Printable PDF Version Full Symposium Just this article Printable PDF Version Full Symposium Just this article
Issues of Validity: Behavioral Concepts, Their Derivation and Interpretation
The Pink Elephant Paradox (or, Avoiding the Misattribution of Data)Jude A. Spiers
Citation information for this section:
Spiers, J.A. (2002). The pink elephant paradox (or, avoiding the mistattribution of data). International Journal of Qualitative Methods, 1(4). Article 4. Retrieved DATE from http://www.ualberta.ca/~ijqm.
The pink elephant paradox refers to the threat to inductive thinking caused by the difficulty of inadvertently proving the existence of a concept or phenomena just because it overtly or insidiously exists in one’s thoughts, leading to misattribution, or miscategorization of data, and thus subverting inductive processes. As Morse and Mitcham discussed in Part I, this is reduced through inductive strategies, including processes of saturation, replication, and verification. In this article, I present a story of how the phenomenon of interest in nurse-patient interaction evolved and emerged through a number of qualitative projects. At each stage, concepts were identified, explored, and developed in order to more elucidate the central phenomenon. I will show how, while at times I could identify and avoid the pink elephant, at other times there were one or a herd lurking in the shadows or rampaging through my work. I think that discussing both the successes and pit falls is one way to acknowledge and address the fact that, although we accept the evolution in ideas and thought processes in qualitative research, we still may not be comfortable in articulating the far more complex and insidious threats to inductive processes.
Some schools of qualitative inquiry consider analysis of the literature a hindrance—in fact an invalidity—before commencing fieldwork. To the contrary, when a researcher is studying a concept rather than letting a concept emerge from a setting, it is essential to undertake a thorough theoretical and conceptual analysis of the literature (Morse, 2000; Morse et al, 1996). In my own program of research, the concept analysis was a study in, and of, itself, with the purpose of examining the maturity of concepts, and the explicit and implicit theoretical and research models. The literature constituted data that could be analyzed and formed the basis for a reconceptualization of the original concept by contrasting it with the theory derived from the fieldwork studies.
My area of interest is interpersonal communication in nurse-patient interaction. Specifically, I am interested in understanding how nurses and patients with uniquely different paradigms of understanding illness experience can, within very short spaces of times, make profound interpersonal connections, perceive and avoid unnecessary interpersonal conflict, and, at times, address issues of significant personal vulnerability.
This dimension of nurse patient interaction was conceptualized by Christensen (1990) as a paradoxical determinant of the context of nursing partnerships. She called it anonymous intimacy, or the significant degree of immediate socially sanctioned closeness between strangers. She described this as the ability for nurses and patients, who are strangers, to forge a high degree of intimacy as the patient surrenders privacy for nursing care. The essence of anonymous intimacy is that patients identify with nursing and nurses rather than individuals.
Although the concept was described well in her data set of 87 nurses and 21 patients in a hospital setting, it was not developed theoretically—the meaning, definitions, assumptions of intimacy and anonymity, and the means through which anonymity and intimacy were combined or resolved in interaction, were unclear. Nevertheless, it was a very interesting concept and one that immediately attracted a sense of recognition from nurses. In my conceptual exploration of anonymous intimacy in the literature, it quickly became clear that although Christensen’s (1990) conceptualization was unique, it in fact represented a way of co-orientating to desired ways of relating to each other. It was a style of interacting, not a contextual feature of interactions. It was about using common social knowledge of nursing and patient roles, along with individual ability and desire to enhance a more personal relationship in order to increase or decrease social distance (Spiers, 1994). Of critical importance was the notion of being able to change the degree of interpersonal space in an interaction according to the flow of events in the encounter.
The ability to manipulate the degree of familiarity is important because many of the activities nurses do on a daily basis creates social and personal discomfort and vulnerability for the patient. We know this well; that is why we have different ways of communicating the same information in different contexts. Each approach recognizes the need for diplomacy, politeness, directness or indirectness. But it is not just the patient's sense of vulnerability at issue–nurses, too, deal with needs for privacy, boundaries, and formality and also can be vulnerable in their interactions. From the standpoint of a theoretical understanding of anonymous intimacy, it became evident that the ways nurses and patients interact has something to do with trying to save face in interaction, to prevent or minimize interpersonal discomfort and embarrassment. Thus, in order to comprehend the essence of anonymous intimacy, the concept of face became important.
