REFLECTIONSEthnographic Content AnalysisDavid L. AltheideArizona State UniversityABS TRACT. An ethnographic approach to document analysis is offered based on principles of qualitative data collection and analysis. It is proposed that numeric as well asnarrative data be collected when studying such documents as TV new and movies. Ethnographic content analysis is briefly contrasted with conventional modes of quantitativecontent analysis to illustrate the usefulness of constant comparison for discovering emergentpatterns, emphases and themes in an analysis of TV news coverage of the Iranian hostagesituation. It is suggested that an ethnographic perspective can help delineate patterns ofhuman action when document analysis is conceptualized as fieldwork.It has been claimed that all research directly or indirectly involvesparticipant observation in the selection of a topic, method of study, datacollection, analysis and interpretation (cf. Cicourel, 1964; Johnson, 1975;Hammersley and Atkinson, 1983). While it may seem evident that anysustained inquiry is constituted through a complex and reflexive interaction process, it is also apparent that some research methods, e.g., ethnography, embrace this process; while others, e.g., survey research andcontent analysis, disavow it. In what follows, I suggest that severalaspects of an ethnographic research approach can be applied to contentanalysis to produce ethnographic content analysis (ECA), which may bedefined as the reflexive analysis of documents (cf. Plummer, 1983). ECAhas been less widely recognized as a distinctive method, although variousfacets of this approach are apparent in document analyses by historians,literary scholars, and social scientists (cf. Plummer, 1983; Glaser andStrauss, 1967). However, to my knowledge, the method foraccomplishing such “grounding” with documents has not been set forthThis draft has benefitted from comments and suggestions by Robert Emerson, ShulamitReinharz, and three anonymous reviewers. An earlier draft of this paper was presentedat the Annual Meeting of the Pacific Sociological Association, Albuquerque, NM, April17—20, 1985. Address requests for reprints to: School of Justice Studies, Arizona StateUniversity, Tempe, AZ 85287.Qualitative Sociology, 10(1), Spring 198765 1987 Human Sciences Press

66QUALITATIVE SOCIOLOGY(cf. Starosta, 1984). A brief comparison of this approach with conventional content analysis will be proceeded with examples of the use ofECA in several research projects.Ethnography in ContextIn general, ethnography refers to the description of people and theirculture (cf. Schwartz and Jacobs, 1979). In this sense, the subjectmatter—human beings engaged in meaningful behavior—guide the modeof inquiry and orientation of the investigator. However, if the meaning ofan activity remains paramount, ethnography can also be considered amethodological orientation independently of a specific subject matter.Products of social interaction, for example, can also be studied reflexively, looking at one feature in the context of what is understood aboutother features, allowing for the constant comparison suggested by Glaserand Strauss (1967).An Overview of Content AnalysisEthnographic content analysis (ECA) may be contrasted with conventional, or more quantitative content analysis (QCA), in approach to datacollection, data analysis and interpretation. Table 1 provides an overviewof these approaches on several dimensions.Quantitative Content Analysis (QCA)Originating in positivistic assumptions about objectivity, QCA provides a way of obtaining data to measure the frequency and variety ofmessages (cf. Berelson, 1966). QCA analysis has been used to determinethe “objective” content of written and electronic documents, e.g., TV cartoons (cf. McCormack, 1982). As summarized by Starosta (1984:185),Content analysis translates frequency of occurrence of certain symbols intosummary judgments and comparisons of content of the discourse.whatever “means” will presumably take up space and/or time; hence, thegreater that space and/or time, the greater the meaning’s significance.Units of space most commonly are seen as countable, and therefore, mea

Reflections: Ethnographic Content AnalysisTABLE 1Quantitative (QCA) and Ethnographic (ECA) Content AnalysisQCAVerificationECADiscovery; VerificationReflexive ityResearch GoalProgression from Data SerialCo1lection, Analysis,InterpretationReflexive; CircularPrimary ResearcherInvolvementData Analysis andInterpretationAll PhasesSampleRandom orStratifiedPurposive ing Required toCollect DataLittleSubstantialType of DataNumbersNumbers; NarrativeData Entry PointsOnceMultipleNarrative Descriptionand CommentsSeldomAlwaysConcepts EmergeDuring ResearchSeldomAlwaysData AnalysisStatisticalTextual; StatisticalData PresentationTablesTables and Textsurable. And, even though early proponents of QCA made it clear thatimputation of the speaker’s (writer’s) motive was unwarranted, themethod has been used to relate messages to the source’s intent (Berel3on, 1966).67

