 5.66); and type C studies analyzed the highest num- ber of frames (M = 7.84, sd = 9.03). Studies that reported propositional- level frames analyzed fewer items than studies measuring frames at the article level. Analysis of more than 500 items occurred in 29% of type D and 41% of type B studies, but only 19% of type C. Furthermore, type B studies most frequently cited the definitions by Iyengar (n = 9), Semetko and Valkenburg (8), Entman (6), and Cappella and Jamieson (4). In con- trast, type C studies mostly cited Gamson and Modigliani ( n = 13), Entman (13), and others (7). Type D studies mostly cited Entman ( n = 25), Gamson and Modigliani (17), Gitlin (8), Goffman (7), and others (12).

Theoretical Status. RQ4 aims at examining the extent to which media framing studies are theory-driven. The results show that the majority of studies (68%) did not test hypotheses regarding frames. Interestingly, 74% of studies that assess generic frames on the article level (type B studies) and 2 out of 5 type A studies reported hypotheses, com- pared to 30% of type C and 27% of type D. Neglecting the small sample sizes, this hints that generic frames are more suited for hypothesis testing as they can be more easily compared across media sources, media types, or periods of time.63 In contrast, issue-specific frames (no matter the propositional level) tend to result in a rather descriptive focus.

Discussion of antecedents and consequences of frames is another indicator of theory. The majority of studies addressed neither antecedents (79%) nor consequences (80%). Type C studies were less likely to ignore

356 IOURNALISM b MASS COMMUNICATION QUARTERLY

Page 9: What's in a Frame?

antecedents (65% did not take antecedents into account), and type B studies were less likely to ignore consequences of frames (61%) com- pared to the other types of studies. When it comes to antecedents, 13% discussed the frame-building process without presenting other sources of data (i.e., how the frames were crafted by strategic actors). Four per- cent presented interviews with journalists. Two percent included factu- al data, and another 2% analyzed press releases. Similarly, 12% of all studies discussed possible effects of frames, but only 7% presented sur- vey data and less than 1% reported experimental data. Therefore, con- tent analytical framing studies seldom draw connections to other fields of framing research.

Methods for the Analysis of Media Frames. From 1990-1999,19% of studies derived frames deductively from the literature, and from 2000-2005, this number jumped to 37%. Similarly, an increase in hypoth- esis testing appeared: In the first time period, 14% of studies tested hypotheses in regard to media frames, but in the second period 39% per- formed tests. There was also an increase in use of inferential statistics (1990-1999: 14%; 2000-2005: 38%; overall: 31%). These results suggest that quantitatively orientated research is on the rise.

RQ5 results indicate 46% of studies were text-based (non-quanti- tative). Of text-based studies, almost all derived frames inductively from the material (93%), and none conducted computer-assisted analy- sis or use data-reduction techniques. The remaining 54% of studies used quantitative analysis; 47% of these work in an inductive manner. In con- trast to text-based studies, 14% of the quantitative studies performed computer-assisted analysis, and 23% use data-reduction techniques.

To develop a typology of methodological approaches, a hierarchi- cal cluster analysis with the Ward-algorithm was performed.@ Four nominal variables were included text-based vs. number-based, induc- tive vs. deductive, manual coding vs. computer-assisted coding, and data-reduction techniques vs. direct coding of the frame. The elbow-cri- tenon led to a four-cluster solution.ffi Cluster 1, the biggest cluster (n = 80), includes mostly text-based (and a few number-based) studies that extracted frames inductively without use of data-reduction techniques and computer-assisted content analysis. This cluster can be termed inductive qualitative studies, with an interpretative focus. The second largest cluster (n = 33) includes data-based, quantitative studies that mostly used a deductive frame-extraction strategy with manual coding and no data-reduction techniques. Previously defined holistic frames (e.g., thematic or episodic frames) are coded for their presence in a quan- titative content analysis. These can be called deductive quantitative stud- ies. Third, a small cluster (n = 8) is similar to the second, but instead, data-reduction techniques were used for extraction of deductively defined frames. This group of articles is called deductive quantitative-clus- tering studies. The last cluster ( n = 8) is called inductive quantitative com- puter-assisted studies; these included data-reduction techniques and no manual coding.

