Feminist Content Analysis and Representative Characters
Patricia Leavy

The Qualitative Report, Volume 5, Numbers 1 & 2, May, 2000


This paper details textual, visual and audio-visual content analysis from both a general and feminist perspective. It provides a backstage look at how I took an abstract idea of analyzing the larger socio-cultural-political American 1990s/2000 context through a representative fictional character, Ally McBeal, and created a manageable project outline for how to achieve my research goals using content analysis in multiple ways.


When it came time to choose a topic for my next research project, I found myself with an idea for the substantive content of the paper but unsure of how to methodologically fulfill my goals. As my work is generally qualitative and this particular project lent itself to content analysis, I decided to construct a proposal of how I was going to carry out my project working from within a feminist content analysis framework. I researched the method of content analysis while specifically focusing on how I would use it for my project and 1) detailed some of the complexities and individual research options within content analysis, and, 2) demonstrated how to construct a reasonable research strategy by working through numerous methodological issues prior to active research.

What follows is the background/backstage work I did prior to carrying out my project. The goal of this paper is to provide both general information about textual, visual and audiovisual content analysis research and also give researchers a road map for how they might take a large seemingly abstract idea for a project and detail a feasible outline for its completion.

The Goal of the Project

I will analyze Ally McBeal as a representative character through which I will discuss the larger social-cultural-political American context in which the show and character have emerged. Issues to be explored through this selected representative character are: feminism in the 1990s, body image issues including the American female body "ideal", comparisons between various forms of media in the 1990s, the blurring of reality and fiction within a specific historical and spatial context, and the (re)production and consumption of images within postmodernity. When addressing each of these issues the "imprinting" of history, following Foucault's tradition, will be analyzed as a part of the "theoretical" backdrop. While the theoretical framework for the paper will be primarily feminist post-structuralist, the methodological design will consist of a complex and systematic feminist content analysis.

While this analysis will address the methodological framework for this upcoming project, I will begin by briefly addressing the notion of "representative character" that my research is based upon. Once this issue has been crystallized, the use of content analysis as my selected method of data collection and analysis can be addressed as it pertains to the specific intent of my project.

My conception of "representative character" is based largely upon the work of S. Paige Baty (1995) (who analyzed Marilyn Monroe as a representative character in her compelling book, American Monroe: The Making of a Body Politic. Drawing upon Donna Haraway's theory of the hybrid subject and Roland Barthes' analysis of "Myth Today", Baty states:

"a representative character through whom to approach the political cultural condition of our time. Remembered as product or story or some hybrid of the two, the representative character operates as a site on which American political culture is written and exchanged. Consequently, the representative character is not simply available as shared story but also for sale as product. Particularly striking is the excessive commodification that accompanies, and is made possible through, iconographic rememberings of representative characters. Once the representative character is imaged as icon, she is easily reproduced on any number of objects... The mass-mediated representative character operates as a figure through whom multiple meanings, references, and roles are remembered..." (Baty, 1995, pp.10-11)

While Baty's (1995) analysis/definition of representative characters contains further description then what I have presented, I offer only this brief definition as I think it is sufficient for the purpose of this paper. In order to utilize a representative character as a vehicle, or focal point from which to begin, to discuss the larger social-cultural-political landscape of a specified time and place, the selected character must be analyzed in the multiple forms in which it is presented/represented.

In order to understand how, and, the scope to which Ally McBeal can serve as a representative character I will analyze the television show, tabloids, magazines, newspapers and an internet discussion group. In addition to having a multitude of data sources to analyze, I will also investigate the written, visual and audiovisual components as appropriate to each data source. As I will be employing a variety of content analysis techniques, I will now turn to a general discussion of content analysis and the ways in which I intend on using this method.

An Overview of Content Analysis

To begin, what exactly is content analysis? A standard textbook definition is as follows:

Content analysis is a technique for examining information, or content, in written or symbolic material... In content analysis, a researcher first identifies a body of material to analyze... and then creates a system for recording specific aspects of it. The system might include counting how often certain words or themes occur. Finally, the researcher records what was found in the material. He or she often measures information in the content as numbers... Content analysis is used for exploratory and explanatory research but is most often used in descriptive research. (Neuman, 1997, p.31)

In short, content analysis is the systematic study of various forms of communication and usually aims at description rather then explanation. While this standard definition of content analysis is a sufficient starting place, as stated earlier, I will be conducting a feminist content analysis using both a multitude of data sources and analysis techniques. This requires a much more complex look at the premise of content analysis including its particular advantages. Moreover, what will make my content analysis "feminist" in nature?

While I have read in standard textbooks that content analysis has existed as a method of social investigation for the last decade, in the 1830s Harriet Martineau already understood and utilized the power of studying "things" (Reinharz, 1992, p.145). Firstly, why study "things" as I plan on doing? Furthermore, how am I going to connect the "things" I choose to analyze to the larger social-cultural-political context?

