Professional Documents
Culture Documents
research-article2017
QHRXXX10.1177/1049732317750127Qualitative Health ResearchMorgan
Commentary
Qualitative Health Research
David L. Morgan1
Abstract
Although theory building is often described as the ultimate goal of qualitative research, an examination of articles in
Qualitative Health Research (QHR) shows that themes are actually the typical format for reporting results. In addition,
articles that rely on themes often present low-level theories in the form of models that connect these themes. Because
models have received less attention than either themes or theories, this article concentrates on summarizing four
different kinds of models: hierarchies, timelines, processes, and cycles. In each of these cases, it presents both a general
illustration of such a model and a realistic example from a published article in QHR. It concludes with a call for greater
recognition of the role that models play in capturing the results of qualitative research.
Keywords
reflexivity; theory development; methodology; qualitative analysis
Defining Themes, Theories, and the use of themes in the nursing literature and produce the
Models following composite definition: “A theme is an abstract
Themes entity that brings meaning and identity to a recurrent
experience and its variant manifestations. As such, a
Themes or the equivalent (e.g., “core concepts”) made up
the basic building blocks for reporting results in well over 1
Portland State University, Portland, Oregon, USA
half of the articles that I examined in QHR. Although the
Corresponding Author:
role of themes in presenting qualitative research has not David L. Morgan, Department of Sociology, Portland State University,
received prominent attention, it has not been ignored, USA.
either. In particular, DeSantis and Ugarriza (2000) review Email: morgand@pdx.edu
340 Qualitative Health Research 28(3)
My own definition is as follows: Themes are meaning- Figure 2. An illustration of a model based on hierarchies.
ful patterns in the data, which researchers use to interpret
that data for an audience. This definition has two parts:
first, the same reliance on patterns that is common ele- could be built into theories. Even so, there was little
ment in other definitions, and second, an emphasis on full-scale theorizing in these articles. This was obvious
communicating with audiences. More specifically, from the consistent failure to describe why a given set
authors construct themes to present the core patterns they of themes was the most useful or powerful way of
use to interpret the data. My definitions, thus, treat themes interpreting the data. Instead, the “Results” sections of
as tools that authors use to help their readers understand these articles almost always began by announcing that
the most important aspects of the data. there was some number of themes (typically three–
Of course, no one definition can capture the wide range of five) in the data, and then worked through this list of
uses that authors have made of themes. Sandelowski and themes, one by one.
Barroso (2003) provide a useful distinction between present- Of course, creating a high-level, abstract theory is an
ing qualitative research through thematic surveys versus con- almost impossibly large goal for an article-length presen-
ceptual/thematic interpretations, in which the former offer tation, but this is not to say that these articles did not do
interpretive summaries that are closer to the data whereas the any theorizing. Instead, they worked with connections
later move toward a more integrated reframing of the data. between themes to create what I will call models.
This points to a tension between the extent to which themes
describe the data or to interpret it. Thorne (2016) has Models
addressed this issue through an approach that she describes
as “interpretive description,” where mere description of the I define models as systematic depictions of the relation-
data is not enough, but abstract interpretation can lead to ships among a set of concepts. As such, they amount to
results that are too removed from the data. As my own defini- low-level theories, with a minimum level of abstraction.
tion indicates, I prefer an approach based on interpretation, For qualitative research, these models simply accept a list
but I also would argue against an overly abstract approach to of themes as the elements that go into a model, without
creating and using themes. Instead, I would reserve abstrac- attempting to explain why those themes are the basis for
tion for the level of theory rather than themes. the model. Models, thus, stand between themes and more
abstract theories.
My rough estimate is that about a fourth of the QHR
Theories articles that used themes made the next step of connecting
One thing that is clear across the range of uses for themes those themes into an explicit model. At present, these
is that they fall short of theories. I will follow a classic models are most likely to appear at the beginning of the
definition, where a theory specifies the links between a “Discussion” section, but one can make a strong case that
set of key concepts. Collections of connected concepts they should be treated as part of the “Results” section,
amount to low-level theories. More powerful theories go because they represent the end point of the analysis and
two steps further: They explain why those particular con- interpretation of the data.
cepts are the key elements that make up the theory, and Interestingly, nearly all these models involved some
they explain why those concepts are related to each other kind of graphic. In all likelihood, this reflects the fact that
in the ways that they are. complex sets of relations are easier to show than to
In the articles I examined, it is fair to think of the describe with words, so the point of using a graphic was
themes they reported as representing concepts that indeed to replace a thousand words.
