Beyond description to interpretation and theory in qualitative social work research

Reflexivity statements

Being reflexive enhances the quality of descriptions, interpretations, and theorizing. When researchers are reflexive, they reflect upon what their topic means to them and speculate about what the topic means to informants and others who have a stake in the research. They consider the socio-cultural influences on the topic of interest, the origins and consequences of the issues they are considering, and what the topic may mean to the various audiences they address. Furthermore, being reflexive in an on-going way supports an ethical stance, including raising awareness of power issues such as the relationships between researchers and informants, between members of research teams, in field relations, and in dissemination and uptake of research findings. Being reflexive also increases clarity about what accounts and descriptions mean to researchers and to multiple others (Gilgun, 2006, 2014a). Once the research is underway, reflexivity statements can become part of fieldnotes and group analysis of data.

I recommend that researchers engage in reflexivity from the beginning of their inquiries until the end, from initial brainstorming to write-ups and revisions. Researchers are reflexive in three main ways: conversations they have with others about the research, writing out reflexivity statements, and reflecting upon the meanings of the research that happen spontaneously, often while going about usual daily activities such as during leisure time.

Definitions

In this paper, I define descriptions as excerpts from interviews, observations, and/or documents, including videos and photographs that researchers choose to present in research reports as well as the summaries that researchers provide of the accounts. In research reports, descriptive material has a minimal amount of researcher interpretation. There is always some because researchers’ choices of what to present and what to leave out are based on interpretations. Descriptions can be thought of as a first level of analysis, as close to the actual accounts as possible.

Interpretations are researchers’ commentaries on descriptions and are a second level of analysis. These commentaries are rooted in and gain their credibility through their connections to descriptive material. Researchers may draw on a variety of sources for their commentaries, typically personal and professional experience and their general store of knowledge. Commentaries often contain second-order concepts that organize related processes and other material contained in the data. These second-order concepts are often called themes and categories. They are called second order because they are extractions from first-order concepts that compose the descriptions and thus are one step removed from descriptions. In published articles, themes, categories, and interpretations typically are presented before descriptive material, and the descriptive material illustrates and supports the conceptual material.

Theorizing is a third level of analysis and builds upon the first two. In the construction of theory, researchers engage in a two-step process: 1) the development of working hypotheses that they believe represent what they see in the accounts, and 2) situating this conceptual material into related research and theory. Working hypotheses are statements of relationships among concepts where researchers have not yet situated concepts and hypotheses within their scholarly traditions. Working hypotheses can also be called emerging theory. Theory is a statement of relationships among concepts that are situated within scholarly traditions and is not the same as emerging theory.

Theory-based qualitative research is inquiry that begins with conceptual material such as sensitizing concepts and working hypotheses or theory. Sensitizing concepts are ideas with which researchers begin their inquiries and that alert researchers to what might be important in the topics of interest (Blumer, 1969). Gender, for example, is a sensitizing concept for research on family violence. In previous writings, I called theory-based qualitative research deductive qualitative analysis (Gilgun, 2014b).

Group analysis

Group analysis of data goes back to the origins of qualitative research (Olesen et al., 1994; Webb and Webb, 1932), and for good reasons. Each of us has a limited point of view. If we do analysis with others, we increase the likelihood that we will identify multiple dimensions of informants’ accounts and have checks and balances on our individual perceptions. As is well known in qualitative research, our own ideas and worldviews influence what we notice and do not notice. Working with even one more person adds another worldview and set of ideas. What one researcher may notice, another may not. Through discussion, two or more researchers can broaden and deepen each others’ perspectives and add depth and breadth to descriptions.

What we decide to include in research reports is based on our understandings. If researchers have limited perspectives, then we are at risk to leave out what could be important dimensions of descriptions of informants’ experiences. Any interpretations and theorizing will not be as comprehensive as they would have been through group analysis.

For a variety of reasons, researchers may find that they have no one else with whom to analyze data. In those circumstances, researchers can present their emerging analysis in classes they teach, in workshops and presentations, and in papers they give at conferences before they seek publication. I do all of this, and I also may talk informally to knowledgeable people, such as other researchers and professionals. I typically share some of my interpretations with informants, such as saying, “This is how I understand what you are saying. Did I get what you said?” Informants respond well to this, possibly because they rarely find people interested in understanding them. It also satisfies their curiosity about how we researchers view their stories and shows that we are concerned that we understand what they told us. We are showing respect.

An example of a descriptive finding

Through reflection and analysis, researchers develop their understandings of informants’ accounts. An example is the following from Kranke, Jackson, Taylor, Landguth, & Floersch (2015) who wrote about the experiences of adolescents with mental illness who had a positive outlook and a sense of empowerment. As is usual practice in qualitative research, they first present their interpretation and then they illustrate the interpretation with descriptive material.

