The problem of interpretation in vignette methodology in research with young people

Relationship to the ‘real’ and methods of interpretation

A key assumption that is frequently made when vignettes are used in research in the social sciences is that ‘narrative representations of emotional events can be treated as functionally comparable to the corresponding real-life encounters’ (Parkinson and Manstead, 1993: 296). Thus, when working on the construction and development of vignettes, it is important to make sure that the materials are meaningful and realistic for participants. This is a consideration in the design of vignettes and a key aspect of critiques of the method. Sleed et al. (2002) illustrated this in their concern about the composition of vignettes:

Can such brief written narratives adequately capture the reality of the context, and thus elicit responses that would be similar to reactions to real world situations?

Some researchers who employ the method acknowledge that it cannot ‘fully capture the elements of reality under study’ (Hughes and Huby, 2004: 45) but suggest that the method is useful precisely because of the schematic nature of the material. The lack of detail in vignettes that are given to participants means that they are led to fill in the gaps that may reveal important data for the research project. Thus, the participants’ interpretation of the vignette material becomes valuable material for study rather than a weakness in the design of the research instrument.

Another assumption that is commonly made is that participants project their feelings, views and social norms onto the character in the vignette. While there is developed literature on projective testing (see, e.g. Miller (1998) for an account of projective testing in developmental psychology), the majority of recent researchers using vignettes do not provide a theoretical account of the basis of these assumptions. They do not, for example, espouse the psychodynamic framework that has been applied to the psychological interpretation of projective responses. However, when researchers have used vignettes to attempt to study how people might behave in real life, there is an underlying cognitive theorization, which assumes a straightforward link between stated attitudes and behaviour. Different researchers have reported different findings in relation to links between stated behaviour in their vignette research and actual behaviour. For example, Rahman (1996) argued that in her work on female carers of older people, their responses were very similar to what their real-life responses would be. Similarly, McKeganey et al. (1996) found that intravenous drug injectors were more likely to talk about sharing needles in the vignette study than in a self-report study, therefore, apparently giving truthful information and not responding to the vignettes in a socially desirable way. However, in a study about child protection referrals, Spratt (2001) argued that emotion played a key part in decision-making in ‘real life’, which was not possible to reflect in a vignette methodology. Hughes and Huby (2004) discussing Spratt’s work suggest that the vignette methodology did not give a full context for the case/referral or in which decisions were made under high pressure, as typically experienced in professional practice. It is evident that the social workers had limited knowledge of the case described in the vignette, and this would have impacted their decisions. Vignette methodology cannot easily replicate ‘real’ conditions to test what would actually happen in practice. However, an alternative way of framing a researcher’s interpretation of participants’ responses to vignette material may be to explore the dominant shared representations/understandings of practice that are expressed and the less dominant ones that may emerge due to the personal trajectories and specific experiences of a participant (e.g. having had experience of a similar case). The focus would not then be on what participants would actually do in a specific situation but on their subjective perceptions, feelings and experiences.

Explaining diverse voices in the data: differing theoretical views

Traditionally, the focus for discussion around vignettes has been in two areas – first, the correspondence between participants’ ‘real-life’ behaviours and their comments on the vignettes, and second, whether opinions stated in the research represent a consensus view of the topic (Hughes and Huby, 2004; Parkinson and Manstead, 1993). Where results are not what were expected, some researchers have discussed the findings as indicating a problem with the methodology. One explanation is that the materials may not be written in sufficient detail and do not provide sufficient contextual information to meaningfully address the reality of the topic. Neff (1975) suggested that the more hypothetical vignettes appear, the less likely it is that reactions will correspond to the actual behaviour. Another explanation is that participants may feel embarrassed to reveal their true feelings, relying instead upon socially accepted values and beliefs. It has been argued that vignette techniques often tap into ‘socially desirable patterns of responding’ (Hughes and Huby, 2004: 43), where participants often express less personally threatening publicly accessible morality (Finch, 1987; Whittaker, 2002). In this article, we argue that a participant’s view cannot be separated out from a socially sanctioned version.

Most authors who review the vignette technique point to the problems of interpretation from the viewpoint of the participant. Hughes and Huby (2004) argue that participants may give reactions to the vignettes that they think others would give rather than risk giving their own view. Hughes (1998) discusses the potential for the researcher to encounter difficulties analysing data where some participants respond as if the character is himself or herself, while others respond in the third person. It seems evident that ‘some social stake is always involved in the version of reality that is presented’ (Parkinson and Manstead, 1993: 320) and that participants may choose to give a diverse range of responses to a topic, varying from their own self-disclosure, to socially stigmatized views and repeating socially acceptable explanations.