The typology of frame measurement can be useful in describing how researchers extract frames from content analytical material. Beyond

WHAT’S IN A FRAME? 357

Page 10: What's in a Frame?

TABLE 2 Frame Extraction Methods as a Function of Frame Type

Inductive Deductive Deductive Inductive Total Qualitative Quantitative Quantitative Quantitative

Studies Studies Clustem Computer- Studies Assisted

Studies

TypeA 1 (1.60%) 4 (12.5%) 0 (0.00%) 0 (0.00%) 5 (4.500/0) (generic at propositional level)

TypeB

Type c

TypeD

(generic at article level) 3 (4.80%) 15 (46.9%) 5 (62.5%) 0 (0.00%) 23 (20.7%)

(issue-specific at propositional level) 15 (23.8%) 3 (9.40%) 3 (37.5%) 5 (62.5%) 26 (23.4%)

(issue-specific at article level) 44 (69.8%) 10 (31.2%) 0 (0.00%) 3 (37.5%) 57 (51.4%)

Total 63 (100%) 32 (100%) 8 (100%) 8 (100%) 111 (100%)

Note: Total N does not equal 131 due to missing values, i.e., insuffiaent information in those studies to code this information.

that, it is interesting to see how these methodological choices relate to the theoretical understanding and conceptualization of frames. Unfortunately, the small sample sizes do not allow strong generalizations. However, as Table 2 shows, some basic patterns appear. Inductive quali- tative studies tend to analyze issue-specific frames (types C and D), while deductive quantitative studies prefer the article as the unit of analysis. Deductive quantitative-clustering studies mostly work with generic frames at the article level and inductive quantitative computer-assisted studies analyze issue-specific frames at the propositional level.

Reliability Reporting. Fifty-five percent of articles did not report intercoder reliability, 21% reported simple agreement, 11% used Holsti’s formula, 6% Scott’s pi , 3% Krippendorff’s Alpha, and the remaining 4% documented other formulae. When reliability was reported, scholars usu- ally documented an overall value (70%) or a range of values (18%) rather than frame-by-frame (10%) or variable-by-variable reliabilities (2%). From those 57 studies that reported intercoder reliability, 50 included informa- tion about the number of coders, 39 documented the sampling for the reli- ability test, and 32 reported the test’s sample size. From 1990-1999, 30% of all studies documented reliability, and from 2000-2005 it was 53%. Nevertheless, compared to other studies that investigated the reliability data in scholarly articles, these numbers are a cause for concern.&

Three explanations occur for the lack of reliability measures: First, the method of frame extraction might play a role; second, reliability may vary with the type of frame analyzed; and third, journals might have dif-

358 ~ O U R N A L I S M b MASS COMMUNICATION QUARTERLY

Page 11: What's in a Frame?

TABLE 3 Reliability Reporting as a Function of Frame Type

Type A TypeB TypeC TypeD Total

No Report of Reliability 1 (20.0%) 7 (30.4%) 11 (55.0%) 31 (55.4%) 50 (48.1%)

Report of Reliability 4 (80.0%) 16 (69.6%) 9 (45.0%) 25 (44.6%) 54 (51.9%)

Total 5 (100%) 23 (100%) 20 (100%) 56 (100%) 104 (100%)

Note: Type A = generic at propositional level; Type B = generic at article level; Type C = issue-specific at propositional level; Type D = issue-specific at article level (see text for explanation). Total N does not equal 131 due to missing values, i.e., insufficient information in those studies to code this aspect.

ferent standards. To begin with the methodological approach, it is strik- ing that 68% of inductive qualitative studies, 24% of deductive quanti- tative studies, and 43% of deductive quantitative-clustering studies fail to report 1~liability.6~ Table 3 shows that studies working with generic frames are more likely to report reliability than studies analyzing issue- specific frames. For issue-specific frames, reliability reporting does not vary with the unit of analysis (type C and D).

