Advances in Consumer Research Volume 21, 1994 Pages 501-507
USING THE ZALTMAN METAPHOR ELICITATION TECHNIQUE TO UNDERSTAND BRAND IMAGES
Robin Higie Coulter, University of Connecticut
Gerald Zaltman, Harvard University
Understanding consumers' interpretations of brand image has become increasingly important as firms have attempted to boost brand equity. Our paper introduces the Zaltman Metaphor Elicitation Technique, discusses its theoretical grounding, and examines its use as a tool for investigating brand image management activities.
COKE...NIKE...TIDE...LEVI'S...CREST...The shelves of stores and homes alike are lined with these and other brand names. Since the 1950s, companies have assigned brand names to their products as a means of differentiating them from the competition, and for several decades consumer researchers have studied brand names, brand extensions, and more recently, brand equity. The basis for the research lies in the meaning of the brand - its brand image. Recently, Biel (1993) suggested that brand image is composed of the image of (1) the provider (eg., manufacturer), (2) the user, and (3) the product, itself. Reynolds and Gutman (1984) use their means-end chain framework in proposing that brand image is a synthesis of product attributes, consumer consequences and personal values. Farquhar and Herr (1993) discuss the types of associations that brands evoke, including specific product attributes, customer benefits, usage situations and other summary evaluations. After an extensive review, Dobni and Zinkhan (1990) conclude that brand image is the consumer's subjective interpretation of a brand that is formed by marketing activities and context variables. If we think more about COKE, NIKE, TIDE and other brand names, we realize that they are, in fact, the embodiment of the product, the user, the producer, the marketer, and the use situation.
As firms attempt to develop and sustain stronger brands and build brand equity to maintain a competitive advantage in the marketplace, understanding customers' perceptions becomes increasingly important. Dobni and Zinkhan (1990) review a variety of approaches to assessing brand image, noting that the trend has been toward using quantitative methods. Qualitative procedures, including having customers' engage in thought listing procedures, describe a brand's personality, and define "what the brand thinks of you, as the customer" also have been used in brand image studies (Blaxton 1993; Boivin 1986; Durgee and Stuart 1987) and brand extension studies (Aaker and Keller 1990; Park, Milberg and Lawson 1991). Most of these research techniques rely on verbal communication to obtain customer information. Research on communication, however, suggests that over 80 percent of all human communication is nonverbal, and Biel (1993, p.73) notes that "brand images have a strong nonverbal component." Thus, there appears to be an imbalance between how customers think and communicate about brands and how researchers access customers' thinking.
Our purpose is to introduce the Zaltman Metaphor Elicitation Technique, a research tool that uses visual and sensory images to help better understand the meaning of brands.
Briefly, the Zaltman Metaphor Elicitation Technique (ZMET) employs qualitative methods to elicit the metaphors, constructs and mental models that drive customers' thinking and behavior, as well as quantitative analyses to provide data for marketing mix decisions and segmentation strategies. The Kelly Repertory grid and laddering techniques are integral components of the technique, as are visual (eg., photographs) and other sensory images that consumers provide. The constructs elicited during the interviews are aggregated to produce a consensus map which represents most of the thinking of most people most of the time. In addition, consumers use digital imaging techniques to produce summary images of their thinking.
To begin, we provide a literature review on nonverbal communication and metaphors and briefly discuss the asymmetry between how consumers' think about and experience brands and how most market research data are collected. Next, we define the ZMET steps that have been tested and provide an illustration of the technique's use to study brand images. Finally, we discuss the use of ZMET for understanding an array of brand image issues.
Mehrabian (1971) reports that approximately 93 percent of the meaning contained in any message is nonverbal (including paralanguage) and that only seven percent is contained in verbal language. Birdwhistell (1970) notes that no more than 30 percent of the meaning in a social exchange is conveyed by words. Other evidence indicates that most communication occurs nonverbally (Knapp 1980; Seiter 1988; Weisner 1988). As we noted, the rule of thumb among communications specialists is that about 80 percent of all human communication is nonverbal, and much of the meaning of verbal language also is determined by nonverbal cues. Further, research reinforces the idea that nonverbal communication is dominant (Montagu 1986; Stoller 1989; Howes 1991), and when there is an apparent contradiction, nonverbal cues tend to be believed over verbal ones (Knapp 1980, p. 86). Additionally, visual representations are processed differently than verbal messages and hence are not subject to the same logical scrutiny and counterarguing (Biel 1993). As a consequence, the images and information are more likely to be internalized, with the increased potential of affecting attitudes and behavior.