There can be little doubt that vignette-based experiences are different from real-life but whether this in itself makes it an unsuitable research tool depends on the rationale for using it. (Hughes, 1998: 384)

Vignette methodology can be difficult to analyse and can be problematic if researchers wish to disentangle the socially sanctioned, ‘safe’, normative interpretation of the vignette material from what participants ‘really’ think. Similarly, if researchers are using the method as a way of accessing what people really do in practice, the method can be problematic. However, if researchers begin with a theoretical stance that assumes that individuals will be in a constant state of dialogue with the self and others, encompassing many different positions and perspectives (‘voices’) on anything that impinges on them, then vignettes can be used by researchers in a more productive manner. The method can provide ambiguity in the materials, to access different positions on a topic under investigation. Thus, a key property of the use of vignettes in research is their flexibility. Vignettes can enable identification with socially approved views (the ‘generalized other’) or the expression of agreement with normative assumptions, or in some instances, the sharing of stigmatized or self-disclosed information. For researchers with an interest in exploring the range of accessible perspectives, problems of interpreting participants’ voices become an asset of the methodology rather than a problematic feature of the method.

Rather than drawing on a psychodynamic or a cognitive theoretical interpretation of the data produced by a vignette methodology, we argue for an interpretation drawn from theoretical insights offered by Hermans (2002). He drew on Bakhtin’s (1973) argument that monologic approaches to dialogue, that is, looking purely at the words said by the individual, could not tell the whole story because the utterances are not just the creation of the self but also relate to positions and opinions given by others in relation to self. Therefore, singular or monologic approaches to understanding dialogue produced from a well-constructed vignette cannot tell the whole story because they fail to recognize the interactional positioning, which is constantly in the process of being re-created in dialogue. The work of Bakhtin has been taken up by many theorists who have developed understandings of dialogue and the relationship between sociocultural accounts of the relationship between psychological functioning and dialogue (e.g. Hermans, 2002; Skinner et al. 2001; Wertsch, 1991).

Our position in relation to a theoretical and methodological orientation to research using vignettes is also informed by Hermans’ (2002) view that the self functions as a ‘society of mind’, in which the voices of the individual and collective other enter the self-space. The self in this perspective is a dynamic dialogical structure composed of a multiplicity of internal and external identifications. I-positions are voiced and engage in dialogues, which can reveal dynamics such as tension, conflict, agreement and so on. Conflicting positions can coexist within the self, in the same way that they coexist in society. The dominance of one position over the other (such as ‘I as a school child’ and ‘I as my mother’s carer’) is not fixed and can vary both over time and space.

Hermans (2001) has proposed a ‘personal position repertoire’ as a new methodology for studying, both qualitatively and quantitatively, ‘the organization and reorganization of a person’s position repertoire with attention to the personal meanings that are associated with the different positions’ (p. 323). While this is not the methodological approach we adopted in the research discussed in this article, our work does draw on insights from Herman’s theoretical work. The notion of space as a metaphor for understanding different voices and positions is important in Hermans’ work where people in talk not only move between I-positions but also move between ‘me and mine’ and ‘I and you’, illustrating the potential for many possible internal dialogues. We were also interested in the ‘matrix of internal and external positions’, – that we have both internal and external positions, which arise from our ‘person-world interactions’ (Hermans, 2001: 325). Internal positions can be social positions, that is, ‘equalized with the traditional term “role”’ (Hermans, 2001: 330):

Internal positions receive their relevance from their relation to some of the external positions. All these internal and external positions are part of the self (I-positions) and their significance is derived from their mutual transactions over time. (Prokopiou, 2007)

Hermans also addressed personal positions that may be in contrast to societal expectations. In addition, positions shift over time because of changes to their ‘frame’ (Zittoun, 2006). Aveling and Gillespie (2008) argue that drawing on Hermans’ notion of the dialogical self ‘can be used to elucidate identity dynamics within a heterogeneous sociocultural context’ (p. 3). In Aveling and Gillespie’s work, the identity positions of young British people of Turkish origin were outlined, and various identifications, such as ‘I as Turkish’, ‘I as a young Turk in England’ and ‘I as ethnic other’, were explored. Through this process, Aveling and Gillespie could discuss the ways in which a hybridized identity of the ‘young Turk’ existed through various dialogical tensions that allowed or foreclosed different identifications for the young people. They argue that a focus on dialogicality allows them to push the notion of multiple identity identifications further to examine how they fit or fail to fit with self-identification and sociocultural context. Similarly, Grossen and Salazar-Orvig (2006) discussed positions taken by participants in their focus groups. Students discussed five dilemmas of medical confidentiality and HIV infection, positioning themselves in various ways, including ‘I think’, speaking in the name of a particular group ‘we think’ and speaking as the character in the dilemma and generic talk. Drawing on the notion of dialogical and multiple positions of self and identity provides a very powerful analytical tool to re-think the interpretation of findings in vignette methodology, as will be illustrated in the analysis presented in this article.

Design and analysis of a vignette study

The data drawn on in this article form part of a larger study of childhood and what are viewed as ‘atypical’ roles for young people funded by the Economic and Social Research Council (ESRC) (RES-000-22-0549; Crafter et al., 2009). The project collected data in two phases: an initial survey of (very broadly defined) working activities of 1002 young people in two year groups at school (aged 15–16 years and aged 17–18 years) and a second phase of in-depth interviews with 46 young people identified as being either young carers, language brokers or young people who engaged in more ‘typical’ work roles such as having a Saturday job. Participants were recruited from six schools and colleges in the southeast and south coast of England.