Finally, crucial differences appear among the journals in which the studies are published. Sixty percent of studies published in "American flagship journals''68 documented reliability in contrast to 6% of studies in "critical journals."69 European journals7" plus Asian and Canadian journals are in the middle (48%).7l

Much has been written about framing, but no systematic, stan- dardized analysis of the research literature has been conducted. This study aimed to provide more systematic, fine-grained knowledge of how researchers understand and measure frames. The results reveal interesting descriptive patterns, but also some problems. In particular, framing researchers need to critically reflect upon five points to move this burgeoning research area forward.

First, translation of framing definitions to concrete, operational steps is not transparent in a huge part of the literature. Some definitions are general, giving little information about how to operationalize frames. Other definitions provide more precise operational steps but are not always explicitly followed. This is mostly true for Entman's fre- quently cited definition distinguishing frame elements, such as problem definition, causal interpretation, moral evaluations, and /or treatment recommendations. If frames are defined this way, then moral evalua- tions, for instance, should be a part of the frame and coded as such. Clearly, these things need to be made transparent in frame analysis, and single operational steps-that is, the translation of a definition to the exact operationalization-must be explicitly stated. This is not to say

~ i ~ ~ ~ ~ ~ i ~ ~

WHAT'S IN A FRAME? 359

Page 12: What's in a Frame?

that framing scholars failed to have these theoretical ideas, but we do need to make them salient. The typology of framing conceptualizations developed in this study could be helpful for that. Ultimately, this would clarify what needs to be present so we can consider a frame as existing as a unit of analysis.

Second, a clear conceptual divide exists between issue-specific frames and generic frames. Some generic frames describe structural fea- tures of news items, such as conflict or personalization. Others are related to features of topics and issues, such as the economic or the morality frame. We need ”to specify how general a frame must be in order to be classified as generic.”” We also need a conceptual discussion about the differences between generic frames, such as conflict, and news values. This is not to say that generic frames are unable to yield useful insights. However, the term “frame” should be used carefully, and we need a fur- ther scholarly debate about what a frame really is and is not.

Similarly, theorists need to consider whether the frames in media content conceptually match with communicator frames or audience frames. Although framing profits from conceptual and paradigmatic diversity,” scholars must share a common definition when investigating processes of frame building and framing effects. For instance, the notion of contesting frames is inherent to the strategic framing of communica- t o r ~ . ~ ~ Translating this idea of frame contest to the analysis of single news articles, we cannot u priori decide that articles will always have a domi- nant, issue-specific frame. Moreover, strategic framing scholars are most- ly not interested in generic frames, but in issue-specific, evaluative frames. It appears that content researchers analyze frames that strategic framing scholars might regard as rather irrelevant.

Third, most studies are descriptive, not testing any hypotheses regarding framing theory. This is especially true for issue-specific frames. Without doubt, descriptive studies are of great value to the field, and framing is a very useful concept to contextualize and describe media con- tent. However, to advance a theory of framing as a major concept within the field of communication, a less descriptive strategy is necessary. This is also true for discussing the antecedents and consequences of frames.

Fourth, we do not know much about how visual elements are incorpo- rated in frames, though they are arguably quite important for framing theo- ry.” As Graber wrote, “purely verbal analyses not only miss the information contained in the pictures and nonverbal sounds, they even fail to interpret the verbal content appropriately because that content is modified by its com- bination with picture messages.”76 Clearly, framing scholars should con- tribute to the general discourse in the field about the role of visuals.

Fifth, the lack of reliability reporting in a vast amount of studies poses a problem for framing research. As Riffe and Freitag argued, relia- bility is a “necessary condition for validity, and improved reliability reporting is progress.”n Reliability reporting varies with method, frame type, and journal, but it has increased. However, scholars must provide more information about the type of coefficients used, the samples for reli- ability tests, and the number of coders. More rigorous reliability formulae should also be used.

360 JOURNALISM 6 MASS COMMUNICATION QUARTERLY

Page 13: What's in a Frame?

