Issues encountered in a qualitative secondary analysis of help-seeking in the prodrome to psychosis.
ABSTRACT Primary data are rarely used explicitly as a source of data outside of the original research purpose for which they were collected. As a result, qualitative secondary analysis (QSA) has been described as an "invisible enterprise" for which there is a "notable silence" amongst the qualitative research community. In this paper, we report on the methodological implications of conducting a secondary analysis of qualitative data focusing on parents' narratives of help-seeking activities in the prodrome to psychosis. We review the literature on QSA, highlighting the main characteristics of the approach, and discuss issues and challenges encountered in conducting a secondary analysis. We conclude with some thoughts on the implications for conducting a QSA in children's mental health services and research.
- SourceAvailable from: Fang-Pei Chen[Show abstract] [Hide abstract]
ABSTRACT: Parents and caregivers are confronted with many challenges when caring for adolescents experiencing first-episode psychosis (FEP). Understanding and support for parental help-seeking process is essential for adolescents' timely access to treatment. The study aimed to develop a pathway model of parental help seeking for adolescents experiencing FEP and identify crucial time points for intervention.Early Intervention in Psychiatry 06/2014; · 1.65 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: This paper reports on a qualitative exploration of the reasons for differences in pathways to care and duration of untreated psychosis (DUP) in the African, Caribbean and European (ACE) Pathways to Care study from the perspective of respondents to the study and their families. Ontario, Canada. Thirty-four participants in total. Twenty-five young people who had experienced a first episode of psychosis and nine family members. Participants were part of the ACE Pathways to Care study. We implemented six focus groups. Furthermore, we implemented four in-depth interviews with two African-origin young women, one Caribbean-origin woman, and one European-origin woman with lived experience of psychosis. Factors that influenced help-seeking delays across the three groups were: personal awareness of symptoms, family members' knowledge of psychotic symptoms and knowledge of mental health services. Youth and their family members described how stigma played a key role in pathways to care by stopping them from asking for help. The way in which stigma operated on the three groups' members, from feeling ashamed to feeling guilty for their mental illnesses, helped to explain differences in DUP between the groups. Guilt feelings emerged as a prominent theme among members from the African and Caribbean groups and it was not discussed in the European focus group. Delay in entering into first-episode psychosis programmes was also influenced by the stigma perceived by young people in healthcare settings. This had an impact on the therapeutic relationships, disclosure of symptoms and overall trust in the healthcare system. The findings of this paper suggest that stigma, especially internalised stigma, may operate in different ways in European-origin, African-origin and Caribbean-origin groups. These findings could inform the development of more equitable services for people in early stages of psychosis. Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://group.bmj.com/group/rights-licensing/permissions.BMJ Open 01/2015; 5(1):e006562. · 2.06 Impact Factor
Issues Encountered in a Qualitative
Secondary Analysis of Help-Seeking
in the Prodrome to Psychosis
Brenda M. Gladstone
Tiziana Volpe, MSc
Katherine M. Boydell, MHSc, PhD
Primary data are rarely used explicitly as a source of data outside of the original research
purpose for which they were collected. As a result, qualitative secondary analysis (QSA) has been
described as an Binvisible enterprise^ for which there is a Bnotable silence^ amongst the
qualitative research community. In this paper, we report on the methodological implications of
conducting a secondary analysis of qualitative data focusing on parents_ narratives of help-
seeking activities in the prodrome to psychosis. We review the literature on QSA, highlighting the
main characteristics of the approach, and discuss issues and challenges encountered in
conducting a secondary analysis. We conclude with some thoughts on the implications for
conducting a QSA in children_s mental health services and research.
Qualitative secondary analysis
Within the health and social care literature, secondary analysis is typically considered the
purview of deductive, quantitative research methods.1–4In qualitative research, the text produced
Address correspondence to Brenda M. Gladstone, a PhD candidate, Community Health Systems Resource Group, The
Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, M5G 1X8, Canada. Phone: +1-416-8137262. Fax:
+1-416-8137258. Email: Brenda.Gladstone@sickkids.ca.
Tiziana Volpe, MSc, a PhD candidate, Community Health Systems Resource Group, The Hospital for Sick Children,
555 University Avenue, Toronto, Ontario, M5G 1X8, Canada. Phone: +1-416-8137205. Fax: +1-416-8137258. Email:
Katherine M. Boydell, MHSc, PhD, Community Health Systems Resource Group, The Hospital for Sick Children, 555
University Avenue, Toronto, Ontario, M5G 1X8, Canada. Phone:+1-416-8138469. Fax: +1-416-8137258. Email:
Katherine M. Boydell, MHSc, PhD, Community Health Systems Resource Group, Child Health Evaluative Services,
The Hospital for Sick Children, Toronto, Ontario, Canada.
Katherine M. Boydell, MHSc, PhD, Department of Psychiatry and Public Health Sciences, University of Toronto, 555
University Avenue, Toronto, Ontario, M5G 1X8, Canada.