68QUALITATIVE SOCIOLOGYQCA is used to verify or confirm hypothesized relationships. Indeed,QCA protocols are usually constructed on the basis of operational definitions of concepts which yield enumerative data for purposes of measurement (cf. Krippendorf, 1980). Research designs were organizedserially, moving from category construction to sampling, data collection,data analysis and interpretation. As this mode of document analysis wasinfluenced by electronic data processing formats, the researcher’s role wasreduced to setting up the protocol, and then analyzing and interpreting thedata. Data collection and organization (coding) was carried out by noviceshired and quickly “trained” to find, record, and count the “mentions” foreach unit of analysis. Measures of “intercoder reliability” were undertakento show that identical judgmental criteria were used in their selection andenumeration. The upshot of this procedure was that reliability producedvalidity. Indeed, this rationale has led to the institutionalization ofintercoder reliability scores in most content analysis studies.Ethnographic Content Analysis (ECA)Ethnographic content analysis is used to document and understandthe communication of meaning, as well as to verify theoretical relationships. Its distinctive characteristic is the reflexive and highly interactive nature of the investigator, concepts, data collection and analysis.Unlike QCA in which the protocol is the instrument, in ECA the investigator is continually central, although protocols may be used in laterphases of the research. Like all ethnographic research, the meaning ofa message is assumed to be reflected in various modes of informationexchange, format, rhythm and style, e.g., aural and visual style, as wellas in the context of the report itself, and other nuances.ECA consists of reflexive movement between concept development,sampling, data collection, data coding, data analysis, and interpretation. The aim is to be systematic and analytic, but not rigid. Althoughcategories and “variables” initially guide the study, others are allowedand expected to emerge throughout the study. Thus, ECA is embeddedin constant discovery and constant comparison of relevant situations,settings, styles, images, meanings and nuances (cf. Glaser and Strauss,1967). To this end, ECA draws on and collects numerical and narrativedata, rather than forcing the latter into predefined categories of the former as is done in QCA. ECA is oriented to check, supplement, and supplant prior theoretical claims by simultaneously obtaining categoricaland unique data for every case studied in order to develop analyticalconstructs appropriate for several investigations (cf. Schwartz and

Reflections: Ethnographic Content Analysis69Jacobs, 1979). Further, data are often coded conceptually so that one itemmay be relevant for several purposes. In short, while items and topics canstill be counted and put in emergent categories, ECA also provides gooddescriptive information.An Example of an ECA StudyEthnographers approach a topic with a wealth of information and understanding about human behavior. Previous work on TV news provided afoundation in news procedures and perspectives which could be reflexivelyincorporated in a study of TV news coverage. The relevance of reflexiveobservation can be illustrated by a study of network news coverage of theIranian hostage crisis, which involved 52 Americans who were held for444 days (November 4, 1979—January 24, 1981) (cf. Altheide, 1981,1982, 1985a). This was the first study of its kind since previously there hadnot been an extended crisis that was so heavily televised. My task was todescribe the news coverage in a theoretically informed manner whichwould provide data for further conceptual refinement. Theoretical andsaturation sampling were combined. Ultimately, I viewed 925 news reportsabout this highly publicized series of events.The major focus of the study was to examine the role of formats in TVnews coverage of an international crisis. Formats are organizationaldevices to facilitate coordination of the news process. Format refers to therules and procedures for defining, recognizing, selecting, organizing, andpresenting information as news. They are a link to, and a probe of theexternal environment (cf. Altheide, 1985b). TV news formats include shortreports with visual and aural information, presented in a narrative formwith a beginning, middle, and end. Particular attention was paid to thenature, extent and source of visual imagery, and how it was used forthematic emphasis (cf. Epstein, 1973; Altheide, 1976; Tuchman, 1978;Bennett, 1983).While it is possible to adapt almost any event to the narrative format,events with certain characteristics are more likely to be selected forcoverage because they can readily be shaped into such a format (cf.Altheide and Snow, 1979). These characteristics include: accessibility,visual quality, drama and action, perceived audience relevance, and encapsulation and thematic unity. The general dimensions of each of thesecharacteristics can be briefly stated:Accessibility refers to how easily newsworkers can learn about anevent, obtain information about it, get to a site where it occurs, and/orobtain visuals.