In sum, we have gained insights about the state of the art in media framing research. Some are merely descriptive; others can be used as a basis for further scholarly debate. However, there are limitations to this study. Among these, perhaps the most critical is that the results of this study are influenced by the selection of journals, excluding book chap- ters and monographs. In fact, many of the problems revealed in this arti- cle have been addressed in those publications. Furthermore, the cate- gories used here had to be general enough so that they could be applied to every study sampled, possibly not doing justice to the richness of individual studies.

A related shortcoming is the exclusion of (non-content analytical) studies dealing with framing effects and strategic frames of communi- cators. To understand how media frames are related to other contexts, these studies could provide rich insights. Finally, sample sizes for the comparisons made in this paper are sometimes small, and therefore, some results have to be treated with a good deal of caution. It should be noted that the sample includes the first studies on framing published in our field. As some results indicate, many things have changed over the years, and more recent studies might have solved the conceptual and methodological problems of framing research. Their number, however, is still too small to leave a significant mark in this analysis. Indeed, the present analysis might contribute to that change by pushing framing scholars to move forward in order to fertilize, to advance, and to chal- lenge framing research for the next fifteen years.

Conclusion

NOTES

1. Daniel Riffe, “An Editorial Comment,” Journalism & Mass Com- munication Quarterly 81 (spring 2004): 2.

2. Robert M. Entman, ”Framing: Toward Clarification of a Fractured Paradigm,” Journal of Communication 43 (autumn 1993): 51-58.

3. Paul DAngelo, ”News Framing as a Multi-paradigmatic Re- search Program: A Response to Entman,” Journal of Communication 52 (December 2002): 870-88.

4. Stephen D. Reese, ”The Framing Project: A Bridging Model for Media Research Revisited,” Journal of Communication 57 (March 2007): 149.

5. DAngelo, ”News Framing.” 6. Reese, “The Framing Project,” 152. 7. Robert M. Entman, Jorg Matthes, and LYM Pellicano, “Nature,

Sources, and Effects of News Framing,” in Handbook of Journalism Studies, ed. Karin Wahl-Jorgenson, and Thomas Hanitzsch (New York, NY Routledge, 2009), 175-90.

8. Todd Gitlin, The Whole World is Watching: Mass Media in the Making and Unmaking of the New Left (Berkeley: University of California Press, 1980), 7. Another example is Gamson and Modigliani’s frequent- ly quoted definition of framing as the ”central organizing idea or story

WHAT‘S IN A FRAME? 361

Page 14: What's in a Frame?

line that provides meaning to an unfolding strip of events.” See William A. Gamson and Andre Modigliani, “The Changing Culture of Affirmative Action,” in Research in Political Sociology, ed. Richard G. Braungart and Margaret M. Braungart (Greenwich, CT JAI Press, 1987), 143.

9. Entman, ”Framing,“ 52, original emphasis. 10. Entman, Matthes, and Pellicano, ”Nature, Sources, and Effects,”

176. 11. L. Paul Husselbee and Larry Elliot, ”Looking Beyond Hate: How

National and Regional Newspapers Framed Hate Crimes in Jasper, Texas, and Laramie, Wyoming,” Journalism 6 Mass Communication Quarterly 79 (winter 2002): 833-52.

12. Zhongdang Pan and Gerald M. Kosicki, ”Framing Analysis: An Approach to News Discourse,” Political Communication 10 (spring 1993):

13. Cyntha King and Paul Lester, ”Photographic Coverage during the Persian Gulf and Iraqi Ears in Three U.S. Newspapers,” Journalism &Mass Communication Quarterly 82 (autumn 2005): 623-37; Lynda Jean Kensicki, ”Deaf President Now! Positive Media Framing of a Social Movement within a Hegemonic Political Environment,” Journal of Communication Inquiry 25 (summer 2001): 147-66.

14. E.g., Jorg Matthes and Matthias Kohring, “The Content Analysis of Media Frames: Toward Improving Reliability and Validity,” Journal of Communication 58 (summer 2008): 258-79.

15. Frank Esser and Paul DAngelo, “Framing the Press and the Publicity Process. A Content Analysis of Metacoverage in Campaign 2000 Network News,” American Behavioral Scientist 46 (January 2003): 617-41; Marie Hardin, Susan Lynn, Kristie Walsdorf, and Brent Hardin, ”The Framing of Sexual Difference in SI for Kids Editorial Photos,” Mass Communication & Society 5 (August 2000): 341-59; King and Lester, ”Photographic Coverage”; Kensicki, “Deaf President Now!”