Journal of Behavioral Health Services & Research, 2007 * 2007 National Council for Community Behavioral
Issues in Qualitative Secondary AnalysisGLADSTONE et al.
by collecting primary qualitative data is rarely used as a source of data outside of the original
research for which it was intended.2Consequently, qualitative secondary analysis (QSA) is
described as an invisible enterprise for which there is a notable silence amongst the qualitative
research community.5,6This is consistent with the lack of discussion on the topic, including little
published evidence regarding the benefits and limitations of QSA, although this is rapidly
changing. Recently, an increased interest in QSA is suggested by a number of new publications in
this area;7speculation as to the reasons for this increase are the growth in discussions related to
archiving data (qualitative and quantitative) and advances in technology which allow for the
promotion of formal data sharing and data retention.
Informal data sharing and the personal use of a researcher_s (or research team_s) data is a more
common occurrence in QSA. This is evidenced by several early innovative studies, many of
which are included in a critical review of QSA in the health and social care literature by
Heaton,1,3,4indicating that much of this work to date is undertaken by the primary researcher or a
member of the original research team involved in collecting primary data. This was the case in
our study. In this paper, we highlight some of the challenges that arose when we used QSA to ask
a new question about help-seeking activities in the prodrome to first-episode psychosis based on
data we had collected for prior purposes. We trust that the explicit identification of issues related
to QSA will encourage further reflexive dialogue about methodological challenges amongst
researchers.22Although our research team has many years of experience using qualitative
methods in children_s mental health research, QSA was new to us, and we wanted to think
through some of the issues we encountered in relation to the newly emerging literature. An
enduring question for us has been to consider to what extent QSA is a distinct research enterprise,
and whether many of the challenges inherent in this approach are particular to QSA or to
qualitative methods more broadly.
Debates about QSA tend to center on whether or not it is possible, advisable, or even feasible to
conduct a secondary analysis of qualitative data.8There is little guidance focused on the process
of conducting QSA,1,9and few traditions or established principles from which to generate
methodological decisions or evaluative criteria.2,10There is speculation that researchers hesitate
to acknowledge their work as a secondary analysis because they are tacitly aware of the
epistemological, methodological, and ethical challenges posed by this work, particularly in the
absence of a solid body of methodological theory to support procedural decisions.9(p. 548)
Consequently, there are questions as to whether an analysis is a secondary analysis, implying
something above and beyond the simple Bre-use^ of existing data. Moreover, primary analysis is
considered a safer approach to research because most careers within social science disciplines are
promoted based on the production of new knowledge arrived at through the collection of new
data,7or as Sandelowski notes, BWe have become inveterate data collectors, having been imbued
with the idea that research means collecting new data^.10(p. 129)
A preliminary problem, particularly for those working with their own data, is to know where a
primary analysis of qualitative data ends and a secondary study might begin. In other words, how
sufficiently new or distinctive must the analysis be to consider it a QSA? It has been suggested
that a greater awareness of QSA will enable researchers to recognize and define their work as
such.4Heaton1and others2,6have offered a number of typologies for using QSA that help to
clarify the purposes of a QSA; in our study, the secondary analysis focused on an examination of
a new empirical question rather than a theoretical or methodological question. We were clear that
our question did not offer a more in-depth investigation of our original purpose, which was to
study motivation in the context of a first-episode of psychosis; rather, it exceeded this focus,
directing attention to help-seeking activities in the data.
We conducted a secondary analysis of qualitative data based on interview transcripts produced
in our earlier research on young people_s experiences of motivation during a first episode of
psychosis.11,12The secondary analysis focused on parents_ narratives, which included descriptions
The Journal of Behavioral Health Services & Research34:4October 2007
of their experiences of obtaining mental health services for their son or daughter*. Our goal in this
study was to better understand the factors involved in the decision to seek help (or not) from the
mental health system. Sociological studies document that clergy, police, lawyers, and relatives
and friends are critical actors in the social process of seeking care.13,14Moreover, findings suggest
that parents themselves do not seek help for the young person experiencing psychosis without
consulting a number of lay and formal resources along the pathway to mental health care
services,15underscoring the importance of social influence on when, how, and whether
individuals receive care. Individuals in social networks are more than an influence on help-
seeking, they are often the caregivers and advisors. This social influence affects the choices made
by individuals and may take the decision out of the individuals_ hands and place it with family
members or others in the community.
We begin the paper with a brief introduction to QSA, followed by a description of our study
design, which includes an overview of the original study. We used an interpretive interactionist
framework for data analysis in both the original and in the second study, and so a discussion of
this framework is presented. Finally, we review the literature on QSA, highlighting the main
characteristics of this approach and discuss challenges we encountered in our study of help-
seeking activities. We conclude with a discussion of issues related to how we think about context
in QSA, as well as those associated with study design, ethics, benefits, and costs of doing QSA.
The QSA Study Design
Secondary analysis is defined as the reuse of existing data, collected for prior purposes, to
investigate new questions or apply a new perspective to an Bold^ question and as a means of
corroborating, validating, or redefining original, primary analysis.1An immediate conundrum in
speaking and writing about QSA is the difficulty posed by language. As a study in its own right, a
secondary analysis uses primary data collected for prior purposes; to avoid confusion, researchers
generally refer to the primary data or the Boriginal^ study as a way of distinguishing between the
previous work and that to be performed in the secondary analysis. For the purposes of our
discussion, we will use primary data and refer to the first or original study to distinguish this from
the QSA study, which we use to examine some of the methodological challenges of this approach.