70QUALITATIVE SOCIOLOGYVisual quality is the extent and clarity of film, tape or other visualdepictions of the significant action.Drama and action refer to the graphic, visual and aural portrayal ofsome movement which is sued to illustrate the event.Encapsulation and thematic unity refer to the ease with which an eventcan be (1) briefly stated and summarized, and (2) joined to a similar eventor a series of reports over a period of time, or even within the samenewscast.Finally, audience relevance is the interest newsworkers perceive anitem to have for a mass audience. Of course, the extent of the audience’sinterest is also a feature of the way the other format criteria are brought tobear on an event.Together, these format considerations direct as well as reflect newsmessages. However, the selection of a sample and the data collectionprocedures must be theoretically informed by this perspective.Sampling and Data CollectionPrior to systematically selecting newscasts which would represent adequately the more than 14 months of daily coverage, all available newsreports pertaining to Iran on the network evening newscasts during the firstnine days of the embassy takeover (November 4—12, 1979) were viewedand analyzed. This preliminary analysis made it clear that a simple randomsample or stratified sample would systematically distort an understandingof news coverage of this event. This is because coverage was in groups ofdays, such as concentrations around holidays and the tendency to broadcasta certain feature of the ordeal over a period of several days.A few brief examples illustrate what was gained from viewing newscontent reflexively rather than statically. If the QCA approach to samplingand data collection had been followed, important thematic pat-terms wouldhave been lost. Major reports about various facets of the hostage situationoccurred in mini-series, often over a period of several days. Thesignificance and message of one report would be lost when removed fromthe context of other reports. For example, while the families of hostageswere featured in about 12 percent of the total sample of 925 news reports,they were involved in 37 percent of all hostage related reports during thehostages’ first Christmas, and 25 percent of reports about the secondChristmas. The upshot is that the hostages’ families played more of a roleat certain times, and this role had a great deal to do with the format of TVnews.

Reflections: Ethnographic Content Analysis71The final sample was a saturation sample of approximately 112 daysand 26 hours of compiled news reports. Since a simple random samplewould rarely select two or three days in succession, it would obscure thefact that networks frequently stretch a series of reports over several days.The solution was to combine two units of analysis: (1) each network reportpertaining to Iran (which may have included more than one topic)presented in a newscast; and (2) several consecutive newscasts or“clusters.” Thus, the sample consisted of seventeen “clusters” consisting of5—9 consecutive newscasts per network. In addition, care was taken toproportionately represent all 14 months, as well as weeks 1—4 of thevarious months. I also checked news records from the VanderbiltUniversity Television News Archive to determine if the networks weresimilar in the amount and emphasis of their coverage (Altheide, 1982).This check indicated that they were.The original protocol was constructed to provide both numeric and narrative (descriptive) data on the following topics: network; presenter; lengthof report; origin of report; news sources; names and status of individualspresented or interviewed; their dress, appearance, and facility withEnglish; what was filmed; and the correspondence between film, speechand overall emphasis. The narrative portion was particularly helpful fordeveloping a framework for dealing with visuals.With the exception of identifying materials, (e.g., network, date, time),the news broadcasts were viewed without predefined and rigid categoriesfor defining what was relevant. At the same time, prior research andfamiliarity with news procedures provided me with tools to use inobservations and analysis. My general procedure was to view a fewreports, assess the message(s) in terms of news techniques, and then notegeneral categories for this report, and for reexamining several previousreports. Then I checked the quality and quantity of information recorded interms of what was being omitted and what segments or time blocks werenot important for the present focus. This continuously refined explorationand comparison became a substantively informed sampling procedure andtopical guide to data collection. The topics that emerged in this way wereas follows:Hostages: Any report which focused on the hostages’ status, location,health, etc.Families: Any report which focused on the hostage families’ status,health, reaction, plans, etc.Shah of Iran: Any report which focused on the context of the Shah’srule, including political alliances and enemies, as well as his status, health,location, and statements.