"The essence of a metaphor," according to Lakoff and Johnson, "is understanding and experiencing one kind of thing in terms of another" (1980, p. 5). It is only through their metaphors that we can better understand customer thinking and behavior and thus develop and market successful brands, goods and services. Even behavior and thoughts are themselves metaphors for one another. Many metaphors are mental images, and as discussed, the majority of these are visual (Arnheim 1969; Kosslyn et al., 1990).
Discussions of imagery and its relevance to advertising, marketing and consumer research that have implications for branding can be found in Lutz and Lutz 1977, 1978; Rossiter 1982; Rossiter and Percy 1978; MacInnis and Price 1987; and Bone and Ellen 1992.
Customer and Firm Metaphors. Understanding customers' metaphors (or knowledge representations) related to a brand and how these metaphors interact with those of the manufacturing firm or advertising agency is important. For customers, metaphors about a brand are found in their images, which may be visual, verbal, mathematical, and musical, among other forms. These images contain a customer's experience, understanding, and memory Cin short, their knowledgeC about the type of packaging, where the brand can be purchased, and the advertising campaign, for example.
Biel (1993) reminds us that when thinking of brands, visual representations often come to mind. This is particularly the case for well-developed brands. For example, upon hearing "Keebler," Keebler elves might come to the mind's eye, and "McDonald's" may generate a picture of the golden arches. Less well-defined brands are likely to spawn a more diverse set of images.
Because a firm's or ad agency's metaphors about a brand are conveyed in the form of various marketing mix decisions, including advertising, product and package design, product concepts, and distribution channels, these metaphors need to contain components that are both similar and dissimilar to their customers' metaphors. The similarities are necessary because customers are predisposed to find their own metaphors in advertising, the design of a product, or a store setting, for example. Thus, when marketer communications are congruent with customer-generated images, customers are more likely to attend to, process, and comprehend the observed information. The dissimilarities are also necessary; they create a degree of tension that attracts attention and sets the foundation for message comprehension. Creative staffs and product design staffs play a particularly important role in providing the creative "dissimilarities" component.
Pictures as Metaphors and Basic Concepts. Because much communication is nonverbal, pictures can serve as entry points for exploring other customer concepts (Weisner 1988; Ball and Smith 1992). Pictures typically represent basic concepts, which contain extensive information or defining attributes, and people usually recognize such concepts first. After customers recognize basic concepts, then they can infer or identify associated higher-order concepts (Smith 1988; Rosch 1978). For example, the concept of "Hershey's kiss" has more defining characteristics than the higher order concept of "chocolate." The concept of "Hershey's kiss" contains more information, is more readily recognized and is a more efficient way to begin processing information than the concept of "chocolate."
Because pictures are so basic, information-rich and attribute-laden, they can be associated with multiple related higher-order constructs. Pictures represent a natural and efficient way for customers to start communicating higher-order constructs. In fact, some clinical psychologists use a client's photographs as a central part of the therapeutic counselling process (Entin 1981; Krauss and Fryrear 1983; Weisner 1988).
Seeing the Voice of the Customer
As noted, most market research methodologies such as questionnaires and personal and group interviews rely on verbal forms of communication to elicit information. Given the prominence of nonverbal communication, an asymmetry exists between how market research data typically are collected and how customers usually think about and experience their worlds. A better understanding of customers' perceptions requires the use of research tools which engage customers' nonverbal and especially visual "channels" of thought and communication. In this section, we briefly review the literature relevant to photographs and other visual stimuli as research tools and as important components in marketing communications.
Gombrich, Hochberg and Black (1972) provide an excellent discussion of how pictures represent people and things, and about the representation of people and things in pictures. Given the expressive power of pictures, it is not surprising that photography as a research tool has been growing. Excellent discussions of the use of this tool in sociology, psychology, and anthropology can be found in Bateson and Mead 1942; Wagner 1979; Becker 1980; Collier and Collier 1986; Denzin 1989; Ziller 1990; and Ball and Smith 1992. Additionally, Denzin (1989) provides a good discussion of the validity and reliability issues concerning the use of film and photography generally as sociological research tools.
