QBRDM #6

Q

Be a C b C c C

blo

�

International Journal of Nursing Studies 49 (2012) 360–371

A

Art

Re

Re

Ac

Ke

Qu

An

Int

*

(B

00

do

UAGOL: A guide for qualitative data analysis

rnadette Dierckx de Casterlé a,*, Chris Gastmans b, Els Bryon c, Yvonne Denier b

entre of Health Services and Nursing Research, Faculty of Medicine, Catholic University of Leuven, Kapucijnenvoer 35 blok d - bus 7001, 3000 Leuven, Belgium

entre for Biomedical Ethics and Law, Faculty of Medicine, Catholic University of Leuven, Kapucijnenvoer 35 blok d - bus 7001, 3000 Leuven, Belgium

entre for Biomedical Ethics and Law & Centre for Health Services and Nursing Research, Faculty of Medicine, Catholic University of Leuven, Kapucijnenvoer 35

k d- bus 7001, 3000 Leuven, Belgium

What is already known about the topic?

Qualitative data analysis is a complex and challenging part of the research process which has received only limited attention in the research literature.

� During the analysis process of qualitative data, quite a lot of researchers are struggling with problems that com- promise the trustworthiness of the research findings. � There is a lack of guidelines on how to analyze the mass

of qualitative interview data.

What this paper adds

� A theory- and practice-based guide that supports and facilitates the process of analysis of qualitative interview data.

R T I C L E I N F O

icle history:

ceived 27 June 2011

ceived in revised form 14 September 2011

cepted 16 September 2011

ywords:

alitative research

alysis

erview data

A B S T R A C T

Background: Data analysis is a complex and contested part of the qualitative research

process, which has received limited theoretical attention. Researchers are often in need of

useful instructions or guidelines on how to analyze the mass of qualitative data, but face

the lack of clear guidance for using particular analytic methods.

Objectives: The aim of this paper is to propose and discuss the Qualitative Analysis Guide

of Leuven (QUAGOL), a guide that was developed in order to be able to truly capture the

rich insights of qualitative interview data.

Method: The article describes six major problems researchers are often struggling with

during the process of qualitative data analysis. Consequently, the QUAGOL is proposed as a

guide to facilitate the process of analysis. Challenges emerged and lessons learned from

own extensive experiences with qualitative data analysis within the Grounded Theory

Approach, as well as from those of other researchers (as described in the literature), were

discussed and recommendations were presented. Strengths and pitfalls of the proposed

method were discussed in detail.

Results: The Qualitative Analysis Guide of Leuven (QUAGOL) offers a comprehensive method

to guide the process of qualitative data analysis. The process consists of two parts, each

consisting of five stages. The method is systematic but not rigid. It is characterized by iterative

processes of digging deeper, constantly moving between the various stages of the process. As

such, it aims to stimulate the researcher’s intuition and creativity as optimal as possible.

Conclusion: The QUAGOL guide is a theory and practice-based guide that supports and

facilitates the process of analysis of qualitative interview data. Although the method can

facilitate the process of analysis, it cannot guarantee automatic quality. The skills of the

researcher and the quality of the research team remain the most crucial components of a

successful process of analysis. Additionally, the importance of constantly moving between

the various stages throughout the research process cannot be overstated.

� 2011 Elsevier Ltd. All rights reserved.

Corresponding author.

E-mail address: [email protected]

. Dierckx de Casterlé).

Contents lists available at SciVerse ScienceDirect

International Journal of Nursing Studies

journal homepage: www.elsevier.com/ijns

20-7489/$ – see front matter � 2011 Elsevier Ltd. All rights reserved. i:10.1016/j.ijnurstu.2011.09.012

�

1

s w r w to o T s to th th S s h m

n e c H n u c s th c p c s s p ti th

h h h h in p g d m te e in

h e

in

2

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371 361

An experience-based and detailed description of the strengths and pitfalls of the Qualitative Analysis Guide of Leuven (QUAGOL).