School success, strong, functional interpersonal relationships, and a sense of belonging all contribute to the development of a positive outlook and a sense of self-efficacy. To illustrate, one adolescent stated, ‘I’ve made honor roll, moving from making straight F’s to straight A’s’ (p. 110).

Fieldnotes

Fieldnotes also contribute to quality of descriptions, and they help researchers go beyond description to interpretations and theorizing. As Geertz (2000) said, “first you write and then you figure out what you are writing about” (location 54 on Kindle). Writing can clarify our thinking and extend our understandings. We often experience creative thinking through writing, where meanings come together seemingly on their own, and we find words to represent these clusters of meanings.

Ideally, researchers keep fieldnotes throughout the course of data collection and analysis. Fieldnotes are a structured way of reflecting on practice. They typically have four parts: descriptions, observer comments, memos, and diagrams (Bogdan and Biklen, 2007).

Descriptions in fieldnotes are free flowing accounts of what happened during data collection. Descriptions include non-verbals as they occur during interviews and observations. The non-verbals add dimensions to understandings of what informants say during interviews and observations. The following is an example of descriptions in fieldnotes that I wrote following an interview with a sister of a man who had a life sentence for rape and murder. The names are pseudonyms.

I had hoped to be able to see her eyes better, and I could, some. When she looked directly at me, which she did periodically, I clearly could see her eyes, which were clear and intelligent and not at all angry or belligerent. They were kind.

She is somewhat heavyset like Robert. She has clearer eyes, or a clearer face than Robert. Despite her discomfort, she had an open face, mobile, alive. She did not have that mask-like expression which Robert often has, almost like a dead person or the face of a skull.

Memos are the third section of fieldnotes and typically occur at the end of a single entry. They are analytic in nature and build upon descriptions and observer comments. Here researchers compare the various accounts, looking for patterns and exceptions to patterns. This is the beginning of thick description (Geertz, 2000), where researchers begin to develop multiple points of view and multiple dimensions of concepts.

In memos, researchers also speculate about what the accounts might mean in terms of their own general store of knowledge as well as formal research and theory. They consider questions regarding whether the findings that they develop are consistent with, add to, or contradict what is already known. They consider what the findings may mean to multiple others. They may make plans to read related research and theory.

Diagrams of relationships among concepts are common in memos. These diagrams may show connections between the emerging concepts, and they may describe processes, such as how one thing leads to another. I could, for example, show in a diagram how the commission of violent acts (a concept) is something good (concept) for perpetrators. I could also diagram the following working hypothesis that describes a process: Some perpetrators are in dysregulated states (concept) and seek to reregulate (concept) through acts of violence (concept).

Diagrams are important in observational studies. The physical setting, movements in settings, and who interacts whom, characteristics of the interactions, and who is close to or distant from others can tell a great deal about relationships between people researches observe.

In practice, fieldnotes for a given interview or observation may not have all four parts. For example, in an observation of a case consult in a family service agency, I described the interactions and wrote observer comments, I did not write memos. Here is an excerpt from this fieldnotes of this observation. All names are pseudonyms.

Clinical supervisor: ‘Does anyone have any suggestions? I’m at a loss about how to help David.’ [David is a case manager who asked for guidance about what to do regarding a mother who slept in the same bed with her teenage son.]

O.C.

Around this time I spoke up and asked if this case is a logical outcome of lack of action on the part of CPS [child protective services]. … Donna [clinical supervisor] said that this case ‘hovered on the periphery of CPS’ but never went beyond intake because mom perpetrated emotional abuse and this does not fall under traditional CPS. This is under the surface. That this mother slept with Riley [her teenage son] in the basement for years did not rise to level of a CPS case.

The following is an example of a memo I wrote where I reflected on my decades-long research on the meanings of violence to perpetrators. I described my realization that a core concept that has the potential to organize my understanding the idea that violence is good from the points of view of perpetrators (Gilgun, 2014c).

When it came to the meanings of violence to perpetrators, I was a slow learner. Apparently, I did not have cognitive schemas that included the idea that violence can be experienced as good. Too many other discourses dominated. … those that helped me interpret my data as violence as a terrible thing. Look at its effects: lifelong suffering for child survivors, stigma and trauma for survivors of rape, shame for survivors of domestic abuse, fear, and trauma for all who care about survivors of violence. Violence to perpetrators meant power, dominance, control, hegemony, entitlement, male privilege, effects of childhood maltreatment, and ideologies that support violence. These are received ideas that appear to influence what we notice and that may prevent us from noticing novel dimensions of accounts of human experiences.