The methodological choice of content analysis as the appropriate method in light of my research goals is evidenced, as cultural artifacts embody, reflect and mediate the views of the society from which they emerge. In other words, the themes exist within the artifacts independent of and prior to the research process and through a careful process of selection and analysis I will extract those already present themes. Put differently,

The cultural products of any given society at any given time reverberate with the themes of that society and that era. (Weitz as quoted by Reinharz, 1992, p.145)

As stated by perhaps the pioneer of content analysis:

To arrive at the facts of the condition of a people through the discourse of individuals, is a hopeless enterprise. The plain truth is - it is beginning at the wrong end. The grand secret of wise inquiry... is to begin with the study of THINGS... Though the facts sought by travelers relate to Persons, they may most readily be learned from Things. The eloquence of Institutions and Records, in which the action of the nation is embodied and perpetuated, is more comprehensive and more faithful than that of any variety of individual voices. (Martineau, 1995, p.73)

While the "facts" I wish to study through the cultural artifacts surrounding Ally McBeal relate to "Persons" in Martineau's (1995) language, they can be best studied through cultural products which embody "the action[s] of the nation" I wish to link my representative character to. For example, the ways in which various forms of media ranging from tabloids to entertainment magazines to print news addressed-created-perpetuated the rumors that Calista Flockhart is anorexic says something about the larger social-cultural context "we" live in. The way that the media has labeled Ally McBeal (not Calista Flockhart) the "poster-girl for postmodern feminism" also says something about the larger social-political context in a given historical time and space. While I will not be discussing the content of my project at this point I offer these vague examples merely to illustrate some of the themes of society embedded within the cultural products surrounding Ally McBeal.

Not to be missed in Martineau's (1995) statement, which I should note is a dominant theme throughout her book To Observe Morals and Manners, is the contention that studying "things" has the advantage of non-interaction. According to Reinharz cultural artifacts possess two distinctive properties, 1) a naturalistic quality, and, 2) they are noninteractive (Reinharz, 1992, p.147). This is important for a variety of reasons. Firstly, the researcher has not taken part in the construction of the products being analyzed. In this way the data itself has an implicit dimension of validity. Secondly, the themes extracted from the data represent the discourse of those who created the products (including the social context in which that discourse lives) versus the personal/political discourse of the individual researcher. Due to the nature of cultural products feminist researchers seek to examine not only the texts/artifacts, but also the processes of production (Reinharz, 1992, p.145). Lindsay Prior articulates the preceding points made by Martineau, Weitz and Reinharz in the following way:

...in most literate cultures the order of things is often displayed and represented in text as well as in artefacts...we should treat each text as coherent simply because it is there, lying before us as a unified object. The task of the researcher is therefore to investigate 'archaeologically', as Foucault might say, the innumerable accidents and myriad twists and turns of human practice that have brought the text to its present form. Qualitative research in this context, then, is not so much a question of deciding what a given text or textual extract might mean to a thinking subject as a matter of analysing the origins, nature and structure of the discursive themes by means of which the text has been produced... One might even argue that in many spheres of human practice one can only know the world through the representational orders contained within text. (Prior, 1997, pp.66-67)

Before moving on to the problematic(s) in content analysis I will briefly summarize the advantages already discussed. Analyzing cultural products is particularly useful in that the materials are noninteractive and thereby contain a naturalistic quality that function to implicitly gain a certain level of validity, and also, as Prior (1997) decidedly notes, artifacts and texts embody the discourse of their producers (larger context) as opposed to the individual researcher. Given that the data used in content analysis lends itself to a particular kind of validity, how can researchers maintain that validity? As a feminist researcher I especially feel it is important to address this issue of "contamination" particularly when working with data that has an essence of validity to begin with (as in content analysis). Self identified feminist research is likely to receive a great deal of scrutiny and thus it is imperative that feminist researchers working with content analysis make every effort to preserve the credibility of their data.

While this discussion has been leading up to a notation as to the importance of the context of justification (methods) within a feminist content analysis research design, as a feminist researcher I would like to first address the context of discovery within feminist research generally, and the Ally McBeal project specifically. During the initial phase of research, the context of discovery, Sandra Harding asserts and I concur that feminist researchers have an obligation to disclose a brief personal 'biography' including why they have chosen to research a given topic, the vantage point from which they will begin inquiry and the way in which they will gather, analyze and report the knowledge they have produced. So while I have just been addressing the natural element of validity within the data used in content analysis, as a self identified feminist I understand that from the conception of this Ally McBeal project, to the selection of a method and then the selection of data, interpretation and report writing, my own positionality impacts this project. By acknowledging my own position within this research project I however will avoid the tendency of claiming universal truths but rather place my project within a specified context. Having mentioned my understanding that my own discourse inevitably will make its way into this project, which I will actively and openly utilize as opposed to futile attempts at repression/denial, I will now move on to discuss the context of justification.