Interest in the use of QSA for this study was prompted by the richness of parent transcripts from
the original study and our awareness that they contained the Bstory^ of events leading to their
child_s first episode of psychosis, although this was not the focus of our original research question.
The original study
The first study was conducted using primary qualitative data originally collected by all three
authors (BG, TV, KB). The original study used an interactionist framework to focus on the
subjective experience of motivation and schizophrenia.11It focused on understanding the
psychosocial factors influencing motivation while living with schizophrenia or first episode
psychosis for those who were early in the course of their illness.12Participants were asked to
describe how they experienced motivation in their day-to-day lives. Sixty in-depth interviews
were conducted with two groups of individuals: people diagnosed with schizophrenia that were
identified as recovered and young people experiencing a first episode of psychosis. We also
interviewed parents of ill family members and mental health professionals; however, the
participants in this study were not connected to one another.
*Further details regarding the substantive findings related to the QSA are available from the authors upon request.
Issues in Qualitative Secondary Analysis GLADSTONE et al.
This paper focuses on a QSA of a subset of primary data, representing ten in-depth interviews
with parents whose children experienced a first episode of psychosis. The purpose of this QSA
was to ask new questions about parents_ help-seeking experiences based on the primary data
described above. Seven mothers and three fathers were interviewed in the original study; one
interview included both parents. Parents had six sons and three daughters in total, and their ages
ranged from 14 to 22 years at the time of the interview. Parent interviewees were recruited from
clinics serving children with mental health problems in general hospitals, as well as specialty
clinics offering first-episode psychosis services in two separate urban areas in southern Ontario.
Institutional ethics review board approval was obtained following an internal scientific peer
review of the research proposal. It was not feasible to obtain consent for the QSA; therefore, we
decided to anonymize the data to protect the identities of the participants in the original study.
Importantly, the research team felt strongly that no harm would come to participants through this
exploration of parents_ help-seeking stories, although this was not the focus of the study for which
they originally gave their informed consent to participate.
Following the work of Denzin,11an interpretative interactionist framework was used to guide the
secondary analysis. Interpretive interactionism is both a perspective and a method and seeks to
highlight the world of problematic lived experience of individuals. At the heart of interpretative
interactionism lies thick description;16thick interpretation; and deep, authentic understanding.11(p. 33)
Denzin describes this as a method for studying biographical experience through the use of personal
experience and self-stories that focus on key turning points in people_s lives.11(p. 47)A major aim of
of the individual positioned within a social context. The interconnection between private lives and
public responses to personal troubles is addressed.17It is acknowledged that many social care
programs are ineffective because they do not account for the perspectives and attitudes of the
individuals served. Using this approach, researchers can detect strategic points of intervention in
social situations; in our case, this involved identifying barriers to help-seeking to plan appropriate
interventions. As in other qualitative research methodologies, the epistemological posture taken by
interpretive interactionism questions the possibility of value-free inquiry, an objective researcher, and
interpretation based on causality.18,19It assumes that knowledge is socially constructed and that the
concept of truth depends on the perspective one takes in interpretation.20,21This is not to deny the
individual_s experience of first-episode psychosis as something Breal^ but to suggest that these
experiences correspond to the social processes they are embedded within and with the presence of
The following steps describe the analytic procedures we followed in keeping with the
interpretive interactionist framework for the QSA. The first step of the analysis is bracketing,
followed by construction, and then finally the contextualisation of the findings.11
The data generated from the original interviews were analyzed through the process of
bracketing, which involved isolating and examining the essential features of the help-seeking
processes under investigation. The phenomenon of help seeking was Btaken apart, and dissected
so that its elements and essential structures are uncovered, defined, and analysed^.11(p. 55)The
data were bracketed by repeatedly reading through the transcripts to develop a Bstory^ of help
seeking from the parent perspective. In this manner, essential recurring features of the general
pathway narrative were categorized. We found that parents engaged in ignoring, avoiding, hiding,
and explaining away the early symptoms of psychosis in their children. When it became clear that
these symptoms were not going away, parents began to look for answers, and a wide variety of
The Journal of Behavioral Health Services & Research34:4October 2007
professionals were approached as they attempted to make sense of what was going on. In so
doing, the advocacy efforts involved in accessing any form of help were apparent. As one mother
stated, BI knew how to work the system, but we still went through the meat grinder that I think
every family goes through.^
The next step of the analysis was construction, which builds on the bracketing process. It
consisted of arranging the elements of help-seeking into a pattern that reflected what had
happened in the lived experience of parents overall. In essence, the process of construction
Bclassifies, orders, and reassembles the phenomenon back into a coherent whole^.11(p. 58)This
involved listing the bracketed elements of help seeking and locating the same repeated forms of
conduct, experience, and meaning in all of them. These recurring features in the pathway included
multiple turning points—the importance of symptoms, the significance of actors in the pathway,
the role of coercion and persuasion, the function of the health care system, and the ways in which
uncertainty and time played out in the pathway to care. This process led the research team to a
better understanding of the ways in which these various elements interacted in a holistic sense.
For example, for one mother, there was confusion (uncertainty) regarding the onset of
symptoms—Byou don_t know where to begin and you are not looking for mental illness.^ Many
roadblocks were encountered in the search for help and resulted in frustration that professionals
were not identifying the problem (health care system). Only after a great deal of time was her
child committed to hospital by a physician (coercion).