Consumer behavior researchers also have employed photographs as stimuli to elicit consumers' thought processes and/or develop theories relevant to their work. Work by Heisley and Levy (1991) entailed their taking photographs of people and interviewing those people about the content of the photographs. Holbrook (1987), taking on the roles of the researcher and subject simultaneously, took photographs of artifacts around his home and "interviewed himself," asking about the meaning of the pictures. More recently, Wallendorf and Arnould (1991) and their co-investigators did content analyses of photographs of Thanksgiving events to understand the rituals associated with the holiday. The very productive Consumer Odyssey project also made extensive use of still and video images (Belk, Wallendorf and Sherry 1989).
Researchers investigating advertising have examined the visual component of ads, and the interaction between the visual and verbal components. The pictorial content of advertisements has received attention by Rossiter (1982), Edell and Staelin (1983), and Mitchell (1986), for example. Studies have substantiated that pictorial ad stimuli are recalled better than verbal ad stimuli and result in more favorable product beliefs and brand attitudes (Lutz and Lutz 1977; Mitchell and Olson 1981). Childers and Houston (1984) note that pictorial ad content is more effective in delayed recall. Other research has demonstrated that semantically discrepant pictures and words can be combined in a way that yields better recall than that for semantically consistent pictures and words (Houston, Childers and Heckler, 1987). In a recent follow-up, Unnava and Burnkrant (1991) found that when verbal information was low in imagery, the inclusion of pictures consistent with the information increased recall of verbal information on both immediate and delayed post tests. When verbal information was high in imagery, adding pictures did not affect recall of verbal information. In addition, Miniard et al. (1991) discuss the imagery-evoking ability of pictures in ads and subjects' subsequent processing of information in the ad.
Several important points can be drawn from the literature. First, nonverbal communication is more prevalent than verbal communication. Second, visual images are entry points for accessing people's knowledge structures. Third, photography is a powerful tool for accessing consumers' visual images. Fourth, research documents the importance of visual images in marketing communications. Finally, most market research tools rely on verbal communication. Based upon these factors, we developed ZMET, a methodology that relies on visual and other sensory images to elicit customers' metaphors and constructs.
IMPLEMENTATION OF THE ZALTMAN METAPHOR ELICITATION TECHNIQUE
A total of 25 customers typically are recruited to participate in a project. After qualifying for participation in a project (based on screeners), customers are given a set of instructions and guidelines about the research topic, eg., a brand name, a service concept, product use, or product design. They are instructed to take photographs and/or collect pictures from magazines, books, newspapers or other sources that indicate what the topic means to them. Customers are provided with a camera, and a personal interview is scheduled approximately seven to ten days hence. The personal interview involves a guided conversation which we believe yields more valid, more reliable and, importantly, more relevant insights than traditional structured interview approaches (see McCracken 1988; Mishler 1986).
CORE STEPS TESTED
Customer Control over Stimuli
Because ZMET has customers collect their own pictures, the customers (not the researchers) are in control of the stimuli used in the guided conversation. Customer control has a number of benefits. First, customer-generated pictures are especially meaningful because what the eye perceives when it is encoding a sequence of information over time, including the viewing of a static image such as a magazine ad or a point-of-purchase promotion, is guided by existing customer knowledge, beliefs, or expectations. Second, requiring customers to collect the stimuli increases the likelihood of uncovering important, but previously unconsidered customer issues.
In addition to using photographs and pictures as stimuli to elicit constructs, ZMET relies on verbalization to establish and record images the customer wanted to collect but was unable to do so because of time or geographic constraints. The technique also uses an image bank created by the researchers. Other image capture techniques are also effective and the best combination of techniques will vary with the product-usage situation, the target audience, and the theme and objective of the research. Additionally, ZMET makes extensive use of nonvisual sensory images.
The Guided Conversation
The personal (one-on-one) interview or conversation with each customer takes approximately two hours, and it is audiotaped. The conversation focuses primarily on the images customers bring to the interview. During our initial R&D, we examined twelve topic areas, ranging from the meaning of snack foods to the meaning of a specific corporate image. The conversation guidelines include 9 steps, and the input from these steps serve as data for the researchers developing the consensus map. Exhibit 1 describes each step.