Not to be lost after my review of the context of discovery is the conception that cultural products do in fact embody the discourse of the larger society from which they emerge. By investigating both the physical artifacts/texts and the processes of production (Reinharz, 1992, p.145) feminists can connect the products to the larger themes within a given time and place. While studying "things" inherently contains an aspect of validity, in order to avoid undue contamination, the context of justification is crucial in a feminist content analysis.

While to a certain extent the population from which I will extract my data samples (in the instance of the Ally McBeal project the populations are the television show, tabloids, magazines, newspapers and an internet discussion group) is determined during the context of discovery (who and what will be asked), the specific methods of sampling to be utilized are a crucial phase during the context of justification. Once the populations have been selected I must determine how I will sample from within those populations, and furthermore, to what extent will my sampling procedures allow me to make generalizations when reporting my conclusions. In other words, what larger populations can I later make partial truth claims about? The key is to use a systematic sampling procedure in order to maintain the integrity of the data. In other words, I can not merely collect the photographs (as an example) of Ally-Calista, that support my own beliefs, but rather randomly sample from within my population in a pre-determined/disclosed format. I must address negative cases. For example, I could sample the content of the photographs from six specified magazines in a given time period. In terms of the internet discussion group I will be analyzing, after choosing a specific discussion group (my population) I pre-selected an eight-week period and my sample will consist of all responses within that eight-week time frame. This sampling procedure will force me to look at a fair representation of responses.

Authenticity within Feminist Content Analysis

The goal of content analysis is to extract independently existing themes from cultural products. As mentioned the data already has 1) a naturalistic quality, and 2) is noninteractive. Furthermore, the data embodies the discourse of its producers/production process, which is not the individual researcher. Therefor, the raw data used in content analysis has an implicit measure of validity which a researcher must systematically attempt to maintain. While specifics are discussed throughout this paper, at this point I will provide a general overview of how I will maintain authenticity by general feminist standards throughout the research process.

The issue of "contamination" is central to the development of my methodological strategy as one of my tendencies towards content analysis is the implicit dimension of validity within the data. Additionally, as the project has a political component (feminism), maintaining authenticity is not only an issue of methodological quality control, but also a political responsibility.

Systematized sampling procedures are a necessary step in terms of maintaining validity and achieving generalizability. For this project I will conduct random sampling ensuring "negative cases" have an equal probability of being selected as those cases which may support my personal hypotheses. I will also select sample sizes that are, on face value, proportionally large enough, relative to the number of available cases within each universe, to constitute representative samples. These sampling strategies will allow me to generalize from my samples to the universes from which cases were drawn while also aiding in the maintenance of interpretive validity.

After sampling, several methodological techniques will be employed during analysis as quality control will be a critical issue throughout the interpretive process. Using a feminist deconstructionist approach I will be actively deconstructing the data in order to reduce it to its own essence thereby revealing dominant themes already contained within the data. This study relies on multiple methodological techniques, as opposed to a consistent methodological frame, so that I may explore more analytic possibilities.

Sections of data will be interpreted from a "grounded theory" approach where preconceived code categories are not employed but emergent theories are grounded in the raw data. This complies with the goals of feminist content analysis as the distinctive properties of the cultural products are preserved. Again, my goal is to maintain the authenticity inherent in my raw data throughout the interpretive process which will eventuate in a high degree of validity.

While segments of data will be analyzed from an inductive approach as discussed, a deductive frame will also be used. During this phase of analysis preconceived code categories will be employed and the resulting data will be largely quantitative. In an effort to maintain quality by achieving "face validity", the code categories selected will also be reviewed by two colleagues. The three of us will individually examine 1) the appropriateness of the language, 2) the exhaustiveness of the code categories, and 3) the exclusiveness of the code categories.

By focusing on parts of data quantitatively, emergent patterns pertaining to the larger universe can then be examined qualitatively. I will take creativity during analysis, viewing interpretation as an art form and relying on my instincts/impressions, while allowing quantitatively determined patterns to act as my guide. By combining strict methodology with creative authorship I will be remaining both true to the data as well as my own politically influenced interpretations. This is crucial in order to retain external validity while also maintaining authenticity from a feminist perspective.

Finally, in the write-up I will include a detailed discussion of the context of discovery and the context of justification, including a personal biography as it pertains to my role in this project. By actively positioning myself within the production process, I will be attentive to issues of feminist quality control. As I will provide a detailed methodological section in my report, ultimately readers will be able to draw their own conclusions as to my success in terms of authenticity. Full disclosure, as often encouraged in feminist scholarship, will afford readers that opportunity.