The final step of the analysis was contextualization, a process in which greater meaning is
sought across individual experiences. It starts with the critical themes and structures uncovered
during bracketing and construction. This process resulted in uncovering the motives, feelings, and
intentions of parents as they recounted their experiences of help-seeking for their child. In
essence, according to Denzin,11contextualization provides a more thorough understanding of the
phenomenon because it is described from the participants_ perspectives Bin their terms, in their
language, and in their emotions. It reveals how the phenomenon is experienced by ordinary
people^.11(p. 60)In this study, help seeking was contextualized in terms of turning points (or what
Denzin would call Bmoments of crisis^) in the pathway to care. These turning points were
multiple and many attempts were made to access care over a period of time, until the young
person finally arrived at a specialty first-episode clinic or general hospital. Because of these
multiple, unsuccessful attempts at help seeking, we conceived the notion of missed opportuni-
ties—the key recurring feature in all of the pathways to mental health care. Opportunities were
missed by parents, educators, guidance counselors, social workers, and psychiatrists. Symptoms
were Bexplained away as teenage rebellion^ or Bthought to be drug related^.
Traditional criteria for establishing the quality of qualitative research are not readily duplicated
in QSA, and methodological guidance is only now beginning to be developed.9,23We considered
our study trustworthy based on the availability of the original interview transcripts and the
participation of the researchers who designed and conducted the original study. We had access to
knowledge about the context in which the data were collected, which would help to inform the
QSA. We had information about the time and setting of data collection and issues related to
recruitment. We had a common, although implicit, understanding of many factors that shaped our
analysis and interpretation in the first study.
Another important factor in establishing the quality of the study was that the new research
question not only fit with the Bold^ data but arose directly from the primary data. This helped to
validate the importance of this issue as described by the participants in the original study who were
not asked to describe their help-seeking efforts directly, however it emerged in situ. While we
originally sought to understand parents_ perspectives of their children_s experiences of motivation
Issues in Qualitative Secondary AnalysisGLADSTONE et al.
after a first episode of psychosis, parents were intent on telling stories about trying to find help for
their children and referred to experiences that occurred during a much earlier prodromal period.
Specific qualitative approaches such as grounded theory, ethnography, or phenomenology
generate specific kinds of data, which has lead some researchers to question the compatibility or
fit between the purpose of the secondary study and the use of primary data.2In this case example,
we used an interpretive interactionist conceptual framework in both the primary study and the
QSA. The quality of the study was also established using the analytic methods that are commonly
used in qualitative research practice, verifying the consistency of our results within individual
interviews and comparing and contrasting this analysis across all transcripts.24The three authors
reviewed the transcripts at all levels of analysis to establish agreement regarding the coding of
transcripts and the development of themes and concepts.
Examining Issues Encountered in the QSA
In the following discussion we highlight some of the challenges we encountered in conducting
our QSA of help-seeking activities in the prodrome to first-episode psychosis. These issues are
related to how we think about context, a concept that, for many, describes the essential nature and
contribution of qualitative research. We consider dilemmas we encountered associated with study
design, and consider both the benefits and costs of doing QSA. Ultimately, we weigh in on the
positive side of the debate about whether QSA is possible or advisable and suggest some ways to
think about the process of conducting QSA.
Context is a concept that is central to qualitative research.25How we think about context in
relation to lived experience may include factors ranging from individual interactions in local
settings and a focus on the life of specific participants; however, it may also include broader
social, economic, political, and discursive arenas that have their effect on the phenomena of
interest. Holstein and Gubrium consider context(s) as a departure point for thinking about
qualitative analysis, pluralizing the word to denote the myriad ways that it can have an effect on
the analysis of data and encouraging researchers to work with the complexity inherent in thinking
about what context matters in a particular study by specifying those factors that are drawn under
this rubric.25(p. 308)
Criticism of QSA is often based precisely on this issue because it is assumed that QSA will lack
the kind of in-depth information about context that was part of the original study. However,
Bcontext^ is a contingent, complex concept, which is open to interpretation about how Bfixed^ this
notion is from the perspective of different researchers with varying theoretical and disciplinary
approaches to their work. Clifford Geertz_s Bthick description,^ upon which claims are often made
about the quality of qualitative research, refers to context as a way of thinking about how specific
phenomena are linked to something greater than itself, whether at a local or a broader level of
analysis;16thick description does not require the researcher simply to accumulate data for its own
sake. Mauthner and colleagues have expressed concern about the advisability of QSA based on
the nature of qualitative data they define as B...the product of the reflexive relationship between
the researcher and the researched, constrained and informed by biographical, historical,
theoretical and epistemological contingencies^.8(p. 742)Fielding is dismissive, seeing context
and its relationship to the data as a practical rather than epistemological or theoretical issue that is
present in any data.6He argues, B...an essential part of qualitative research has always involved
monitoring the effects of reflexivity and taking account of these in the analysis...Since no one has
ever argued that fieldwork can adequately provide data as evidence for every potential analytic
theme applicable to the data in primary data analysis, we can challenge the idea that context
The Journal of Behavioral Health Services & Research34:4October 2007
effects make secondary data analysis an epistemologically distinct activity^ (emphasis in the
original, p 100).6
Thinking about context has drawn our attention to assumptions in our analysis. While we agree
that qualitative research is always informed and shaped by biographical, historical, and theoretical
contingencies, taking account of their effect in a particular analysis is an important aspect of all
qualitative research. For example, considering the social context of our QSA has led us to think
about historical and biographical changes during the 6 years since the data was collected for the
original study. The field of early intervention in psychosis has changed considerably, and health
care funding for interventions is now largely concentrated in specialty clinics and first-episode
programs worldwide. At the time of the first study our participants were recruited largely from
general hospitals and, as indicated in our field notes, we made a sustained and sometimes frustrated
effort to sample participants from a wide variety of settings because they were scattered amongst a
number of programs and institutions. We are left to consider the implications of these system-wide
changes on our analysis of data that was collected from participantsin the context of a very different
intervention climate and on our own thinking in this area as it has evolved over time. More simply,
we wonder how help-seeking has changed in the interim between the analysis of the data for the
QSA and current research efforts. Our research team has developed and launched a new program of
research that further investigates the findings of this secondary analysis by studying pathways to
care that account for the turmoil associated with the subtle, help-seeking activities. Two separately
designed studies – using a qualitative case study approach that includes the various players and the
interaction betweensystems of care,including formal and informal care – will engage young people
and significant others in understanding pathways to care for those who have recently experienced a
first episode and those who are currently considered to be at an early, prodromal, Bhigh-risk^ state.