16. John Parmelee, ”Presidential Primary Videocassettes: How Candidates in the 2000 U.S. Presidential Primary Elections Framed Their Early Campaigns,” Political Communication 19 (autumn 2002): 317-31; Jisuk Woo, “Television News Discourse in Political Transition: Framing the 1987 and 1992 Korean Presidential Elections,” Political Communication 13 (spring 1996): 63-80; Zhongdang Pan, Chin-Chuan Lee, Joseph Man Chan, and Clement Y.K. So, “One Event, Three Stories: Media Narratives of the Handover of Hong Kong in Cultural China,” Gazette: The International Journal for Communication Studies 61 (summer 1999): 99-112; Federico Boni, ”Framing Media Masculinities. Men’s Lifestyle Magazines and the Biopolitics of the Male Body,” European Journal of Communication 17 (winter 2002): 465-78.

17. Serena Carpenter, ”U.S. Elite and Non-elite Newspapers’ Portrayal of the Iraq War: A Comparison of Frames and Source Use,” Journalism & Mass Communication Quarterly 84 (winter 2007) 761-76.

18. Matthew R. Kerbel, Sumaiya Apee, and Marc Howard Ross, “PBS Ain’t So Different. Public Broadcasting, Election Frames, and Democratic Empowerment,” The Harvard lnternational Journal of PresslPolitics 5 (2000 winter): 8-32.

55-76.

362 ~OURNALISM b MASS COMMUNICATION OUARTERLY

Page 15: What's in a Frame?

19. Daniela V. Dimitrova and Jesper Stromback, “Mission Accom- plished? Framing of the Iraq War in the Elite Newspapers in Sweden and the United States,“ Gazette: The International Journal for Communication Studies 67 (October 2005): 399-417.

20. Kerbel, Apee, and Ross, “PBS Ain’t So Different,” 15. 21. Husselbee and Elliot, “Looking Beyond Hate.” 22. Claes H. de Vreese, ”News Framing: Theory and Typology,”

Information Design Journal + Document Design 13 (2005): 48-59. 23. Laureen R. Tucker, “The Framing of Calvin Klein,” Critical Studies

in Mass Communication 15 (summer 1998): 141-57; Stephen Reese and Bob Buckalew, ”The Militarism of Local Television: The Routine Fram- ing of the Persian Gulf War,” Critical Studies in Mass Communication 12 (March 1994): 40-59; Dhavan V. Shah, Mark D. Watts, David Domke, and David P. Fan, ”News Framing and Cueing of Issue Regimes: Explaining Clinton’s Public Approval in Spite of Scandal,” Public Opinion Quarterly 66 (fall 2002): 339-70.

24. Shanto Iyengar, Is Anyone Responsible? How Television Frames Political Issues (Chicago: University of California Press, 1991).

25. Holli A. Semetko and Patti M. Valkenburg, ”Framing European Politics: A Content Analysis of Press and Television News,” Journal of Communication 50 (spring 2000): 93-109.

26. Paul DAngelo, Matthew Calderone, and Anthony Territola, ”Strategy and Issue Framing: An Exploratory Analysis of Topics and Frames in Campaign 2004 Print News,” Atlantic Journal of Communi- cation 13 (winter 2005): 199-219.

27. Joshua Greenberg, ”Opinion Discourse and Canadian Newspa- pers: The Case of the Chinese ‘Boat People,“‘ Canadian Journal of Com- munication 25 (winter 2000): 517-37.

28. David Roskos-Ewoldsen, “Review of the Book ’Framing Public Life: Perspectives on Media and our Understanding of the Social World,”’ Journal of Communication 53 (December 2003): 340.