Problematic Comparative Studies

In order to illustrate the importance of systematic randomized sampling procedures and the inclusion of any negative cases that may appear, I will discuss as few studies I discovered that failed to disclose their sampling procedures and furthermore neglected to exemplify the use of systematized procedures for sampling and how this, to me, creates problems of authenticity and generalizability. The American Studies Journal, Volume 39, Number 2, Summer 1998 issue contained a section encompassing a third of the issue titled "The Personal is Professional on TV". In this section four television shows were analyzed independently by four individual researchers. Prior to the essays, the four researchers collaborated on an introduction which aims at discussing the overall themes/conclusions in the following articles. The authors make the following statements:

Our essays attempt to understand how "the personal" functions within specific television programs... the TV programs we examine here are representative of a time in which the "professional" has come to appear as an alternative to the "political"... What the four programs share is a sense that anyone can undergo professionalization, regardless of the work they do... Our close-readings seek to show how thoroughly stories about the professionalization of everyday life have saturated contemporary popular culture... whatever their strengths and weaknesses, he shows we examine in these essays provoke us to consider the consequences of this trend. They call us to reflect on a society in which people have a hard time defining themselves without a job description. (Bertsch, Epstein, Newitz, & Sandell, 1998, pp.90-91)

While the authors both collectively and individually make generalizations about the "content" of these programs as well as large scale conclusions regarding how these themes merely re-present themes within the culture from which they have emerged, I am not convinced that their argument is substantiated as a result of their haphazard methodological origins.

Gillian Epstein's (1998) analysis of the television show Beverly Hills 90210 presents itself as if it were a content analysis. Epstein frequently refers to the "themes" and "narrative" of the show. Epstein continues on to draw large scale conclusions about both the show and the historical and spatial context in which the show is produced and consumed. What I then find astounding, is that the entire essay is based upon Epstein's viewing of one episode of the television show. This essay, at best, is the product of an individual case study, yet Epstein professes numerous universals based upon her data. Actually, the episode the essay is based upon is an anomaly of sorts both in Epstein's terms and in my own, which is why she chose to analyze that particular episode. Epstein failed to account for the context of discovery by intentionally selecting only one "special" episode without sufficient justification, failing to look for negative cases and then constructing universals regarding both the entire television series and the larger social context. Furthermore, Epstein does not provide a methodological discussion of her coding procedures. As a reader, I was left wondering how she extracted these themes from the program.

While a lengthy discussion does not seem paramount, I will briefly note that while the three other essays based upon The X Files (Bertsch, 1998), ER (Newitz, 1998), and Friends (Sandell, 1998) discussed several episodes, none of the articles disclosed precisely how many episodes, how those episodes were chosen and what the process of interpretation consisted of (i.e., the coding procedures). I chose these examples of problematic content analysis work as they too, seemingly, aimed at using a television show as the vantage point from which to discuss larger social-cultural-political conditions. One positive note to mention is that these essays do provide a precedent for the kind of work I have proposed.

Interpreting Textual Data

As I have tried to demonstrate the importance of systematic sampling procedures as a measure of preserving the data's validity and in allowing for specified levels of generalizability, I will move on to discuss the analysis stage. Once the units of analysis and the units of observation (Babbie, 1998, p.310) have been determined, how is that data coded and analyzed/interpreted? As the visual and audiovisual data to be used in this Ally McBeal project require specific interpretation techniques, for now, I will discuss the analysis phase in regards to the written texts I will be sampling.

I will begin by discussing how I will analyze the texts obtained from the discussion group I monitored for eight weeks. The discussion group formed on the internet immediately after the media announced rumors that Calista Flockhart was/is anorexic. Respondents are obviously self-selected and I do not have any way of gaining biographical data about individual respondents. Similarly, while respondents log into the discussion group using a name, I can not stratify my sample by gender in that I have no way of being certain whether the names given are the actual names of the respondents and for that matter do not feel comfortable assuming any given name to be only appropriate to one gender (even if I could "trust" that the names given were real). The eight weeks of data (responses) that I am analyzing begin with the inception of the discussion group.

First, I will enter the text into HyperRESEARCH, a computer directed qualitative data analysis program. In terms of entering the text into the program, which HyperRESEARCH does on a case by case basis, each week's worth of responses will be entered in as a case resulting in eight cases differentiated by the dates (seven days) they encompass. It is important to mention this as the sequence of responses may (or may not) prove to be a critical factor in the content of the responses. During the interpretive phase this will also allow for a temporally constructed comparative analysis component. The data for each response is the "source material" based on the HyperRESEARCH model (Hesse-Biber, Dupuis, & Kinder, 1991, p.290).

Once the source material (text) for all eight cases is put into HyperRESEARCH, I will code the data using a grounded theory approach. By employing a line by line coding procedure the code categories that I develop will directly reflect the words of my respondents. Furthermore, the larger themes I eventually extract from the data will be a direct result of my respondents' words as opposed to a theory I developed prior to analysis, as would result if I chose to use a deductive model as opposed to the inductive approach I will use. As I will be utilizing their model for the very reasons that the following excerpt discusses I offer the following which is said by Hesse-Biber, Dupuis and Kinder (1991) drawing upon the work of Charmaz's "Grounded Theory":

A central part of qualitative analysis involves extracting meaning from collected textual materials. Sometimes, a deductive approach is employed to code the data using theoretically pre-assigned code categories. A more intensive, inductive procedure requires an immersion of oneself in the text until themes, concepts of dimensions of concepts emerge. The researcher uses a "focused" coding procedure which allows for the building and clarifying of such concepts. (Hesse-Biber, Dupuis, & Kinder, 1991, p.292)

By systematically detailing my sampling procedure and then utilizing an inductive approach to analysis, the data which is already both non-interactive and "naturalistic", will maintain a high level of validity the importance of which has already been addressed.