In forthcominganalyses ofparent narratives we willwanttothink about howoursecondary analysis
holds up in the current context of children_s mental health services. Following the newly emerging
literature on qualitative longitudinal analysis, our QSA and current program of research documents
perspectives from different periods of time and vantage points, producing a potential archive that
will be a useful comparative basis for understanding and explaining patterns of continuity and
change in help-seeking activities over time.26–28
Current guidelines for acquiring the appropriate contextual information to conduct QSA
has caused us to reconsider the adequacy of our knowledge given certain limitations of our
data associated with funding issues, changes in research personnel, and the development of
new technology to carry out qualitative analysis. A brief summary of extant guidelines
includes: having information about the interaction between the interviewer and the
interviewee (either detailed transcripts or audio/video tapes); access to the Bwhole
interview^; background information about the interviewer(s), including their age, gender,
race, and social class; and documentation of the place, time, and setting of the data
collection and how the participant(s) were recruited, including relevant information about
gatekeepers and issues of access.29
In our case, all three authors were involved in the design, data collection, analysis, and written
representation of both the original study and the QSA. We felt that this enabled us to be sensitive
and reflexive in our approach to contextual issues such as our relationship as researchers to the
participants in our study, our detailed knowledge about issues related to recruitment and access,
and the availability of the data as a whole for the purpose of multileveled analysis and so forth.
However, our involvement was not consistent across all domains. As is the case in many research
projects, we hired research assistants to conduct the interviews in the original study and they
moved on at the end of data collection as funding ended. Apart from graduate-level research,
Issues in Qualitative Secondary Analysis GLADSTONE et al.
fieldwork is often undertaken by hired hands, not by the investigators who will go on to analyze,
interpret, and write up the research. To what degree this matters to the qualitative research field as
a whole, given the emphasis on research relationships as an integral part of understanding the
social context, is a matter for reflexive dialogue amongst qualitative practitioners.
Nevertheless, keeping detailed documentation might be one strategy for overcoming the
inadequacies of particular data. For example, while we had access to fieldnotes and minutes
of research meetings recorded at the time of primary data collection and analysis, we found
they were limited in scope, poorly organized, and archived in an ad hoc fashion. We had not
anticipated future use, which became apparent as we conducted the QSA and were frustrated
by questions we could not answer. Our desire to be sensitive to the background information
we intended to collect about our participants meant that we would include only demographic
data that were salient for our purposes at the time. Many individuals in our previous studies
complained that researchers asked too many personal questions, particularly of a sensitive
nature, and thought they were irrelevant to the study in which they agreed to participate.
Consequently, we are left with a quandary based on the need to anticipate what information
might be pertinent for secondary analysis. Moreover, if we collect data in anticipation of what
we might want to know, are we in danger of becoming inveterate data collectors not because we
seek Bnew^ data, but as a consequence of stockpiling data for as yet unknown future research
During our analysis of the data in the secondary study we had access to copies of the previously
transcribed interviews, but we were unable to produce new, unmarked copies to carry out the
QSA. In the intervening years between the two studies we had updated the computer software that
we use to manage qualitative data, and as a result, we could not reproduce clean copies of our
transcripts. Consequently, we resorted to a rather crude but useful technique of folding over the
older, coded material on the original pages of the transcripts so that it would not interfere with our
secondary coding and analysis. The shelf life of current technology is not long, and therefore,
access to data in the form that was originally used with the primary data may not be available for
QSA. Furthermore, the human and financial resources necessary to organize and manage
databases over time are rarely sufficient. QSA underscores the necessity of keeping detailed
records of observations and reflexive fieldnotes, as well as accurate, Bthick descriptions^ of the
data recorded during the course of a primary study, not only for present use as is the case in all
Bgood^ qualitative practice but also for future, secondary use of qualitative data. In our first study
we did keep field notes; however, in retrospect we find these limited both in volume and content.