The idea of face as personal vulnerability in interaction is an interesting one. Saving face is a well recognized phenomena in many cultural groups, where face refers to preserving or losing one's social standing by deferring to social norms of behavior. As an ethno-linguistic concept, it has been developed by Brown and Levinson (1987) in a model of the work involved in social interaction to protect and address threats to face, or the threats to individual’s sense of public image in social interaction. Interestingly, face is defined more by its loss and threats, than, in fact, what it is.
Face would seem to be a highly pertinent concept for nursing interactions. Think of the number of situations that threaten not only patients face, but that have implications for our own as well—patients becoming embarrassed at the loss of bodily or emotional control, the difficulty of conveying distressing news, how one approaches a procedure never attempted before. Yet, the concept is absent in the nursing literature. The concepts that come closest—such as quality of relatedness, trust, co-creation of meaning—work at a level of abstraction often developed from nurses and patients reflection of their interpersonal relationships (Spiers, 1998). This means that the behaviors, the social actions involved in enacting this dimension of interaction, were still largely obscured because they do not occur at a conscious level of behavior (Byrd, 1995).
Face is related to our sense of personal vulnerability, to our sense of social image or presentation in social interaction (Brown & Levinson, 1987). Moving away from a macro view of anonymous intimacy as a way of relating that minimizes embarrassment and discomfort in interaction by creating a sense of anonymity, it became evident that communicative processes of saving face were the key to understanding nursing interactions. However, it is problematic to just borrow a concept from another discipline without thoroughly investigating it. Thus, my attention turned to the concept of face. I needed to explore the philosophical, theoretical and methodological assumptions underlying the concept of face in order to move forward in my investigations. A paper entitled “The use of face work and politeness theory” was published in Qualitative Health Research in 1998.
I had moved from anonymous intimacy to the concept of face in order to draw closer to comprehending that elusive dimension of nurse-patient interaction. Face represents personal vulnerability in interaction. The work involved in saving face—face work—referred to the continual process of identifying, constructing, and enhancing one’s own and the other person’s sense of face, and avoiding or mitigating situations that threatened face (Holtgraves & Yang, 1990). Face, an interpersonal social phenomenon rather than an intrapersonal psychological construct, is mutually constructed in the interaction and is something that is strategically manipulated in response to the flow of events in the encounter (Holtgraves, 1992). In other words, the context of the encounter, and the events within that encounter, change the nature of the face one wishes to claim for oneself and that one is willing to recognize for the other person. That is why talking about a highly intimate and private topic feels different—and is handled differently—when talking with a best friend, an employer, or a health professional. It seemed to revolve around the idea of vulnerability. To understand that part of the nursing experience Christensen (1990) called anonymous intimacy, it was necessary to establish an interpersonal context of face. In order to understand the nature of face, the concept of vulnerability emerged in my theoretical analysis. So far, there was little sign of the pink elephant.
There are literally thousands of references to vulnerability in the literature as it is a fundamental aspect of the experience of health and illness. Yet, when I started to explore how the concept was used in the clinical and research literature, it was evident that there were two primary approaches, neither of which were of much use in looking at the experience of nursing, or the behaviors related to influencing social distance in interpersonal interaction (see Spiers, 2000). Vulnerability can be used to identify individual and group at risk of harm (Aday, 1993). This is based on epidemiological characteristics that assign people or groups to higher than normal standards of risk. This risk is objectively derived, most frequently by some source external to the person being assessed. Thus, vulnerability is located intra-personally as a personal attribute of some kind of deficiency in comparison to the normative standard that requires intervention in order to protect the subject from harm or endangerment (Ferguson, 1978). Alternatively, vulnerability can be a more experiential and qualitative phenomenon, a sense of challenge to one's sense of personal integrity (Morse, 1997; Stevens, Hall & Meleis, 1992). Being able to distinguish between emic and etic views allows us to differentiate between being at risk and feeling vulnerable (Spiers, 2000). However, the problem that the concept of face in interaction posed remained—the idea of mutual vulnerability as a social construct, rather than an intra-personal state in interaction. The definitions used in the literature were still intra-personal, whether the view was emically or etically derived.