72QUALITATIVE SOCIOLOGYIran: Any report which focused on Iranian government action, plans,reactions, statements, or elections.Iran (internal problems): Any report which pertained to economic,civil, criminal, and demonstration problems, underlying internal revolts.Iran (external problems): Any report which concerned economicsanctions or military threats such as the Soviet invasion of Afghanistan,and Iran’s war with Iraq. U.S.A.: Any report which concerned U.S. government actions, statements, reactions, proposals, or criticisms.International: Any report which concerned international statements,actions, reactions, proposals, involvements, or sanctions, including theactions by the United Nations and the World Court.Iranian Students: Reports about Iranian students in the UnitedStates, reactions to U.S. policy, support of the Iranian government,demonstrations, civil and criminal actions, etc.The progression from data collection to interpretation was intended tobe reflexive rather than serial. While there was an effort to verify thefindings of some prior research which suggested that news organizationsemploy stereotypical angles to encapsulate an event (cf. Epstein, 1973;Batscha, 1975), that focus alone would have precluded important emergentunderstandings about the interaction between formats. One product of anemergent orientation to the data was a set of insights about the source ofreports. Previous research revealed that newsworkers often incorporate“file” film and “old” reports into new ones (cf. Altheide, 1976) so that anyupdate or overview of a report is usually tied to what has gone before. Inthis sense, TV news often reports on itself. The news sources and visualopportunities fluctuated during the hostage ordeal so that new visuals ofhostages and their captors became quite scarce. The networks’ problemthen was to look elsewhere, but where they looked also led to differenttopics, e.g., family members, emphases, and conclusions. An atheoreticalsample would have missed these systematic clusters which show veryclearly that the substance of reports could be predicted by the origin ofreports (cf. Altheide, 1985b:77ff).The fact that the origin of reports and visuals contributed to thethematic emphasis did not emerge until well into the study. ECA offers anapproach for systematically studying the use of visuals and text as featuresof formats (cf. Lang and Lang, 1968; Adams and Schreibman, 1978;Tuchman, 1978). My procedure was to describe the visuals in terms of“what was shown,” “who was shown,” and “what they were doing.”Rereading my descriptions and relating these to certain events (e.g., thehostages’ Christmas) led me to return to previous tapes and add additionaldata to the open-ended protocols. In turn, since I began to under-

Reflections: Ethnographic Content Analysis73stand that the “file tape” was used more often with reports originating fromcertain locations, I saw the pattern between these sources and certaintopics and aspects of the hostage situation. For example, visuals— usuallyfile film—of crowds chanting anti-American slogans were routinely usedwhen a reporter in London would conclude a summary of the day’snondevelopments.Moving reflexively between data collection, analysis and reconceptualization increased my understanding of the relevance of TV news formats,sources, and thematic emphasis. While hostage families had been a part ofthe story from the beginning, they became relatively more prominent overtime. A key factor in the visual focus on families was their portrayal ofgrief, anguish, frustration and anger. Hostage families were often willingto be interviewed on camera, and several invited reporters into their homesfor days at a time. Because they were also available for coverage from thenetwork affiliate stations throughout the United States, they received adisproportionate amount of the coverage originating from U.S.A./Other(not New York or Washington, D.C.). Moreover, the hostage families alsoreceived a good deal of coverage from Washington/New York becausemany of the family members formed a quasi-organization, centered inWashington, D.C. The articulate spokespersons, who made publicspeeches, were adept at reaching the centers of government and the newsmedia.This availability, in combination with visual news format, contributedto the hostages’ families becoming the visual signature for emotion, fear,and administration inaction. Furthermore, hostages’ families received morecoverage than other facets of the event (e.g., Iranian government) whichclearly were necessary for a broader understanding. Reports about thehostages would be joined with a report about the hostages’ families,especially certain family members who became quite familiar to TV audiences. Because the hostages’ families were seen routinely on the evefling newscasts and ABC’s Nightline, they became the visual link to thehostage ordeal.If the hostages’ families were symbolic of the American perspective,visuals of Iranian crowds chanting anti-American slogans emerged as thesymbol of the adversaries. For example, there were 87 film reports (9.4percent of total reports) about Iranians, primarily in street demonstrationsand crowd scenes. Iranians in the United States were similarly presented:31 of 64 reports featured them in crowd activities, including broadcasts ofsome of the most brutal confrontations since the Civil Rights movement.The way this film defined events can be illustrated with a 2:40 report byCBS reporter Liz Trotta in an interview with an Iranian official who wascomplaining about problems in media