INVESTIGATING BRAND IMAGE USING ZMET
Several researchers have discussed a variety of components related to brand image and the interrelationships among these components. Recall, for example, that Reynolds and Gutman (1984) suggested images can be thought of in terms of product features or attributes that yield consequences (benefits or risks), which in turn facilitate values or end-states. Both Reynolds and Gutman, and more recently, Farquhar and Herr (1993) suggested that the aggregation of individual means-end chains results in a network consisting of the brand and its associates (eg., usage situations, product attributes, customer benefits and values). Moreover, the image that a person has about a brand consists of both verbal and nonverbal representations (Biel 1993). These, in turn, are best expressed via verbal and nonverbal channels, respectively. Nonetheless, most research that studies consumers' images of brands typically uses verbal research methods, and thus misses a great deal of what constitutes the individual's brand image.
The purpose of ZMET is to elicit metaphors and constructs and to establish relationships among the constructs using both verbal and nonverbal stimuli. Therefore, we suggest that ZMET is an efficient and effective means for understanding brands. To give the reader a clearer understanding of how ZMET can facilitate the understanding of brand image issues, we herein discuss the details of an interview with Alice, one of the customers who we asked to take and/or collect pictures of what "Tide" meant to her.
Alice, a young mother, collected 14 images. After completing Step 1 (storytelling about each picture), the interviewer asked Alice if there were any pertinent images that she had not been able to collect (Step 2). Alice indicated that she would have liked to take a photograph of a pig sty, noting that she wondered if "Tide" would be able to "clean a dirty pig." In Step 3, Alice sorted her images into three groups: comfort, freshness, drudgery.
In Step 4, the interviewer randomly selected three of Alice's pictures, and asked her, "How are any two of these three pictures similar to each other and different from the third." This surfaced two constructs: unpleasantness and freshness. The interviewer, using the laddering process, helped to elicit additional constructs and their relationships. After Alice's explanation, the interviewer continued to randomly select three pictures and question Alice until no new constructs were elicited.
ALICE'S "TIDE" TRIAD
Exhibit 2 illustrates one of Alice's triads. She indicated that the top picture represented "unpleasantness" and the bottom two pictures represented "freshness." Alice's picture of "two friends making ugly faces" described "unpleasantness." Her comments were "Tide means doing the laundry. It's time consuming and the laundry facilities are not well-maintained. Plus, it's always a hassle to find quarters. I really don't like to do laundry." Clearly, Alice's commentary related to this triad indicated that she associated the brand with the product use situation. Alice then described "freshness": "Using Tide gets my clothes clean and smelling very fresh. I feel more comfortable and refreshed in my clean clothes." Using the laddering process, the interviewer elicited the additional constructs of "makes me feel confident" (self-confident) and "tells people what kind of person I am" (self-image). In this latter case, Alice's comments focused on customer benefits and values.
Continuing with Step 5, Alice indicated that the picture that most represented "Tide" to her was her picture of the sunrise (shown in Exhibit 5). She reported that the sunrise contained several meanings: freshness (as in the smell of her clothing after using "Tide"), brightness (as in the colors of her clothes after using "Tide"), calm or peacefulness (knowing that "Tide" cleaned her clothes put her at ease), and accomplishment (a new day to get things done).
When the interviewer asked Alice what images conveyed the opposite of her image of "Tide" (Step 6), she responded with images of a bottle of acid ("'Tide' is strong, but not harmful") and a porcupine ("'Tide' leaves clothes feeling soft, not harsh").
ALICE'S SENSORY IMAGES
In Step 7, the interviewer asked Alice about other nonvisual sensory images of "Tide." Alice's responses are listed in Exhibit 3.
In Step 8, the interviewer reviewed all of the constructs that Alice had discussed and asked her if they were accurate representations of what she meant, and if any important ideas were missing. Then Alice created a map showing constructs and relationships she saw as related to "Tide" (See Exhibit 4). Alice "walked" the interviewer through her map, noting that "'Tide' brings laundry to mind, and when I think of laundry, I think that it's expensive, time consuming and unpleasant. All of it is very aggravating. Also, when I think of 'Tide', I think of a strong and dependable detergent - a detergent that cleans my clothes leaving them soft and fresh. Knowing that 'Tide' gets my clothes clean means I don't have any worries. Because my clothes are clean and fresh I feel refreshed, and because my clothes are clean and soft I feel comfortable. I feel more self-confident when I'm comfortable and refreshed, and I look better to other people. Finally, even though 'Tide' is strong, it is environmentally friendly, and that's important for the water and the health of Americans."