Memos, observer comments, group analysis of data, sharing descriptions and interpretations with informants and others in a variety of outlets, and reflexivity.

Creative thinking

The procedures discussed so far help ground researchers in the accounts of informants. Following these procedures sets the stage for creative thinking. Many times I have followed these procedures, and, at a single, almost magical moment, core concepts simply are there in my mind. An early example in my own research practice was work I did on the sexual identity development of men sexually abused as children. Three pathways simply stood out. One was heterosexual development before and after the sexual abuse. The second was heterosexual development before the abuse and then sexual identity confusion after the abuse. The third was gay sexual identity development before the abuse and after the abuse. My co-researcher and I documented the three pathways (Gilgun & Reiser, 1990).

Sometimes the identification of core concepts can take longer. The idea of violence as good from the points of view of perpetrators took almost 30 years to occur to me. As I noted in the previous example of a memo, I think the prevailing discourses about violence blocked me from seeing what was in the data all of time (Gilgun, 2014c). I am now in the process of documenting my theory of violence as good from the points of view of perpetrators. Eventually, I hope to have a well-documented descriptions of violence as good from the points of view of perpetrators that will allow me to develop interpretation and theories and to contribute to policy and practice.

Grounded theory

Theorizing while initially setting aside conceptual material is associated with grounded theory. An example is the study of Slatyer et al. (2015) who sought to develop a theory “explaining the experience of caring for patients with severe pain from the perspective of nurses working in hospital wards” (p. 230). The researchers did a preliminary review of the literature in order to develop their focus and approach. They then set aside what they had learned and did interviews and observation. During data collection and analysis, they

generated categories describing the meaning of pain; the nature of nurses’ care; their emotional responses; challenges encountered in the hospital practice environment; and nurses’ consequent actions and interactions (p. 232).

Theory-guided qualitative research

In research that begins with theory, researchers do the same kind of analysis as is done in studies that do not use initial conceptual material once the research begins. In theory-guided qualitative research, however, researchers use conceptual material from the outset. This material can be composed of sensitizing concepts, theory as defined earlier, and working hypotheses developed from a variety of sources including personal and professional experience and prior research. In theory-guided research, researchers are explicit about the concepts, hypotheses, and theories that shape the design of the research, including the interview questions. Researchers code the material using the concepts with which they began, which can be called sensitizing concepts as described earlier. During analysis, the names of concepts change to that of codes. Over the course of analysis, researchers may add codes based on evidence, and they may discard some or all of the codes with which they began. They also may modify or discard initial hypotheses and theory. The procedures of negative case analysis support such developments.

Negative case analysis is traditional in theory-guided qualitative research (Gilgun, 2014b). Negative case analysis involves the active search for evidence that adds to, contradicts, and undermines the emerging analysis. In this way, researchers seek to account for patterns and exceptions to patterns and to improve upon the initial conceptual material by expanding its scope, depth, and dimensions. Because the research begins with conceptual material, the results are embedded in conceptual material and are therefore theoretical in nature. Example of research I did that was theory-guided are Gilgun (1995) where I used the concepts of justice and care in an analysis of the accounts of incest perpetrators and Gilgun (1999) where I used codes that another analyst developed to analyze a case of family murder.

A recent example of theory-guided research is the work of Kranke et al. (2015) who reported on research on adolescents with mental illnesses who did not self-stigmatize, discussed earlier. They began their analysis with theories of stigma, self-empowerment, and their previously constructed model of adolescent self-stigma (Kranke et al., 2011, as cited in Kranke et al., 2015). Through their analysis, they identified four themes and “an empowerment model that has three components: ‘diminish,’ ‘normalize,’ and ‘attribute’” (p. 110). As they said, their findings are preliminary. They, therefore, did not situate their findings in scholarly contexts to develop theory.

An example of research that conceptual material throughout the study and write up is in my research report on incest perpetrators using the initial concepts of justice and care (Gilgun, 1995). I first reviewed research and theory on care and then developed a hypothesis from this material. The hypothesis with which I began the study is as follows.

Incest perpetrators have special regard for themselves and do not have regard for the impact of incest on their victims (p. 270).

After I did the analysis, I revised the hypothesis. They are

(a) Perpetrators have special regard for the deep pleasure they find in incest. (b) Concepts of romantic love and mutuality are prominent in many but not all incestuous relationships. (c) Many interpret their behaviors as promoting children's welfare and not hurting them. (d) The sense of love and caring that many perpetrators express for the children are contradicted by behaviors which are sometimes unresponsive and cruel (p. 270).