74QUALITATIVE SOCIOLOGYpresentations of cultural differences. While one of the aims of the reportwas to depict other aspects of life in Iran, only about :20 of “non-hostage”daily life was actually presented.The visual emphasis further compounded an already strange encounterwith adversaries embedded in a different culture, with a different worldview and a different religion. Media personnel were not unaware of thistendency to heighten dissimilarities. The push to deliver exciting film andto maintain the “story” on the embassy contributed to a distorted view ofIran that precluded understanding of the context in which the hostagesituation emerged and persisted. On January 16, 1980, the followingexchange took place between an anchorperson and a reporter in Iran.Anchor: What is happening there? Do Americans really understand that is happeningthere?Reporter: I think . . . that the impression we convey from the scenes in front of theembassy, all the fist shakers yelling “Death to Carter, Death to America,” weconveyed a picture of a nation in the grip of madness, and yet just a few blocksaway from the embassy gates people are going about their lives in a normal fashion.Mothers are taking their babies to the park. Businesses are opened. Tehran is prettymuch working as normal. (1/16/80; NBC Special Report: “Crisis in Iran: 1 YearAfter the Shah, Day 75”).The excerpt suggests that reflexive appraisal of documents such as TVnewscasts are done by newscasters themselves. The aim of ECA is to placedocuments in context just as members do, in order to theoretically relateproducts to their organizational production.ConclusionA rationale for an ethnographic and reflexive approach to documents issimilar to the rationale of ethnographic research in general. Samplingprocedures are informed by theory while constant comparison and discovery are used to delineate specific categories as well as narrativedescription. Situations, settings, styles, images, meanings and nuances arekey topics in the analysis of news documents. I have found that structureddata collection based on a protocol combined with ethnographic field notessupports a theoretically informed account of media content.Structured protocols used alone hide critical questions and issueswhich may become apparent only later. If ethnographic materials areincluded, it is usually possible to return to the data when other questions

ERRATATable 2 of David Altheide’s article in this issue was inadvertentlyomitted at the time of printing. This table, show below, shouldhave appeared on page 75.TABLE 2Time:15:18:18:12:22StatementThere are about 150 foreign newsmenin Iran now. They come from all over theworld to cover this major story, and arepermitted to work after they are accredited by the ministry of national guidance.For the moat part, journalists here havebeen accorded the freedom of movementthey enjoy in Western countries. But nowthat freedom is threatened by an announcement from the government thatWestern journalists may be expelled fromIran in the near future.Our revolution has created a vast culturalgap. From what West understands lifeshould be, and we’re trying to close thisgap and this is where the media can helpa great deal.Indeed, there is a culture gap. The verynature and ritual of Islam is an intellectual confrontation for the West.And the new religious militancy inspired by the revolution has only heightened the challenge.As with most stories, there are negativeaspects in the telling of the Iran story:bloody challenges to the Khomeinigovernment inProvence; the mistrust and defianceof the new regime in . Province;a serious threat to Ayatollah Khomeini.From a official point of view, theWestern press focuses too closely onthese apects.SpeakerLizTrottaFilmJournalists andcamerasGhotbzadeh andjournalistswalkingSadeghSadegh sittingbehind deskTrottafilm of crowds,self-flaggelation,praying in massTrottacrowd, chantingand wavingknives, weaponspeople runningman holding a gunclose-up of manpeering throughwrap aroundface; man on roofwith a gun; crowdWhen I have an American reporter fromSadeghone of the major television stations comingbehind his deskfrom Rome, and he says, my God, I thoughtthe whole country was falling apart.I was even afraid of how lam going to get:26from the airport to the hotel, and then Icame and I saw everything cairn and quiet.I was surprised. That reflects the kind ofnegative reporting, or reporting out of con.text . . . In the midst of these politicalTrottavendor; autos;and religious clashes, there is a normalcywoman makingin the city of Tehran. The legendary trafficpurchase fromjams are still a game of nerve and skill;street vendor;the businessman and housewives mainbeets cooking;thin the daily life; street vendors cookpeople walking;:20their hot red beets in the markets.crowd; burningBut against this background one must weighflag; chantingthe headlines: Afghans attack the Sovietcrowdembassy in Tehran; Armenians attackchains on embassythe Turkish embassy in Tehran; Afghansgate; cameraattack the Afghanistan embassy in Tehran;pointed at:21And above all, Americans still captive in theembassyUS embassy. Liz Trotta, CBS News, Tehran.