Of course, they attend to what mattered most at the time of the first study and we can, for
example, track discussions about data analysis and interpretation. They are not, however,
exhaustive, and cannot answer all the questions we currently have about our data in light of the
new question we are asking in this study. This highlights the need to consider how much
information is enough information for future work based on primary data and, more importantly,
to consider how we think about context in light of an analysis that we want to be a Bthick
analysis^. However, we note that we did not have access to the original audio-tapes used to
produce the transcripts as they were destroyed as per our agreement with the ethics review board
and the consent obtained from our participants in the original study. This may be a limitation of
the study from the perspective of a QSA. In our experience, this is common practice in qualitative
research, a discussion of which has implications for research more broadly.
In our current work, we have become quite concerned about the Bflat^ quality of transcripts
based on oral interviews and what is lost in translation between listening to an interview and
reading one for the purposes of analysis and interpretation. Other qualitative researchers have
documented this concern.30,31Being able to listen to the original audio or video recordings at the
time of primary data collection and analysis in conjunction with reading the Bverbatim^ transcript
itself may be more fruitful for our interpretations and representations. It is particularly important
The Journal of Behavioral Health Services & Research34:4 October 2007
to consider how this holds true for future Breadings^ of primary data. Alternatively, Watson writes
convincingly about the advantage of working with transcribed data, described as temporally
dislocated from the immersion of an interview in such a way as to allow for a different, and
potentially constructive analytic relationship to the data to develop.32The passage of time is
beneficial because it allows the analyst a different, perhaps more distant perspective on the
interview and the relationship between the researcher and the participant.
While the three authors reviewed the transcripts at all levels of analysis for the QSA and
reached agreement regarding the coding of transcripts and the development of themes and
concepts, the idea of consensus can be misleading because published reports often leave readers
with a rather sanitized impression of the analytic process. Only after laborious and time-
consuming discussions, in which we (re)visit the data, accept and reject coding schemes, and
(re)develop our concepts, do we actually arrive at an analysis that we are willing to put forward
for that particular account. However, this is also true for qualitative research overall. Moreover,
no one analysis can ever account for all that might be possible in interpreting data; there is always
something, a particular theme or higher-level concept that might be left aside for the immediate
purposes of that project or manuscript. This might also account for the fact that published reports
of a qualitative research project often give the impression that an analysis of data is complete,
whereas a secondary analysis of that same data set may be warranted, or perhaps even a third or
fourth depending on the purposes of those using the data.
Informed consent, confidentiality, and anonymization
Ethical considerations in conducting QSA may appear easier at the outset because there are no
face-to-face interviews or observations of human behavior undertaken. Yet, issues of informed
consent and confidentiality, for example, are not eliminated; rather, they are highlighted in
different ways by the possibility of conducting QSA. Thorne has suggested that researchers must
consider whether asking new questions of previously collected data actually violates the consent
obtained at the time at which the original study was conducted.9BEthical secondary research
therefore requires a defensible judgment as to the scope of the original consent as well as
an analysis of the specific conditions under which secondary analysis would be
appropriate^.9(p. 551)Current practice on institutional ethics review boards in many jurisdictions
now make provisions for participants to consent to the future use of data for other purposes;
however, we have concerns about how participants give informed consent to future studies. This
was not the case at the time our original study was conducted. We took responsibility for this
decision and felt reasonably assured that no harm would come to our participants through
questioning the data in the way that we intended; rather, we discussed our obligation to tell stories
that may be more important to our participants as evidenced in this case by parents_ narratives of
help-seeking activities. Currently, this decision has often been left up to the professional judgment
of the individual researcher; however, a number of institutional review boards now include
secondary analysis within their remit and are beginning to grapple with the unique set of
challenges that arise in conducting QSA. For example, in contacting former participants for QSA
and considering the implications of this process, does the researcher follow the commitments
made in the original protocol? Moreover, what guidance do researchers require to approach
people whose lives have gone on in the interval between studies and may include potentially
stressful life events and/or participants who no longer wish to participate in the research? At the
time we conducted the original study, no such provision was made to obtain consent for future use
of our data, and it was also not feasible to obtain consent for the QSA. We chose to anonymize
our data to protect participants_ identities.33However, because the research team was involved in
all aspects of the original study, we also felt confident that anonymization would not completely
jeopardize our understanding of issues salient to considering context as discussed earlier;
Issues in Qualitative Secondary Analysis GLADSTONE et al.