Alice, with the help of a technician, used digital imaging techniques to create her summary image of "Tide" in Step 9. Her digital image consisted of components of five of her pictures and visually depicted her story about "Tide." Exhibit 5, Alice's digital image, "shows my two baskets of laundry... and how I feel about doing laundry...I don't like to do it! You have to separate clothes and find lots of quarters. It takes forever to do laundry in my apartment building - someone's always forgetting to take care of their clothes. The picture also shows a sunrise which means freshness to me - a new day. If you look in the tree, you'll see a teddy bear - he represents how my clothes feel after they are washed in 'Tide', soft and comfortable. The teddy bear is holding a rose to let you know my clothes smell good when washed with 'Tide'."
ZMET Data Compilation
ZMET produces several types of information, including a consensus map, visual and sensory dictionaries, animated vignettes and a set of digital images. After completion of the interviewing, typically consisting of 20 to 25 interviews, a consensus map is developed which represents (a) most of the thinking of (b) most people (c) most of the time. Producing the map involves several steps. First, each interview transcript (selected randomly) is systematically examined, and the elicited constructs and the number of new constructs added by each additional customer are recorded. The construct list then is refined and condensed through content analysis. Finally, the researchers aggregate the data across customers and then assess the relationships (as given by respondents) among the pertinent constructs. (For a discussion of meta-ethnographic synthesis techniques, see Noblit and Hare, 1988.) The result is a consensus map which diagrammatically illustrates the linkages among the elicited constructs. These linkages are the reasonings which connect the constructs.
ALICE'S MENTAL MAP OF "TIDE"
ALICE'S DIGITAL IMAGE OF "TIDE"
Our R & D objectives for the "Tide" application did not require many interviews; hence, we did not develop a consensus map. However, consensus maps contain three types of constructs. First, some constructs are originating points in a reasoning process, these constructs lead to other constructs. Other constructs are destination or ending points in a reasoning process. Still other constructs serve as transmitters or linkages between originator and receiver constructs. Knowing the status of a construct as an originator, transmitter, or receiver has important implications for the use of a consensus map and the development of marketing strategies. A manager interested in a particular construct should not consider it in isolation, but also should be aware of the constructs that causally precede and follow it. A number of analytical tools and associated software are available to analyze network patterns among constructs and extract still more information from the consensus map.
In addition to the consensus map, ZMET produces nonverbal sensory dictionaries to accompany each construct in the consensus map. These user-friendly dictionaries are made up of the pictures and sensory data that different customers use to convey the constructs. The dictionaries provide a basis for gaining visceral or empathic feelings about the customers, because the relevance of each image to a particular construct is described in the customer's voice. Further, customers' digital images provide visual images that integrate their important constructs. Many images also convey the reasonings connecting constructs.
ZMET is a hybrid methodology grounded in a broad body of literature. The use of photography as a research tool has a fifty year history that crosses many disciplines, and the validity and reliability of photography as a general research tool has been discussed extensively by Denzin (1989). The in-depth personal conversation has significant merit in its own right (Mishler 1986; McCracken, 1988), and other research conducted by Griffin and Hauser (1992), Silver and Thompson (1991), Robinson (1991) and Fern (1982) suggests the efficacy of personal interviews compared with focus groups. The validity and reliability of the Kelly Repertory Grid and laddering technique as means for eliciting constructs are well established (Kelly 1973; Gutman 1982; Reynolds and Gutman 1988).
To-date, our R & D with ZMET involving over 400 interviews has yielded some important insights. First, nearly all participants regardless of demographic characteristics, such as educational or occupational experience, successfully engage in all steps. Steps 7, 8 and 9 which a priori might seem difficult because the tasks are unfamiliar, are undertaken quite readily.