Although I will not be using a computer directed analysis program, once I have obtained a sample of the other written texts I will examine (including newspapers, magazines and tabloids), I will code the material using both a deductive and inductive approach in search of both qualitative and quantitative data. Many feminist researchers who work with content analysis as their method use a combined form of analysis as I intend to do. For example, Francesca Cancian and Steven Gordon systematically selected articles they were going to analyze in regards to "emotion norms" and then "analyzed the material qualitatively and quantitatively" (Reinharz, 1992, p.151) using a multi-method approach to analysis. In this case, they found that the quantitative data supported the qualitative analysis. I will be using quantitative analysis, such as word counts, "to identify patterns" (Reinharz, 1992, p.155) and qualitative data to place those patterns within an interpretive context discussing large scale themes.

While this preceding discussion is meant to serve as a summary of my intended overall approach to the remaining textual information I will analyze, I will now move towards a more specific discussion of the interpretation techniques I will be employing. Firstly, once the internet discussion group data has been analyzed I will be left with magazines, newspapers and tabloid. Important to note is that this discussion is only in reference to the textual data I will obtain from a sample of each of these sources. The accompanying photographic data will be analyzed using specific techniques developed for analyzing visual/audiovisual data. In my final project I will discuss the visual data in relation to the textual data.

After systematically selecting my samples I will break the texts down into categories (i.e., women's magazines, entertainment magazines, etc.). Each sub-sample of texts will be analyzed and then all of the samples will be compared/contrasted. In terms of utilizing a deductive approach to analysis, from a feminist and feminist post-structuralist vantage point, I will create code categories prior to analysis and look for evidence of the presence of data fitting into my pre-constructed categories. The categories will include (but are not limited to) references to Ally (the character) having an eating disorder (versus Calista), labeling Ally a feminist, positive references to Calista's physical appearance, negative references to Calista's appearance, etc. I have provided this brief list of code categories merely as examples. I will both count and record the instances in which data fits into these code categories. Later I will discuss these "parts" (specifics) of the texts in regards to their actual prevalence and also in relation to the overall themes, patterns and messages that will be addressed in the qualitative section of my analysis. As Reinharz says, "feminist analyses can treat a document as whole or analyze its parts" (1992, p.159). I am going to combine these strategies by beginning with an analysis of the "parts" from a deductive, and in part, quantitative approach and then move towards studying the whole text/texts qualitatively.

In terms of qualitatively interpreting the texts as complete documents, I will utilize a "feminist intertextual deconstructionist" approach. I should note that one of the reasons I have selected this interpretive framework is that it corresponds with the theoretical framework the final analysis will be placed within (and I should disclose that it suits my own interests/perspective which will be noted in the paper). As previously noted a significant aspect of my analysis will be the comparison of how various forms of media, with statistically varied consumers, present the same "issues". As I am both using multiple texts and comparing the same "events" as delineated within those texts from a feminist point of view, a "feminist intertextual deconstructionist", or as Graham says, a "feminist multi-text analysis" seems the appropriate framework for qualitative data analysis.

This form of interpretation seeks to look "for contradictions within or between texts that illustrate the pervasive effects of patriarchy and capitalism (Reinharz, 1992, p.149)". Some of the contradictions I imagine encountering based upon my initial gathering of data sources are: claims that Calista Flockhart is anorexic versus the repetition of rumors of an eating disorder, references to Ally's eating disorder versus Calista's, references to Ally as a feminist versus references to Calista as a feminist, presenting the same photographs with multiple interpretations, etc. Again, this list is not exclusive but rather a few examples. Also, as my approach at this point of my analysis will be inductive, I have no way of knowing the nature of all of the contradictions within or between texts that I will discover.

Graham encourages feminists to deconstruct dominant texts in the following way (She references a multi-text analysis she was working on):

Dominant readings need to be deconstructed in order to make sense of the specific ways texts teach their audiences to structure personal systems of meaning. Through deconstruction, readers can find in each text the information to construct oppositional readings... As I read and reread each of these, I found myself repeatedly tempted to compare each to the one that came before... Instead of succumbing to this realist temptation to identify a "true" version, my idea is to be reflective about this nasty tendency and to put it to use methodologically...By finding the points of discontinuity between the texts, once can illuminate the mythologizing strategies and tools unique to that text. (Graham as quoted by Reinharz, 1992, p.149)

While I think it is probably self evident, in the interest of explicitly outlining any of my underlying assumptions it seems worth mentioning at this point that all of the texts I will be investigating are produced for mass consumption by leaders in the news, magazine and tabloid industries. These texts are all components of contemporary mainstream culture, and as such I will be defining them as "dominant readings" in Graham's terms. Through a detailed deconstruction of the multiple texts I have selected during the methodological phase, and the construction/placing of my findings in a feminist post-structuralist frame during my theoretical linkages, I find this particular form of feminist content analysis will best allow me to locate the dominant, hidden and excluded themes and contradictions surrounding Ally McBeal. I will then link this representative character to the larger social-political context.