however, the impact of anonymization in a QSA is not a wholly different consideration from that
of anonymization of qualitative data overall, in which understanding the analysis and
interpretation of a particular study may be compromised by stripping the report of certain
features of the phenomenon or situation under consideration.34
QSA: benefits and costs
We found that a number of arguments in favor of QSA also rang true in our experience
conducting a study of this kind. For example, we found QSA to be an appropriate and productive
method for generating new knowledge because it allowed us to ask a new question of our data,
expanding our understanding of a particular phenomenon within current theories about parenting,
mental health, and first-episode psychosis.2,5,35In this case, asking a new question using QSA may
have decreased the expectations placed on participants, many of whom are enrolled in other
studies because first-episode treatment clinics are often funded for evaluative as well as
intervention purposes. This is a benefit to researchers working with sensitive, vulnerable, or hard-
to-reach populations because it limits the overall burden placed on particular participants,
especially to Btalk^ more.10It is postulated that QSA allows researchers to remain sensitive to
people who may be leading stressful lives, such as caregivers for people with dementia35or
individuals living in difficult, hard-to-access, restricted environments, such as prisons.36Our
research data and on-going work with young people experiencing a first-episode of psychosis and
their caregivers suggests that they often fall into this category of people who are leading stressful
lives and also may be Bvulnerable^ and Bhard to reach.^ We also know from working in the
mental health field that the burden to participate in research can sometimes be overwhelming to
this particular group of participants. At the time this QSA was being conducted, several quali-
tative studies were underway in the same clinic where we had recruited participants from earlier.
In addition, gatekeeper effects can lead to difficulties, not only because parties seek to protect
people they consider vulnerable but because of professional territorial issues and the scientific
capital that is accrued in a field37where the emphasis is on seeking Bnew^ data from a limited
population of potential recruits. This is a conundrum as we are not advocating dispensing with
primary data collection, nor are we suggesting that QSA is always appropriate; our emphasis here
has been to suggest that there are questions that can be answered using data collected for other
purposes and that this may have a salutatory effect on an already burdened research population.
Of course, benefits do not preclude concerns about the costs of doing a QSA, which we will
touch on briefly here. For example, how do we consider funding studies that are based on QSA?
This QSA did not receive independent, external funding, and although the data were already
Bcollected^ (at least from the standpoint of the original study), there is an enormous investment of
time by the research team in coding, analysis, interpretation, and written representation of the data
for this study. In addition, qualitative databases entail a considerable investment of time and
money by researchers and funding bodies alike.2,5,9,10QSA is said to avoid duplication of efforts
and cut costs (although in the broad field of QSA, the resources required for archiving and
formally contracting shared data may suggest otherwise). On this latter point, it would be difficult
for one study such as ours to make claims about duplication of effort or cost, although the
considerable investment of time and money in the collection, maintenance, and storage of the
original data is not debatable.
Implications for Behavioral Health
Conducting a QSA of help seeking in the prodrome to first-episode psychosis has been
instructive in a number of ways that we think would be helpful to researchers working with their
own qualitative data. First, it is important to decide what kind of data are available to undertake a
The Journal of Behavioral Health Services & Research 34:4 October 2007
QSA depending on the purposes for the study. Developing informal criteria may also help to
establish when it is not possible or feasible to do so, and to come to terms about what is Bgood
enough^ data for a secondary analysis. Second, decisions made during primary data collection
will have implications for what kind of data will be needed for future use, keeping in mind that
accumulating data in anticipation of potential research may overwhelm current efforts both
practically and financially and possibly burden participants during the primary phase of data
collection. Third, in addition to questions about how many and what kind of fieldnotes,
transcripts, or audio or video tapes will be required for the QSA, storing such data in a safe,
confidential place over time will also be necessary alongside the human resources required to
manage the data. Researchers will have to anticipate potential technological changes and the
impact this might have on studies using primary data archived for future use. Moreover, the
individual practices and policies of various institutional review boards will govern how qualitative
data is handled after a primary study is complete, which will have an impact on the conduct of
QSA in different research environments. Fourth, researchers must consider context(s) in relation
to their work, how the substantive field may have changed in the interim between primary data
collection and a secondary analysis, and the impact these changes may have on the field as a
whole but also on the researchers and their theoretical, methodological, and practical approaches
to their research. Fifth, compensation for QSA is both a financial concern, as funding is necessary
to do the analytic work, and an academic one, as most disciplines currently recognize those
professional activities associated with primary data collection.
Our enduring question – considering to what extent QSA is a distinct (qualitative) research
enterprise – may only be partially answered. While we consider QSA to be a viable approach to
research in its own right, the process of conducting QSA generates many more questions that
apply not only to QSA but to qualitative research methods overall.
1. Heaton J. Reworking Qualitative Data. London: Sage; 2004.
2. Thorne S. Secondary analysis in qualitative research: issues and implications. In: Morse JM, ed. Critical Issues in Qualitative Research
Methods. Thousand Oaks: Sage; 1994:263–279.
3. Heaton J. Secondary Analysis of Qualitative Data: A Review of the Literature. 2000. Available at: http://www.york.ac.uk/spru.
Accessed June 24, 2004.
4. Heaton J. Secondary Analysis of Qualitative Data. Social Research Update 22. Guilford: University of Surrey, Institute of Social
Research; 1998. Available at: http://sru.soc.surrey.ac.uk/. Accessed June 24, 2004.
5. Corti L, Thompson P. Secondary analysis of archived data. In: Seale C, Gobo G, Gubrium JF, Silverman D, eds. Qualitative Research
Practice. London: Sage; 2004:327–343.
6. Fielding NG. Getting the most from archived qualitative data: epistemological, practical and professional obstacles. International
Journal of Social Research Methodology. 2004;7:97–104.