This, then, was the theoretical background for my research on the nature of vulnerability in the interactions between home care nurses and their patient's. It seemed that none of the frameworks I had explored—from anonymous intimacy, to quality of relatedness, to face—were adequate conceptualizations of that elusive dimension of nursing: the ability to move a sense of intimacy and distance in order to deal with the interpersonal implications of the event in-the-moment. The concept of face, while useful and interesting, was defined more by what it was not, and the categories within the model were largely fixed. If I had used this, I would have run the risk of approaching my interpretation deductively, with a priori definitions, and with categories of behavior that were largely decontextualised because face, in Brown and Levinson's (1987) approach, specifically addressed only one distinct dimension of social interaction. The various conceptualizations of vulnerability in the nursing and health literature were likewise problematic, limiting my ability to combine both etic and emic views.
It is important to emphasize that in going to the literature, I was not developing a conceptual framework but trying to clarify assumptions and perspectives to put together the beginning of the skeleton that would give my study shape and direction. To make the fieldwork viable, I needed to have some clarity and a theoretical understanding of the kinds of concepts and phenomena at work in constructing the topic that piqued my interest. Creating this skeleton through systematic concept analysis processes allowed me to articulate my assumptions and perspectives. This would provide direction in sampling and data collection. It had started to build the internal structure for my study. Sometimes, these assumptions were more questions than beliefs—could vulnerability be an interpersonal phenomenon? As a mutual experience related to the events in the interaction, could it be observable in the behaviors of the nurse and patient. These are the ideas that sparked the phase of inductive clinical fieldwork. If I had not done this, but had just leapt into fieldwork, I would have been at extreme risk of floundering–of seeing everything as related to my phenomenon of interest–which, at the beginning, was extremely poorly delineated. Without this theoretical work, not only would the pink elephant have entered the picture, it would have picked me up, set me on its back and we would have merrily ridden away.
As is common in qualitative work, researchers seek the context in which we can best see the phenomenon of interest. I was looking for nursing situations in which the nature and characteristics of vulnerability would be highly apparent. To do this, I videotaped home care nurses visits to patients. The unit of analysis was the speech or communication act within the interaction, captured in 31 videotaped visit providing more than 19 hours of video data. Now, it is important to remember that I was not seeking representativeness and generalizability, but an indepth understanding of common social experiences in home care nursing situations. This is where the issue of pink elephants, or issues of inductive/deductive traps, truly began to raise its head–or trunk.
Morse and Mitcham, in the first section of this article, talked about the importance of scoping and focusing, to find the balance between entering the research with such a wide view that the researcher is left to fumble in the dark and walking in knowing what to look for and where to find it. Issues of bracketing, as they noted, were difficult to resolve, especially when one has invested so much time and energy in theoretical concept exploration. I had tried to avoid the pink elephant through my evolving concept explorations and analyses. The problem was that now I was trying to explore vulnerability, and I had an idea of what it could look like. It was clear that nurses and patients experienced episodes of difficulty in their interactions–both very minor and quite major difficulties. Yet, I could not make sense of my data. Despite an excruciating level of description of my data, and extensive challenges from my colleagues, it did not make sense; the idea of vulnerability simply did not match what I thought I saw in the data. It seemed that everything could be related to vulnerability.
As I continued to look at different interactions and nurse-patient dyads, it became clear that until I could understand what it was the nurse and patient were trying to achieve, the notion of vulnerability was meaningless. As my study evolved, the research questions became not what is vulnerability, but how the patients and nurses paradigms of understanding or worldviews were co-constructed through their interaction. I had to explore the kinds of goals in terms of co-created meaning that both nurse and patient were working toward in order to understand the interpersonal conditions in which vulnerability could be manifested (Spiers, in press).