76QUALITATIVE SOCIOLOGYand inquiries arise. For this reason, I have been able to use “old” data setsto answer new questions.My recent attempts to integrate findings from studies of the mass media(especially TV news) with the nature of media in general in social life. Ihave examined non-mass media institutions, settings and practices todevelop a conceptual about the effect of mass media on a wide range ofactivities. These concepts clarify the influence of temporal andspatial features of what appear,, at first glance, to be nonmediated occasions. This approach yielded “format” as a common concept in studies ofmass media and other types of mediation.I have looked for instances of mediation in situations which may not beassociated conventionally with media principles and theory. In recentyears this has led me into settings and issues involving social definitionsand applications of “justice.” The settings I have examined include TVcoverage of courtroom activity, the use of “keyboards” and otherterminals by police officers and other criminal justice agents(Altheide, 1985c), and TV viewers’ requests for assistance from anaction line “troubleshooter.” In brief, ethnography offers a perspectivefor analysis of human action in the field and in documents; the key is toreconceptualize the latter as the former and vice versa.ReferencesAdams, W. C. and Fay Schreibman1978 Television Network News: Issues in Content Analysis. Washington, D.C.: GeorgeWashington University.Altheide, David L.1976 “Creating Reality: How TV News Distorts Events.” Beverly Hills, Calif.: Sage.1982 “Three-in-one news: Network coverage of Iran.” Journalism Quarterly 48:476—90.1985a “Format and ideology in TV news coverage of Iran.” Journalism Quarterly Summer 62:346—51.1985b Media Power. Beverly Hills, Calif.: Sage.1985c “Keyboarding as a social form.” Computers and the Social Sciences 1985:(Fall).Altheide, David L. and Robert P. Snow1979 Media Logic. Beverly Hills, Calif.: Sage.Batscha, Robert M.1975 Foreign Affairs News and the Broadcast Journalist. New York: Praeger.Bennett, W. Lance1983 News: The Politics of Illusion. New York: Longman.Berelson, Bernard1966 “Content analysis in communication research.” Pp. 260—66 in Bernard Berelsonand Morris Janowitz (eds.), Reader in Public Opinion and Communication. New York:The Free Press.Epstein, Edward J.1973 News From Nowhere. New York: Random House.

Reflections: Ethnographic Content Analysis77Glaser, Barney and Anselm Strauss1967 The Discovery of Grounded Theory. Chicago: Aldine. Hammersley, Martyn andPaul Atkinson1983 Ethnography: Principles in Practice. London: Tavistock. Holsti, Ole R., et al.1963 Content Analysis. Evanston, Ill.: Northwestern University Press. Johnson, John M.1975 Doing Field Research. New York: The Free Press.Krippendorf, Klaus1978 “The Expression of Values in Political Documents.” Journalism Quarterly:510—18. Lang, Kurt and Gladys Engel Lang1968 Politics and Television. Chicago: Quadrangle.McCormack, Thelma (ed.)1982 Studies in Communications: Vol.2 Culture, Code and Content Analysis. JAI Press.Plummer, Ken1983 Documents of life: An Introduction to the Problems and Literature of aHumanistic Method. London: George Allen & Unwin.Schwartz, Howard and Jerry Jacobs1979 Qualitative Sociology. New York: The Free Press.Starosta, William J.1984 “Qualitative Content Analysis: A Burkean Perspective.” Pp. 185—94 in WilliamB.Gudykunst and Young Yun Kim (eds.), Methods for Intercultural Communication.Beverly Hills, Calif.: Sage.Tuchman, Gaye1978 Making the News. New York: The Free Press.

Ethnographic Content Analysis (ECA) Ethnographic content analysis is used to document and understand the communication of meaning, as well as to verify theoretical relation-ships. Its distinctive characteristic is the reflexive and highly interac- tive nature of the investigator, concepts,