Second, considerable interviewing practice is required to conduct the interview. Each step can be introduced in different ways and individual differences among customers may call for somewhat different approaches. Third, interviews average about two hours, a considerable amount of time to spend with a single customer discussing a very specific topic. This time frame affords an opportunity to learn not only about people's initial thoughts (much as might be obtained by a structured questionnaire or focus group), but also about the deeper meaning of a topic. At the same time, the interviewers must be sensitive to the issue of customer fatigue. Finally, after the interview, participants are given a follow-up survey (and a self-addressed stamped envelope) which assesses their experiences with ZMET. Typically, we get a response rate of at least 80 percent, and the participants indicate that the entire procedure, from taking/collecting images to the completion of their digital image, is enjoyable, interesting and involving.
ZMET has the potential for understanding many aspects of brand image management. First, as we have seen with the "Tide" example, ZMET can be useful in eliciting and understanding the sensory metaphors associated with brand images. These metaphors potentially are useful in a variety of marketing mix decisions, including developing advertising strategies, determining the shape and type of packaging, and selecting distribution channels. The information also might be useful for examining possible brand extensions. A dual application of ZMET would be needed - one investigating the current brand image and a second examining the potential extension. Researchers then could examine the two maps to determine their image-congruity. Similar maps imply that the prospective extension would be compatible with the existing brand image, whereas dissimilar maps may be a sign that consumers would have difficulty integrating the extension into their current knowledge structure. ZMET also is valuable for brand repositioning. Again, a dual application would be needed, the first to assess the current brand image and the latter to assess an "ideal" product image. Examining the parallels and gaps between the two maps would surface issues or components of image that warrant alteration. Implementing only the latter application, i.e., the "ideal" product image, would be worthwhile in establishing a new brand.
Conducting ZMET for multiple target audiences can establish each segment's consensus map. As a consequence, the marketing group would be able to determine whether a single strategy or multiple strategies were needed to communicate to the groups. We are currently investigating the segmenting of markets by decomposing a consensus map, i.e., considering whether all participants have the same mental model or if there are distinct clusters of respondents who have unique brand association networks. Finally, a dual application directed at corporate image and the brand image would provide information about their image-congruity. Clearly, there are numerous issues related to brand image and other facets of consumer behavior and marketing that could be addressed with ZMET.
Please contact the first author for a complete reference list.
Biel, Alexander L. (1993), "Converting Image Into Equity," in Brand Equity & Advertising: Advertising's Role in Building Strong Brands, eds. David A. Aaker and Alexander L. Biel, Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 67-82.
Blaxton, Max (1992), "Beyond Brand Personality: Building Brand Relationships," in Brand Equity & Advertising: Advertising's Role in Building Strong Brands, eds. David A. Aaker and Alexander L. Biel, Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 113-124.
Collier, John, Jr. and Malcolm Collier (1986), Visual Anthropology: Photography as a Research Method, Revised and expanded edition. Albuquerque, NM: University of New Mexico Press.
Denzin, Norman K. (1989), The Research Act: A Theoretical Introduction to Sociological Methods, 3rd ed. Englewood Cliffs, NJ: Prentice Hall, Chapter 9.
Dobni, Dawn and George Zinkhan (1990), "In Search of Brand Image: A Foundation Analysis," in Advances in Consumer Research, Vol. 17, eds. Marvin E. Goldberg, Gerald Gorn and Richard W. Pollay. Provo, UT: Association for Consumer Research, 100-119.
Farquhar, Peter H. and Paul M. Herr (1993), "The Dual Structure of Brand Associations," in Brand Equity & Advertising: Advertising's Role in Building Strong Brands, eds. David A. Aaker and Alexander L. Biel, Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 263-277.
Knapp, M. (1980), Essentials of Nonverbal Communication, New York, NY: Holt, Rinehart and Winston.
Kosslyn, Stephen M. and Oliver Koenig (1992), Wet Mind: The New Cognitive Neuroscience, NY: The Free Press.
Lakoff, George and Mark Johnson (1980), Metaphors We Live By, Chicago: The University of Chicago Press.
Mehrabian, Albert (1971), Silent Messages, Belmont, CA: Wadworth, 42-47.
Mishler, Elliot G. (1986), Research Interviewing: Context and Narrative, Cambridge, MA: Harvard University Press.
Stoller, Paul (1989), The Taste of Ethnographic Things: The Senses in Anthropology, Philadelphia: The University of Pennsylvania Press.
Wagner, Jon, ed. (1979), Images of Information: Still Photography in the Social Sciences, Beverly Hills, CA: Sage Publications.