Morgan 341
Figure 3. An example of a model based on hierarchies (Germeni & Sarris, 2015).
Making Models are organized. The most common use for this kind of
model is to show a relatively complex combination of
The last phase in my informal content analysis of the themes and subthemes.
articles was to sort through the models they contained to Typically, these models are quite basic—to point that
create a basic typology. This produced four categories of they may not require a graphic when all they do is restate
models: hierarchies, timelines, overtime processes, and a simple pattern that could easily be summarized in the
cycles. This section looks at these options, showing both text. Instead, the goal of diagrams such as Figure 2 should
an idealized version of each and a realistic example be to provide a visualization that makes it easier to follow
drawn from QHR. what would otherwise be an overly complex description.
However, note that organizing the coverage of models In addition, as the real-world example in Figure 3
around these four types should not be interpreted as shows (Germeni & Sarris, 2015), it is possible for links to
restricting more innovative options. For example (again, connect themes either within a single level or across lev-
drawing from QHR), Anderson, Foster, Freeman, els. I would argue, however, that it is best to limit the use
Luetsch, and Scott (2017) use a teeter-totter to show a of such complex paths unless it is important to describe a
question involving balance; Dayal, Weaver, and Domene complicated pattern of flows, such as the centrality of
(2015) use a spiderweb to show a central problem with “being left high and dry” in Figure 3.
multiple additional layers; and Tornblom, Werbart, and
Rydelius (2015) use cogwheels to show a set of interlock-
ing issues. Thus, the four categories of models I cover Timelines
here represent only the most common formats I encoun- As shown in Figure 4, models based on timelines connect
tered, and future efforts may well produce a richer range a series of themes, where one precedes another in a
of possibilities. sequence. In Western cultures, timelines are usually rep-
resented as moving from left to right, so the starting point
is on the far left and the terminating point is on the far
Hierarchies
right. These steps may be treated as “causal” so that ear-
Models based on hierarchies, such as Figure 2, lay out a lier themes produce later ones as consequences, or they
relationship between upper level and lower level themes. may simply capture a flow through time.
As the figure indicates, this category of models almost Various forms of diagrams such as Figure 4 were com-
always uses a top-to-bottom flow to show how the themes mon in QHR, but many of these were merely descriptive
342 Qualitative Health Research 28(3)
Figure 7. An example of a model based on complex processes (Edge et al., 2017).
Germeni, E., & Sarris, M. (2015). Experiences of cancer care- Thorne, S. (2016). Interpretive description: Qualitative
giving in socioeconomically deprived areas of Attica, research for applied practice (2nd ed.). New York:
Greece. Qualitative Health Research, 25, 988–995. Routledge.
doi:10.1177/1049732314554098 Tornblom, A., Werbart, A., & Rydelius, P. (2015). Shame and
Hayles, E., Harvey, D., Plummer, D., & Jones, A. (2015). gender differences in paths to youth Suicide: Parents’ per-
Parents’ experiences of health care for their children with spective. Qualitative Health Research, 15, 1099–1116.
cerebral palsy. Qualitative Health Research, 25, 1139– doi:10.1177/1049732315578402
1154. doi:10.1177/1049732315570122 Vann-Ward, T., Morse, J., & Charmaz, K. (2017). Preserving
Miles, M., & Huberman, A. (1994). Qualitative data analysis self: Theorizing the social and psychological processes
(2nd ed.). Thousand Oaks, CA: Sage. of living with Parkinson Disease. Qualitative Health
Morse, J. (2008). Confusing categories and themes. Research, 17, 964–982. doi:10.1177/1049732317707494
Qualitative Health Research, 18, 727–728. doi:10.1177/
1049732308314930
Sandelowski, M., & Barroso, J. (2003). Classifying the findings Author Biography
in qualitative research. Qualitative Health Research, 13, David L. Morgan is an emeritus professor of sociology whose
905–923. doi:10.1177/1049732303253488 interests include both focus groups and mixed methods research.