To return to Morse and Mitcham’s idea of a conceptual skeleton, it turned out that I had the bones the wrong way up. It was only through attention to preserving and ensuring principles of inductive reasoning that I came to realize this problem. It was only by suspending ideas from face work theory and models of vulnerability that I could see this, and then more productively use the concepts of face and emic-etic vulnerability later in my inquiry to explore the communicative means by which the interpersonal contexts of mutual interpersonal vulnerability were created and resolved. The next sections are some very concrete and pragmatic examples of the pink elephant threats in this phase of my research.
A necessary design feature in my research, dictated by the need to understand the vulnerability as part of co-creation of meaning, meant that my sampling and data collection were extremely broad. Remember that my unit of analysis was not the nurse-patient dyad, but the speech act—the smallest unit of meaning, verbal or nonverbal, which could be indicative of successful or unsuccessful co-creation of meaning, and thus vulnerability. In each interaction, there could be anywhere from 500 to 2500 speech turns. This was a huge amount of data. However, this breadth was necessary to describe the context of what I was interested in–co-creation of meaning across the nursing and patient paradigms of understanding, and situations in which this did not occur and which was evident in only some data.
I needed to look at multiple levels of the interaction and from different perspectives. For example, I needed to move between very macro perspectives of identifying the activities and tasks they engaged in, from wound care, to pain management, to coordination of services, to types of interaction, from very rote and apparently superficial, to highly attentive interactions, to ones in which each person juggled the degree of involvement. All of this was layered with the immediate and longer terms goals of interaction that were part of every action. By doing this, I could work out what nurse and patient were trying to do, and the nature of the interaction and consequences when this was not successful, or when one person’s attempts or goals were not recognized or matched by the other. At this point, my skeletal framework enabled me to more successfully sensitize me to instances of vulnerability as both a process and outcome, or, even more likely, instances of near misses.
The near miss instances were very important to avoiding the pink elephant. Essentially, vulnerability emerged as a result of the nurse and patient's inability to co-create common meaning and understanding of the situation or the intentions of the other. Vulnerability was more often a potential manifestation rather than an actual one. Why? Because of the communicative skill of nurses and patient's in averting problems in the interaction that could result in overt vulnerability and, often, communication breakdown. The following example illustrates this.
A major type of work in the interactions was creating and sustaining an amicable working relationship. This involved negotiating the level of formality as nurse and patient, and familiarity and liking, as individuals. It was deciding how, and to what extent to get to know each other. Both nurses and patients volunteered information about themselves, and showed interest in finding out about the other. This could range from finding an acceptable level of social talk to inviting or offering self disclosure. For example, one patient deflected all personal probes from the nurse but would happily engage in detailed conversation about their mutual tastes and habits in their community of shops and restaurants. Mutual vulnerability occurred when someone was trying to establish personal boundaries without appearing rude, dismissive, or offended. If a question was declined in a way that was respectful, it identified the boundaries of the relationship. If it was not performed tactfully, then it had the effect of rejecting the other person. In one dyad, the patient was always interested in flattering her nurse and validating the importance of the exclusivity of their relationship. This created difficulties, because sometimes the nurse could not visit, and a substitute was sent. In trying to offer a compliment to the usual nurse, this could be construed as criticism of the other nurse, creating a situation of mutual vulnerability.
N: Well, I was away last week.
P: I missed you too!
N: I know <both laugh>
P: Don't get me wrong!
N: No!
P: Nothing wrong with the other nurse
N: Yeah.
P: She’s nice-
N: I know!
N: -She’s just not as friendly. She doesn’t laugh like you or I do.
N: Yeah. Yeah, yeah, she’s more- she’s different.