For example, by deconstructing these multiple dominant texts I will begin to uncover the processes by which Ally McBeal, a fictitious television character, came to be labeled "the postergirl for postmodern feminism". As a feminist I am both personally and intellectually concerned when a fictitious character is defined as the 1990's symbol of feminism. Or put differently, perhaps defined as the feminist ideal for a generation. Who has been responsible for the processes of production that constructed these conceptions/images? To what extent have these definitions permeated mainstream culture?

By deconstructing these dominant texts I will, metaphorically, peel back the layers of an onion and reveal both the themes in each layer of the texts as well as the processes of production that developed-reflected-maintained these themes. This work is, I believe, a contribution to feminist research (beyond methodologically) in that the mass production of the symbol of feminism in the 1990s which is then commodified, impacts conceptions and discussions surrounding feminism on multiple scales. When thinking about this issue one can merely look at the cover of TIME magazine that deemed Ally McBeal the feminist of the 1990s as the heading read: "IS FEMINISM DEAD?". I find this to be a crucial issue in need of feminist exploration.

In addition to my positionality regarding feminism, as a post-structuralist I find this blurring of fact and fiction, the selling of the past to the present (i.e., TIME cover), and the commodification and consumption of social justice issues both intellectually fascinating and profoundly frightening. Nevertheless, through a feminist multi-textual deconstruction of these materials I will ascertain information about the ways in which this creation of a "feminist" symbol/image illustrates "the pervasive effects of patriarchy and capitalism" (Reinharz, 1992, p.149)".

While Ally McBeal has been constructed in the media as the feminist "ideal", similarly Calista Flockhart, (at least I think Calista and not Ally), has been constructed-re(presented) as the culture's "beauty ideal". Both of these feminist issues will be explored in order to reveal the presence and power of white heterosexual male domination in the processes of image production and distribution. Moreover, how have these two symbols/images/ideals, both stemming from Ally McBeal, influenced (both reflecting and mediating) mainstream culture in terms of gender related issues. As a final note on this matter, when looking at how Ally/Calista's image has been used to represent both postmodern feminism and the body beautiful, I will be working from Prior's conception of "text as product", which she explains in the following way:

...a representation should be understood not as a true and accurate reflection of some aspect of an external world, but as something to be explained and accounted for through the discursive rules and themes that predominate in a particular socio-historical context. The task of the researcher is to disentangle the rules of association by means of which the representation is structured, the genealogy of the various elements contained in the text (such as the points at which new terms and concepts enter the text), and the image of 'reality' which the text projects... In many respects one might say that discourse empowers certain agents to create representations, and thereby to authoritatively pronounce on the shape and form of the world. (Prior, 1997, pp.70-71)

I will now address one final issue regarding the content analysis of textual data prior to moving into a discussion of the analysis of visual/audiovisual data. I have collected two pieces of data which I would categorize as anomalies of sorts and thus while these data sources have not been randomly selected, but rather intentionally acquired, they will be examined as critical factors in my overall analysis. I will briefly address each of these anomalies in order to clarify their special place within this project.

The first is the June 19, 1998 TIME magazine which served as the impetus for Ally McBeal being defined as the "postergirl for postmodern feminism". In fact, many of the other articles I will examine specifically refer to that issue of TIME magazine. This particular magazine is therefore an integral part of my analysis and the content (visual and textual) will be interpreted in great detail. In order to place the article into a larger context, as it is the only issue of TIME that I will analyze (and the only article of its specific sort), I will discuss (within my literature review) a journal article titled "The Transformation of Time Magazine: From opinion leader to supporting player". This article written by James L. Baughman (1998) was published in the Fall 1998 issue of Media Studies Journal. While I will not delve into the content of the article at this time, in short, it assesses the place of TIME magazine as an "opinion leader" at this point in history. As my analysis of TIME magazine's June cover story will explore how the issue helped shape public conceptions regarding feminism, placing my analysis into the context of Baughman's work will be valuable.

The other anomaly that I want to briefly address is the way I will be looking at two specific pre-selected issues of People magazine. One of the issues was published immediately prior to the public rumor that Calista Flockhart was/is anorexic and the other was published in response to those rumors. Calista Flockhart's picture appears on the cover and inside of both issues in addition to written text. What makes these two magazines so important is that they both utilize the same photographs of the actress however the written analyses oppose each other (and remember, they both appear in different issues of the same magazine). Given the theoretical framework of my analysis, (i.e., Berger: how we see as effected by what we know or believe we know to be true), the comparison of these two issues of People magazine is crucial. I will be addressing the importance of temporally sequential analyses in my discussion of visual data and therefore will move on at this point.