29. Entman, ”Framing.” Antecedents and consequences are also taken into account in other meta-studies, e.g., Daniel Riffe and Allan Freitag, “A Content Analysis of Content Analyses: Twenty-Five Years of Journalism Quarterly,“ Journalism &Mass Communication Quarterly 74 (au-

30. Robert M. Entman, “Cascading Activation: Contesting the White House’s Frame after 9/ 11,” Political Communication 20 (winter 2003): 415- 32; Regina M. Marchi, “Reframing the Runway. A Case Study of the Impact of Community Organizing on News and Politics,” Journalism 6 (winter 2005): 465-85.

31. Trudie Richards and Brent King, “An Alternative to the Fighting Frame in News Reporting,” Canadian Journal of Communication 25 (win- ter 2000): 479-96; Parmelee, ”Presidential Primary Videocassettes.”

32. Hans-Bernd Brosius and Peter Eps, “Prototyping through Key Events. News Selection in the Case of Violence against Aliens and Asylum Seekers in Germany,” European Journal of Communication 10 (September 1995): 391-412; John D. Richardson and Karen M. Lancendorfer, ”Framing Affirmative Action. The Influence of Race on

tumn 1997): 515-24.

WHAT’S IN A FRAME?

Page 16: What's in a Frame?

Newspaper Editorial Responses to the University of Michigan Cases,” Harvard International Journal of Press/Politics 9 (October 2004): 74-94.

33. Claes H. de Vreese, ”The Spiral of Cynicism Reconsidered,” Euro- pean Journal of Communication 20 (September 2005): 283-301; Rens Vliegenthart, Andreas R. T. Schuck, Hajo G. Boomgaarden, and Claes H. De Vreese, “News Coverage and Support for European Integration, 1990-2006,“ International Journal of Public Opinion Research 20 (winter

34. The terms ”inductive” and ”deductive” are used in a restricted sense here. They refer to frame extraction only. Inductive means that frames are generated as a result of the analysis; they are not (theoretical- ly) derived beforehand. Deductive means that pre-defined frames are coded, no new frames are generated. Thus, both terms do not refer to a study’s general reasoning or to broad epistemological orientations. Studies that are treated as inductive (or deductive) here do not necessari- ly take a purely inductive (or deductive) approach.

35. Douglas Downs, “Representing Gun Owners. Frame Identification as Social Responsibility in News Media Discourse,” Written Communi- cation 19 (January 2002): 44-75; Alice Hall, “The Mass Media, Cultural Identity, and Perceptions of National Character: An Analysis of Frames in US and Canadian Coverage of Audiovisual Materials in the GATT,” Gazette: The International Journal for Communication Studies 62 (summer 2000): 231-49; Reese and Buckalew, “The Militarism of Local Television”; Tucker, “The Framing of Calvin Klein.”

36. E.g., Nancy Worthington, “A Division of Labor: Dividing Maternal Authority from Political Activism in the Kenyan Press,” Journal of Com- munication Inquiry 25 (April 2001): 167-83.

37. Roya Akhavan-Majid and Jyotika Ramaprasad, ”Framing Beijing. Dominant Ideological Influences on the American Press Coverage of the Fourth UN Conference on Women and the NGO Forum,” Gazette: The International Journal for Communication Studies 62 (February 2000): 45-69.

38. Adam Simon and Michael Xenos, “Media Framing and Effective Public Deliberation,” Political Communication 17 (October 2000): 613-24; Husselbee and Elliot, “Looking Beyond Hate.”

2008): 415-39.

39. Iyengar, Is Anyone Responsible? 40. Baldwin van Gorp, ”Where is the Frame? Victims and Intruders in

the Belgian Press Coverage of the Asylum Issue,” European Journal of Communication 20 (fall 2005): 485-50.

41. Matthes and Kohring, ”Content Analysis.” 42. M. Mark Miller, Julie L. Andsager, and Bonnie P. Riechert,

“Framing the Candidates in Presidential Primaries: Issues and Images in Press Releases and News Coverage,” Journalism b Mass Communication Quarterly 75 (summer 1998): 312-24; M. Mark Miller, “Frame Mapping and Analysis of News Coverage of Contentious Issues,” Social Science Computer Review 15 (winter 1997): 367-78; Julie Andsager, “How Interest Groups Attempt to Shape Public Opinion with Competing News Frames,” Journalism b Mass Communication Quarterly 77 (autumn 2000): 577-92.