7. Cisneros Puebla C, Mruck K, Roth W. Editorial: The FQS Issue on BQualitative inquiry: Research, archiving and reuse^. Forum
Qualitative Sozialforschung/Forum: Qualitative Social Research [on-line journal], 6(2), Art. 45. Available at: http://www.qualitative-
research.net/fqs-texte/2-05/05-2-45-e.htm/. Accessed August 15, 2005.
8. Mauthner NS, Parry O, Backett-Milburn K. The data are out there, or are they? Implications for archiving and revisiting qualitative
data. Sociology. 1998;32:733–745.
9. Thorne S. Ethical and representational issues in qualitative secondary analysis. Qualitative Health Research. 1998;8:547–555.
10. Sandelowski M. BTo be of use^: enhancing the utility of qualitative research. Nursing Outlook. 1997;45:125–132.
11. Denzin N. Interpretive Interactionism (II). Applied Social Research Methods Series Volume 16. Newbury Park: Sage; 1989.
12. Boydell KM, Gladstone BM, Volpe T. Interpreting narratives of motivation and schizophrenia: A biopsychosocial understanding.
Psychiatric Rehabilitation Journal. 2003;26:422–426.
13. Clausen JA, Yarrow MR. Pathways to the mental hospital. Journal of Social Issues. 1955;11:25–32.
14. Friedson E. The Profession of Medicine. New York: Dodd Mead; 1970.
15. Czuchta DM, McCay E. Help-seeking for parents of individuals experiencing a first episode of schizophrenia. Archives of Psychiatric
16. Geertz C. Thick description: toward an interpretive theory of culture. In: Geertz C, ed. The Interpretation of Cultures. New York: Basic
17. Mills CW. The Sociological Imagination. New York: Oxford University Press; 1959.
18. Guba EG. The alternative paradigm dialog. In: Guba EG, ed. The Paradigm Dialog. Newbury Park: Sage; 1990:17–30.
19. Lather PA. Research as praxis. Harvard Educational Review. 1986;56:257–277.
Issues in Qualitative Secondary AnalysisGLADSTONE et al.
20. Greene JC. Three views on the nature and role of knowledge in social knowledge. In: Guba EG, ed. The Paradigm Dialog. Newbury
Park: Sage; 1990:258–276.
21. Lincoln YS. The makings of a constructivist: A remembrance of transformations past. In: Guba EG, ed. The Paradigm Dialog. Newbury
Park: Sage; 1990:8–104.
22. Williams CC, Collins AA. The social construction of disability in schizophrenia. Qualitative Health Research. 2002;12:297–309.
23. Hinds PS, Vogel RJ, Clarke-Steffen L. The possibilities and pitfalls of doing a secondary analysis of a qualitative data set. Qualitative
Health Research. 1997;7:408–424.
24. Braun V, Clarke V. Using thematic analysis in psychology. Qualitative Research in Psychology. 2006;3:77–101.
25. Holstein JA, Gubrium JF. Context: working it up, down, and across. In: Seal C, Gobo J, Gubrium J, Silverman D, eds. Qualitative
Research Practice. London: Sage; 2004:279–311.
26. McLeod J. Why we interview now-reflexivity and perspective in a longitudinal study. International Journal of Social Research
27. Thomson R, Holland J. Hindsight, foresight and insight: the challenges of longitudinal qualitative research. International Journal of
Social Research Methodology. 2003;6:233–244.
28. Yates L. Interpretive claims and methodological warrant in small-number, qualitative longitudinal research. International Journal of
Social Research Methodology. 2003;6:223–232.
29. Van den Berg J. Reanalyzing qualitative interviews from different angles: the risk of decontextualization and other problems of sharing
qualitative data. Forum Qualitative Sozialforschung/Forum: Qualitative Social Research [on-line journal], 6(1), Art. 30. Available at:
http://www.qualitative-research.net/fqs-texte/1-05/05-1-30-e.htm/. Accessed July 7, 2005.
30. Lapadat JC, Lindsay AC. Transcription in research and practice: from standardization of technique to interpretive positionings.
Qualitative Inquiry. 1999;5:64–86.
31. Poland B. Transcription quality. In: Gubrium JF, Holstein JA, eds. Handbook of Interview Research: Context and Method. Newbury
Park: Sage; 2002:629–649.
32. Watson C. Unreliable narrators? FInconsistency_ (and some inconstancy) in interviews. Qualitative Research. 2006;6:367–384.
33. Thomson D, Bzdel L, Golden-Biddle K et al. Central questions of anonymization: A case study of secondary use of qualitative data.
Forum Qualitative Sozialforschung/Forum: Qualitative Social Research [on-line journal], 6(1), Art. 29. Available at: http://
www.qualitative-research.net/fqs-texte/1-05/05-1-29-e/. Accessed August 21, 2005.
34. Nespor J. Anonymity and place in qualitative inquiry. Qualitative Inquiry. 2000;6:546–569.
35. Szabo V, Strang VR. Secondary analysis of qualitative data. Advances in Nursing Science. 1997;20:66–74.
36. Fielding NG, Fielding JL. Resistance and adaptation to criminal identity: using secondary analysis to evaluate classic studies of crime
and deviance. Sociology. 2002;34:671–689.
37. Bourdieu P. The specificity of the scientific field and the social conditions of the progress of reason. Social Science Information.
The Journal of Behavioral Health Services & Research34:4 October 2007