Working with Visual Data

In Visual Anthropology: Photography as a Research Method Collier and Collier (1996) outline the multiple ways in which researchers can make use of visual/audiovisual data. They discuss both the advantages and complexities involved in working with visual/audiovisual data and offer a variety of methodological techniques researchers can employ during all phases of visual content analysis. The following discussion will be, primarily, based upon their text.

I will be sampling visual and audiovisual data from three populations. As discussed earlier, given that I am utilizing a multi-textual and multi-media approach (and the notion of a representative character) in this project I think it is important to sample visual/audiovisual data from the multiple sources that utilize Ally/Calista's image. Collier and Collier also state the importance of selecting material in such a way that "provides a sufficient reflection of cultural circumstance from which to establish a reliable perspective (1996, p.162)". As issues of validity and reliability are particularly paramount within feminist research, I have selected a comprehensive group of populations from which to randomly select the smaller samples I will be working with. In this same vain, Collier and Collier also state, "The whole view can only be viewed and recorded in the form of a responsible, selective composite" (p.163).

First, I will analyze the visual components of the texts I have already discussed and correspondingly they will be chosen as a result of my previously discussed sampling procedures. Second, I will systematically obtain a sample of photographic images of Ally/Calista that do not have textual components. These pictures will include promotional shots of Ally McBeal and fashion photographs of Calista Flockhart. Lastly, I will obtain a random sample of four episodes of Ally McBeal which I will interpret. Prior to analyzing these selected episodes of Ally McBeal I will provide a discussion of the program's general/reoccurring themes. I will disclose that I am doing so not from intensive investigation but rather the vantage point of someone who has seen every aired episode of the program.

At this point I will address several of the "principles" in visual research as described in Visual Anthropology (Collier & Collier, 1996). The author's state that a researcher can analyze the behavior represented in visual data when provided with a reasonable context including the identity of the participants. Ideally, a researcher will have detailed information regarding both the contextual setting of the visual document and the process by which the photograph was produced. Sequential records, as will be employed in the Ally McBeal project (previously discussed), can reveal developing patterns (Collier & Collier, 1996, p. 166). This is particularly important to my work as one aspect of the project is the way in which the same images have yielded very different interpretations based upon the "knowledge" available at the time.

A basic assumption of visual research, and an underlying assumption of my entire content analysis based endeavor, is that there is an independent voice in visual data (Collier & Collier, 1996, p. 170). This is because photographs (re)present primary experience.

As in all qualitative research, the analysis of visual data moves from raw data to conclusions. As such, a process of reduction occurs. Collier and Collier fear that a highly micro-analysis of the parts of the text can distort a researcher's vision of the whole document including the structural components and patterns within the content. They also contend that when a researcher breaks up the fluency of the record the conceptual stimulus is destroyed. Researchers must transcend the details in order to conceptualize the record in its pure form (1996, p. 170). In light of this, the exclusive use of reductionist codified approaches should be avoided and rather a combination of open and focused interpretive procedures should be employed. Ideally, by conducting analysis in this manner the data will lead to its own conclusions.

As mentioned, while analysis moves from raw data to conclusions, visual analysis involves decoding visual components into written means of communication. This process of translation is a key component of visual content analysis. Of course, as Collier and Collier point out, researchers must be careful not to assume that a visual document can be fully transferred into written form. This process of decoding provides a transitory phase where a document or illustration becomes the basis for systematic knowledge (1996, p. 170). This is possible due to the authentic voice within the data, but again, can be diminished by reductionist approaches to analysis. The "authentic" voice that Collier and Collier speak of is parallel to Reinharz's contention that documents contain a naturalistic dimension and are non-interactive. Regardless of the medium of the data, we can see that many of the same principles thus guide all forms of content analysis. As some fundamentals of visual content analysis have now been noted I will turn to analysis techniques.

The basic model for analysis consists of the following four stages: 1) Open Immersion 2) Inventory or Logging Process 3) Structured Analysis 4) Complete Viewing (Collier & Collier, 1996, pp. 178-179).

Working with Audio-Visual Data

Collier and Collier assert that moving images are the most comprehensive visual records (1996, p. 176). They do however provide specific complexities in that they combine visual, sound and nonverbal behavior (and the relationship between the three). The primary challenge in using audiovisual data is the volume one must manage which could be hundreds of frames. They offer the following suggestions when working with moving images (in my case, episodes of Ally McBeal):

  1. Slow motion viewing may be a key process in analysis
  2. Viewing at high speed may also help the researcher to see different patterns
  3. When logging using video footage counters can aid in precision
  4. Moving images can be made into still images
  5. During the structured analysis, counting and measuring groups of sequences can aid in credibility
  6. Looking at the relationship between sound and images can lead to new insights about the visual record. One can particularly note pacing, peaks of communication and peaks of nonverbal behavior
  7. While a solely micro-analysis would be wholly reductionist, a micro-analysis can be useful during the structured analysis to examine the relationships between visual, sound and nonverbal behavior
  8. Survey viewing can be used to make the records more manageable (1996, pp. 177-193).