P: She’s mostly (XX) on her work and THAT’S IT. Nothing else
N: Yeah…, you know. (yeah) <both laugh>
One of the most apparent deductive-inductive threats in this research was working out the extent to which vulnerability represented an idiosyncratic characteristic or a phenomenon related to the interpersonal context. In dealing with this, my main strategy was pursuing comparative cases. In order to ascertain the extent to which the vulnerability was related to the flow of events in the encounter, rather than to the people, I needed to change the context to see if the nature and characteristics of the vulnerabilities I had identified held across different nurse-patient dyads. It was very interesting to explore how, for example, the kind of vulnerability that was demonstrated in a very well established dyad had significant commonalties and dimensions, as well as differences, with dyads that were the opposite—a new dyad, a first encounter. I was fortunate in being able to observe different nurses with the same patient, which was another way of determining the extent to which the kinds of vulnerabilities or interactive events were intra-personally situated or, as I was discovering, idiosyncratically influenced, but located as mutual interpersonal concern.
Another very interesting conundrum I faced was my ability to attribute meaning to particular behaviors. I noted earlier that I was working on the premise that vulnerability was observable, indicated in not only the content of the interaction but also in the flow of the interaction. In other words, vulnerability or otherwise was not only evident in what is said, but how it was said. This placed me in an interesting position when, from my observers stance, my interpretation of what happened, and how, differed markedly from the patient, the nurse, or both. I had tried to minimize this problem by incorporating interviews in my data collection. I would talk to the patient and nurse each after each visit, asking them about what happened, what they each were trying to do, why and how, and their perceptions of the other person. And of course, I asked if there had been any difficulties. As I expected, they were very rarely able to give me the kind of information that would help me in my interpretation. Why? For two primary reasons, the first being that the kinds of interactive behaviors I was interested in were simply not at a conscious level of awareness or recall. We are so accustomed to dealing with the hiccups and transient communication difficulties that can occur that we simply do not notice them. So, while major communication problems, such as becoming angry or making accusations, were available to my participants for conscious recall, much of what I ultimately found interesting did not exist at a conscious level.
Second, there are interpersonal implications for vulnerability in the researcher-participant relationship and interaction, not just the interaction I was trying to examine. It is bad enough having a researcher observing a nurse’s or patient’s faults or stupid acts, misunderstanding, being inappropriate or coercive during the actual interaction, let alone having to talk about it afterwards. In this example, the patient’s challenge to the nurse’s claim to be able to anticipate the physician’s actions, her subsequent attempts to explain her assumptions, and the patient's realization that he had embarrassed her by disputing her right–and competence—to do this created acute embarrassment for both:
<Nurse is engaged in changing the dressing of the Patient’s abdominal wound>
N: I wonder if he’ll take the rest of those staples out. I know. Are they kind of pulling- can you feel them?
P: No, no, that’s one thing I don't do- tell him how to do his job. No-no-no-no-no <fast sing song voice>
N: No- No! I’m just curious!
P: No, I’m just- I’m not even curious!
N: I know- whenever he’s ready to take them out, that’s (yeah) ok by you!
P: Yes, that’s find. I tell you it doesn’t bother me one way or the other.
N: yeah, yeah. Sometimes people- they irritate, you know, they kind of pull and–but he’ll take- he might take- I wouldn’t be surprised-
P: Oh, whatever. That’s what he gets paid big bucks for.
N: Yeah, that’s right.
Reactions to this kind of interaction would be discounting, denying, laughing it off. Ideally, it could have been useful to take the actual video back to the participants, although I do not think it would have overcome the difficulties inherent in seeing one's own behavior as an observer (Lomax & Casey, 1998). Thus, there was always the risk of misjudging the intent and meaning of the actions. I addressed this in my analysis by being extremely detailed in my description of behaviors and then in my writing, by a textual rendering that tried to draw the reader into the participants world I was observing.
I want to sum up by reiterating that the process I engaged in to pursue a clinical phenomena of interest was an exciting voyage of discovery that has demanded flexibility and willingness to pursue a number of productive and less productive routes. All of the processes and stages of these projects were directed toward developing a skeletal framework to guide and refine my research, to provide purposeful seeking and sensitivity to know what is relevant to build up flesh around the skeleton. The result, to date, of vulnerability as an interpersonal phenomenon in nursing relationships is still excessively broad and there are many areas where the boundaries and attributes are less clear than is desirable. The value of this research focusing on vulnerability within the home care nursing context, however, is that it is generating far more specific directions for even more focused research that, cumulatively, will develop the idea of mutual vulnerability in nursing interactions further.