43. Miller, Andsager, and Riechert, ”Framing the Candidates.”

364 ~OURNALISM b MASS COMMUNICATION OUARTERLY

Page 17: What's in a Frame?

44. Amy E. Jasperson, Dhavan V. Shah, Mark Watts, Ronald J. Faber, and David P. Fan, “Framing and the Public Agenda: Media Effects on the Importance of the Federal Budget Deficit,” Political Communication 15 (March 1998): 205-24; Shah et al., “News Framing”; Dennis T. Lowry, ”Network TV News Framing of Good vs. Bad Economic News under Democrat and Republican Presidents: A Lexical Analysis of Political Bias,“ Journalism & Mass Communication Quarterly 85 (autumn 2008): 483- 98.

45. Jasperson et al., ”Framing and the Public Agenda”; Shah et al., “News Framing.”

46. See also Riffe and Freitag, “A Content Analysis,” that used similar variables in their content analysis of content analyses.

47. Since many studies reported more than one definition, up to three definitions were coded. If an operational definition was presented and other definitions were named as well, only the operational definition was coded. If several operational definitions were equally present, they were coded according to their order in the text.

48. Regina L. Lawrence, “Game-Framing the Issues: Tracking the Strategy Frame in Public Policy News,” Political Communication 17 (sum- mer 2000): 93-114; Christiane Eilders and Albrecht Luter, ”Germany at War: Competing Framing Strategies in German Public Discourse,” European Iournal of Communication 15 (September 2000): 415-28; Lynda Jean Kensicki, “No Cure for What Ails Us: The Media-constructed Disconnect between Societal Problems and Possible Solutions,” Journalism 6 Muss Communication Quarterly 81 (spring 2004) 53-73.

49. In contrast to this coding, Riffe and Freitag, “A Content Analysis,” also treated research questions as indicators of theory. However, hypothe- ses aim at theory testing and research questions aim at theory develop- ment. Thus, hypotheses are a stronger indicator of a theory‘s presence than research questions. Furthermore, it was not distinguished if a study reports both hypotheses and research questions. When hypotheses were present, ”hypothesis use = 1” was coded no matter if research questions were also present. When no hypothesis were present, “hypothesis use = 0 was coded.

50. James W. Tankard, ”An Empirical Approach to the Study of Media Framing,“ in Framing Public Life: Perspectives on Media and Our Understanding of the Social World, ed. Stephen D Reese, Oscar H Gandy, and August E Grant (Mahwah, NJ: Lawrence Erlbaum Associates, 2001), 98.

51. Matthes and Kohring, “Content Analysis.” 52. In a first full round of coding, a reliability check of an 8% random

sample with a second coder yielded insufficient reliability. Therefore, all problematic variables were carefully recoded in an extensive second round of coding working with a more refined codebook. Before the sec- ond round, a second 10% random sample reliability check was conducted with a graduate student who underwent extensive training on the code- book.

53. Andrew F. Hayes and Klaus Krippendorff, ”Answering the Call for a Standard Reliability Measure for Coding Data,” Communication Methods

WHAT’S IN A FRAME? 365

Page 18: What's in a Frame?

and Measures (1, 2007): 77-89. 54. This resulted in reliability of a = .98 for the identification variables

(average score), a = .82 for the type of frame, a = .75 for the unit of analy- sis, a = .95 for the number of frames, a = .82 for the definition of frames (average over three definitions), a = .75 for the presence of multiple or single frames, a = .75 for use of definition (average over three definitions), a = .80 for use of visuals, a = .90 for hypothesis tests, a = .80 for antecedents and consequences (average), and an average a = .83 for method variables.

55. Entman, ”Framing”; Robert Entman, Projections of Power: Framing News, Public Opinion and US. Foreign Policy (Chicago: University of Chicago Press, 2004).

56. Gamson and Modigliani, “The Changing Culture of Affirmative Action“; William A. Gamson and Andre Modigliani, “Media Discourse and Public Opinion on Nuclear Power: A Constructionist Approach,” American Journal of Sociology 95 (July 1989): 1-37; also William A. Gamson, Talking Politics (Cambridge: Cambridge University Press, 2002).