When making conclusions researchers can make comparisons of similar situations and/or contrast different ones. One should not be afraid to shuffle the images and/or create montages that link the records in new ways. Finally, in accordance with my own beliefs, Collier and Collier remind the researcher to use her own creativity in this process. Ideally, the researcher will combine an artistic approach with meticulous analysis (1996, p. 199). In the end, the researcher will reduce volumes of information into, hopefully, its own essence. This is how I have constructed my project by analyzing some of the data with a computer driven program and some data by "hand" (as previously discussed) I will maintain a creative and individualistic approach to my analysis although it is still systematized.

Concluding Thoughts

As seen many of the principles and techniques involved in analyzing visual data parallel those regarding textual data; however, visual and audiovisual data do present their own set of complexities. I will be experimenting with the unique strategies outlined throughout this paper in order to link my representative character to the larger social-political-cultural context. By combining intensive methodological procedures with an established theoretical framework, and a little bit of creativity, this project should at the least prove to be interesting.

Post Script

As the project outlined in this paper has now been completed I would like to briefly comment on the effectiveness of the methodological framework employed. Overall, the project was successful in terms of meeting my goals. I found that the methodology was 1) appropriate to the study, 2) comprehensive (accounting for all phases of research) and 3) systematic yielding trustworthy results. I was also able to exercise creative liberties and maintain political integrity from a feminist perspective.

I would however modify one aspect of data analysis. As opposed to interpreting part of the data inductively from a grounded theory approach and part deductively employing preconceived code categories I would eliminate the latter deductive method and employ a grounded theory approach consistently throughout the study. The benefits to using both methods, as detailed in this paper, seemed to be outweighed by the cost when put into practice. Although I used peer review to gain "face validity" when constructing my code categories, this interpreted data lacked an authenticity in relation to the raw data that did not occur when grounded theory was used and emergent theory developed directly from the data. As I was working from within a feminist deconstructionist frame, the use of preconceived code categories felt contrary to the theoretical and methodological principles upon which my work was based. However, while in the future I would reconsider this aspect of my methodology, the process that lead me to this conclusion has served me well.


      Babbie, E. (1998). The practice of social research (8th Ed.). Blemont, CA: Wadsworth Publishing Co.

      Baty, S. Paige. (1995). American Monroe: The making of a body politic. Berkeley, CA: The University of California Press.

      Baughman, J. L. (1998). The transformation of Time Magazine: From opinion leader to supporting player. Media Studies Journal, 12(3), 120-127.

      Bertsch, C. (1998). The personal is paranormal: Professional labor on The X-Files. American Studies, 39, 107-128.

      Bertsch, C., Epstein, G., Newitz, A., & Sandell, J. (1998). The personal is professional on TV: Introduction. American Studies, 39, 87-92.

      Collier, Jr., J., & Collier, M. (1996). Visual anthropology: Photography as a research method. Albuquerque, NM: University of New Mexico Press.

      Epstein, G. (1998). Crafting the politics of "Special Space": 90210's scripting of media enjoyment in the private sphere. American Studies, 39(2), 129-139.

      Hesse-Biber, S., Dupuis, P., & Kinder, S. (1991). HyperRESEARCH: A computer program for the analysis of qualitative data with an emphasis on hypothesis testing and multimedia analysis. Qualitative Sociology, 14(4), 289-306.

      Martineau, H. (1995). How to observe morals and manners. New Brunswick, NJ: Transaction Publishers.

      Neuman, W. L. (1997). Social research methods: Qualitative and quantitative approaches (3rd Ed.). Needham Heights, MA: Allyn and Bacon.

      Newitz, A. (1998). ER, professionals, and the work-family disaster. American Studies, 39(2), 93-105.

      Prior, L. (1997). Following in Foucault's footsteps: Text and context in qualitative research. In D. Silverman (Ed.), Qualitative research: Theory, method and practice (pp. 63-79). Thousands Oaks, CA: Sage.

      Reinharz, S. (1992). Feminist methods in social research. New York: Oxford University Press.

      Sandell, J. (1998). I'll be there for you: Friends and the fantasy of alternative families. American Studies, 39(2), 141-155.

Author Note

+Patricia Leavy, Ph.D. candidate in Sociology at Boston College specializing in qualitative methods, feminist theory, postmodern theory and deviance/social control. Lecturer of Sociology, Criminal Justice and Women's Studies at Curry College in Milton, MA. E-Mail: leavyp@bc.edu.

Article Citation

      Leavy, P. (2000, May). Feminist content analysis and representative characters. [70 paragraphs]. The Qualitative Report [On-line serial], 5(1/2). Available: http://www.nova.edu/ssss/QR/QR5-1/leavy.html

Patricia Leavy
2000 copyright

Return to the top of the paper.

Return to the Table of Contents.