The second phase of our research project used individual interviews built around four story vignettes, two depicting young people engaging in typical work roles for British young people such as babysitting (an informal arrangement where teenagers are paid to care for children for a few hours) and having a Saturday job (in a British context this is typically a part-time weekend job) and two depicting the atypical work roles under investigation – language broker and young carer. The vignettes were designed to represent aspects of children’s work that were identified by the research team to be of theoretical significance. The characters were 14 years old, slightly younger than the participants to allow them to identify with the characters and to discuss ways in which they related to the character. The vignettes were piloted with participants in a college comparable to the schools and colleges used in the project. The vignette materials were given to a group of 16- to 17-year-old students in the pilot college. The students were asked to give feedback on the clarity, appropriateness and plausibility of the materials. The vignettes implied a variety of cultural backgrounds (indicated by culturally specific or ambiguous names such as Samuel and Mira). For example, ‘Mary’ a young carer:

Mary is 14 years old and lives with her dad and her brother who is 15 years old. Mary’s dad is disabled and needs help during the day with activities such as getting out of bed, getting dressed and making lunch. Mary loves her dad and is happy to be there for him. However she also misses school some days if her dad has a bad day and needs extra help. Sometimes Mary wishes that she could see her friends after school like her brother does.

There were standard questions about each vignette (such as ‘What advice would you give Mary if she was your friend?’ and ‘What do you think their teacher would say?’ as well as questions that involved the participant comparing the four stories (such as ‘Which child has the hardest job and why?’, ‘Do you know anyone like these young people?’ and ‘Which character is most like you?’). Thus, the questions were designed to encourage dialogue between various positions open to the participant, including their own ideas and positions that they thought were held by key figures in their world. In this way, the structuring of the questions was planned to stimulate responses that would refer to different forms of ‘self-identifications’ (e.g. identifying with the character) and ‘other-identification of the self’ (e.g. how key figures such as friends and teachers would position the character).

Interviews were conducted by two of the research team in the student’s school. The participants were given the opportunity to participate or withdraw from the study. A number of students were interviewed from each institution, information about the kinds of jobs performed by the participants was kept strictly confidential and all of the interviews were conducted in the same way.

Louise’s positions on the young carer vignette

Louise was an 18-year-old White British young woman who cared for her disabled mother. In the following section, we illustrate the various moves Louise made through positions of identification with, and making sense of, the vignette of Mary a young carer. The final section discusses the contradictions that arise from Louise’s adoption of different positions.

Discussion

In this article, we have used empirical work from our recent research project to argue that it is possible to encounter difficulties in how to interpret responses when participants shift between discussing the vignettes as themselves, the character and what ‘ought’ to happen. By considering a theoretical position that assumes poly-vocality, the purpose of the method can be reinterpreted. In this article, we have illustrated how when using dialogical self–theory, the shifts between positions can be seen as instances of dialogue between different positions of the self and also between internal and external positions of the self. In the data presented, we have demonstrated that the methodology allows differing positions presented by participants to be documented and scrutinized. The key issue for researchers is to design the materials with appropriately structured questions that enable and facilitate the exploration of participants’ voices and I-positions. The processes of vignette methodology need to address the data collection process and the social context of the research. For example, in the research presented in this article, we examined the context of young carers through a school setting. Thus, the interview was designed to stimulate talk about key characters in this setting – the teacher, school friends and, as we wished to capture issues related to roles that conflict with demands from the family, the parent.

Hermans argued strongly that multi-voicedness does not mean that participants do not have a stable sense of self or do not know their own minds. Rather, it demonstrates the move between various legitimate social and personal positions that become relevant at different points to make sense of the topic in question. Grossen and Salazar-Orvig (2006) argue that the self is fundamentally heterogeneous (made through different voices) and also through systems of activity. Thus, determining ‘who speaks’ is not as self-evident as it might seem (Grossen and Salazar-Orvig, 2006). Wertsch (1991) argued that taking into account heterogeneity of voices requires a consideration of why particular voices take centre stage or are privileged in certain settings. The reason for emphasizing this point is that one of the issues in dialogical self-theory is that only recently it started taking into account the wider social context (O’Sullivan-Lago and Abreu, 2008).

Skinner et al. (2001) argued that ‘the methodology of performing any analysis of dialogical process is still in its infancy’ (p. 3). There have been some significant contributions to addressing dialogicality. Hermans (2001, 2002) has outlined a research method to document and analyse dialogicality. It produces a personal position repertoire in grid style. Aveling and Gillespie’s work also offers a strong framework for the analysis of dialogical identity. Their work was based on individual and group interviews. We would argue that vignette methodology may also be used as a medium for theoretically engaged work through which to explore issues of dialogicality.