Visual Methods in Social Research. Marcus Banks
sister unless we know her too or were told. If we are shown the same photograph in the pages of a magazine, with certain textual elements attached, then we are more likely to assume the two are actors, dressed up for an advertising shoot to sell wedding attire. Either way, we draw upon internal and external narratives in our reading: in the one case to tell ourselves a story of romantic love within a familial context; in the other to tell ourselves a story about consumption and the commodification of romantic love within a possibly global context. What is apparently the ‘same image’ can be a presentation of self (this is how I want to be seen, such as a photograph posted on Facebook) or a commercial action designed to influence others (an advert). We note too that these distinctions can be applied not only to photographs and films but also to clothing and accessories worn by humans and their online avatars (see Section 1.4.1 below).
As scholars of the visual we want to encourage researchers to take visual evidence seriously. The general approach can be applied to the study of fashion and adornment or body art, to advertising images and to the images an individual posts online of parties and the food they eat. It also applies to the images that a researcher may produce in the course of undertaking a research project.
1.4.1 Formal readings5
Although our discussion so far has rested on the interpretation of an image’s meaning or context, there are more formal ways in which this can be pursued. Often known as content analysis, these forms of analysis can be quantitative or qualitative in nature (see also Rose 2012: Chapter 5).
To give an idea of the topic consider the claim going back to Durkheim that styles of dress and adornment are ‘social facts’; that although each individual makes personal choices they are swayed by fashion – the spirit of the times which skews their choices in a certain way. So the hem lines of women’s skirts rise and fall with fashion, and the tendency of men to sport beards or moustaches changes over time. Robinson (1976) studied this by counting how many men had beards in photographs published in the Illustrated London News, over a period of more than a century, from 1842 to 1972. Goffman (1979) and many others have looked at how gender roles are presented in advertising imagery. Content analysis adds substance to the claims of bias. However, we must always remember that simple counts of items (such as beards) in the manifest content of an image – that is, the obvious ‘things’ in the picture – may not be meaningful or informative. Not only is the latent content of images significant (as observed in image composition, such as where women are placed in an image) but the mode of display, and how the audience uses the images are important factors that must be taken into account. Concerns about the amount of sociological presumption in much content analysis where a coding scheme is developed and applied to a body of ‘data’ e.g. a collection of photographs or a corpus of films (see Bell 2001: 17), has led David Altheide to develop what he calls ‘Ethnographic Content Analysis’ (1987), which is inductive, in the spirit of grounded theory (Glaser and Strauss 1967; see also Section 5.3.2 for another example), building from the material under consideration rather than judging it against pre-existing criteria (we recognize that there will be problems with this approach in comparative research projects, or those that involve the analysis of highly heterogenous material). Altheide has applied this to analyze the US broadcast news coverage of the US Embassy hostages in Iran from late 1979 to early 1981. The analysis is systematic (indeed partly quantitative) and sensitive to the specificities of the material under consideration.
There are, broadly, two approaches to content analysis. On the one hand, that of scholars such as Bell (2001) relies on an enumeration of features identified in the content of a film or a corpus of photographs created by others, which can serve to reveal latent features which may not have been intentionally placed or given prominence by the film’s maker or photographer, but which can be uncovered through coding and analysis. While Bell’s approach employs quantitative techniques, not all content analysis approaches do so. For example, Iedema demonstrates how his analysis of a documentary film about failings in the Australian hospital system contains a number of ‘patterns’ (typically, repeated sequences of shots with similar content) that serve to semiotically promote one particular interpretation of the failings and their resolution (Iedema 2001: 184).6 In order to identify these ‘patterns’ (for example, hospital administrators ‘usually’ appear alone in shots and sequences) Iedema needed to compile a descriptive shot list for the entire 30 minute film, something he notes is ‘laborious’ (2001: 200). Iedema does not apply any form of statistical analysis (unlike Bell) although he could have done so. There is, however, a rough and ready sense of building a case from the weight of evidence: he notes for example that shots of the hospital administrators are ‘often’ filmed from a low angle, which he suggests denotes their power (2001: 185). (We discuss further examples of content analysis approaches to the study of film in Section 6.1.)
This is in marked contrast to a completely different kind of content analysis, championed primary by Christian Heath and other colleagues working within an ethnomethodological paradigm (for example, Heath et al. 2010). Their form of analysis is self-consciously not quantitative (contra Bell but apparently in keeping with Iedema). In their recent book on video and video analysis (2010), not only do they not perform any kind of quantitative analysis, they explicitly reject it, something they note is markedly at odds with many other approaches in the social sciences (2010: 84). The main reason for doing this is simply that they generate the video data themselves; therefore while they could analyze it for latent meaning, deducible via content analysis of the former kind, this would, on the face of it, be a pointless task as they have already been quite self-conscious in how they set the camera up, framed the shot, and so forth. They are explicit about the fact that they pre-select ‘perspicuous settings’;7 that is, they choose to film arenas of social action most likely to generate the data they are seeking to capture. Therefore, as they have selected these arenas, and gained the advance permission of the actors to be videotaped in those arenas – such as doctor-patient interactions in the consulting room (2010: 30, 45) – there is little point in scrutinizing the ensuing video tapes for latent meaning, at least at the level of authorial intention. The authorial intention is blatantly manifest to the researchers precisely because they knowingly created it. There may indeed be latent meaning to be deduced from their video tapes at a meta-empirical level, but this is probably a task best left to future historians of social science epistemology.
For Heath et al. content analysis (not a term they use themselves) is about the meticulous transcription and analysis of micro-components of social action. There is still, however, an unspoken concern with latent meaning, this time with the latency of the subjects’ comprehensions and motivations as captured by the researchers’ video, rather than the latency of authorial intention.8
1.5 Planning a research project with visual methods
Good visual research rests upon a judicious reading of both internal and external narratives, whether by formal methods such as content analysis of the internal narrative, or less formal ‘readings’ of internal and external narratives. Approached as objects all visual objects represent nothing but themselves; their very existence in the world as material objects is proof of nothing but their autonomy (see Chapter 3). Yet, despite their uniqueness as material objects, they resemble all other objects in their class (as individual tokens of a particular type), although their uniqueness as particular manifestations of that class needs to be assessed by an initial reading of the internal narrative. All films, photographs and artworks are the product of human action and are by this fact entangled to varying degrees in human social relations; they therefore require a wider frame of analysis in their understanding, a reading of the external narrative that goes beyond the visual text itself.
One way of thinking about social science research methods starts with the seemingly bland generalization that our methods