57. Gitlin, The Whole World is Watching. 58. Iyengar, Is Anyone Responsible? 59. This problem occurred when frames were just described without

60. It was not distinguished, however, if generic frames were given

61. The author wishes to thank an anonymous reviewer for suggesting

62. Studies coding visual units only are excluded for this typology due to small sample size. Type A studies are not further discussed for the same reason.

63. However, this suggestion cannot be tested with the present data because of the small sample sizes.

64. The aim of this cluster analysis is to find groups of articles that have similar (not necessarily identical) approaches. Of course, grouping articles might do injustice to the uniqueness of methods that are applied in single studies. But cluster analysis is the method of choice to develop a broad, standardized, and data-based typology.

65. Merging the fourth with the third cluster would result in a step increase of heterogeneity within the cluster.

66. Riffe and Freitag, “A Content Analysis”; Matthew Lombard, Jennifer Snyder-Duch, and Cheryl Campanella Bracken, “Content Analysis in Mass Communication: Assessment and Reporting of Inter- coder Reliability,” Human Communication Research 28 (October 2002): 587- 604. Riffe and Freitag analyzed 486 content analytical studies in Journalism b Mass Communication Quarterly from 1971-1995. According to their results, 56% of all studies reported intercoder reliability. For the period of 1991-1995, however, 72% report reliability data. In a similar study, Lombard, Snyder-Duch, and Bracken assessed the reporting of reliability data in 200 studies in the mass communication literature between 1994 and 1998. They found that 69% of all articles contained any report of inter- coder reliability.

detailing how they were coded or extracted.

different names but meant the same thing. This is hard to reliably code.

this typology.

366 JOURNALISM 6 MASS COMMUNICATION QUARTERLY

Page 19: What's in a Frame?

67. There is, of course, no reliability reporting in computer-assisted content analysis.

68. Journalism b Mass Communication Quarterly, Journal of Communi- cation, Communication Research, Political Communication, Mass Communi- cation b Society, Harvard International Journal of PresslPolitics, American Be- havioral Scientist.

69. Critical Studies in Mass Communication, Journal of Communication Inquiry, Journalism.

70. European Journal of Communication, Infernational Communication Gazette, Communications.

71. Although the small sample sizes must be taken into account, it can be stated that Journalism b Mass Communication Quarterly is a benchmark journal in terms of reliability reporting: 80% of all articles published in that journal include details about intercoder reliability.

72. Dennis Chong and James N. Druckman, "Framing Theory," Annual Review of Political Science 10 (2007), 107.

73. DAngelo, "News Framing." 74. Robert D. Benford and David A. Snow, "Framing Processes and

Social Movements: An Overview and Assessment," Annual Review of Sociology 26 (2000): 611-39.

75. Robert M. Entman, "Framing U.S. Coverage of International News: Contrasts in Narratives of the KAL and Iran Air Incidents," Journal of Communication 41 (winter 1991): 7; Pan and Kosicki, "Framing Analysis," 61.

76. Doris Graber, "Content and Meaning. What's It All About?" American Behavioral Scientist 33 (November /December 1989): 144-52.

77. Riffe and Freitag, "A Content Analysis," 522. 78. Patti M. Valkenburg, Holli A. Semetko, and Clees H. de Vreese,

"The Effects of News Frames on Readers' Thoughts and Recall," Communication Research 26 (October 1999): 550-69.

79. Erving Goffman, Frame Analysis: An Essay on the Organization of Experience (New York Harper & Row, 1974), 21.

80. Joseph N. Cappella and Kathleen Hall Jamieson, Spiral of Cynicism. The Press and the Public Good (New York, Oxford: Oxford University, 1997).

81. James W. Tankard, Laura Handerson, Jackie Sillberman, Kriss Bliss, and Salma Ghanem, "Media Frames: Approaches to Conceptualization and Measurement" (paper presented at the AEJMC annual convention, Boston, MA, 1991).

WHAT'S IN A FRAME? 367