. Introduction

Imagine, a study about nurses’ involvement in euthana- ia.1 The data are collected through in-depth interviews ith nurses having experience in the care for patients

equesting euthanasia. The first respondent is a man, orking in a neutral hospital, with a positive attitude ward euthanasia. He has 10 years of experience in

ncology care and has been involved in 8 euthanasia cases. he man speaks fluently and with conviction about the ubject. ‘Respecting the patient’s euthanasia request’ seems

be the main focus of his care. The most important role of e nurse, in his opinion, is to gain absolutely certainty that e euthanasia request is really what the patient wants.

ubsequently, the nurse must be sure that all the necessary teps of the procedure are taken. He tells you that the ospital protocol serves as checklist, which is for him the ost important instrument in the euthanasia care process.

The second respondent is a woman, working in a eutral hospital. She also has a positive attitude toward uthanasia. She has 5 years of experience on a geriatric are ward and has been involved in 3 euthanasia cases. ere, you are confronted with a quite different story. The urse tells you how important it is for her to be able to nderstand the patient’s request. Her most important oncern is: what is the right attitude for me in guiding and upporting the patient and the patient’s family through

is process? How should I be? Her primary focus in the are for these patients is to show respect for the patient as erson in the broad sense (a person with a specific haracter, particular life history, own wishes, fears, coping trengths and relationships). She describes in detail how he enters into a close and personal relationship with atients and their family in order to create a communica- onal atmosphere, within which she helps them spend eir final days together in a good way. A next respondent, again a man, working in a catholic

ospital, with a negative attitude toward euthanasia. He as 5 years of experience in a palliative support team and as been involved in 12 euthanasia cases. This time, you ear an emotional story, underlining the emotional tensity of being involved in euthanasia. Caring for a

atient requesting euthanasia is intense, difficult and rave, according to this nurse. ‘Truly helping the patient to ie serenely’ is the central message in his story. ‘As a nurse I ust do everything in my power to contribute to this’, he lls you in the interview. His story makes clear that a

uthanasia care process is only successful when everyone volved is able to make one’s peace with the situation.

The next participant is a woman, working in a neutral ospital. She has a pro-attitude and has 3 years of xperience on an oncology unit; she has been involved

in 2 euthanasia cases. You are confronted with a young nurse telling, again, a totally different story about nurses’ involvement in euthanasia. Her story is one about the organization of care. ‘Caring for a patient requesting euthanasia requires, first of all, an efficient, practical organisation of care’, she tells you. According to this nurse, the responsibility of the nurse is to find out what to ‘do’ to make this care process successfully.

And you can go on. You are confronted with pages and pages of interview data. Every respondent has his or her own unique story that can help you understand the nurses’ involvement in euthanasia care processes. How to analyze and interpret all these different data? How to understand their meaning and draw legitimate conclusions? How to grasp the essence of these data while protecting the integrity of each story when responding to the research question? These questions point to the real challenge of qualitative data analysis.

Data analysis is a complex and contested part of the qualitative research process, which has received limited theoretical attention (Savage, 2000). Researchers are often in need of useful instructions or guidelines on how to analyze the mass of qualitative data, but face the lack of clear guidance for using particular analytic methods (Hunter et al., 2002; McCance et al., 2001). Most available guidelines or checklists related to qualitative studies are critical appraisal tools or focus on reporting qualitative research such as the CASP (Public Health Resource Unit, 2006), COREQ (Tong et al., 2007), Malterud’s guidelines (2001), and McMaster Critical Review Form (Letts et al., 2007). They do not provide researchers with clear instructions on how to analyze, interpret and summarize qualitative data.

In trying to meet this need and fill this lack, we should not, however, forget to be careful. For on the one hand, there is growing consensus that understanding or using a prescribed method of analysis is not enough to generate new insights. Qualitative data analysis is very complex, and any description of the practical aspects of the analysis process runs the risk of oversimplification. There is no one right way to work with qualitative data. Essentially, qualitative data analysis is a process best ‘learnt by doing’ (Froggatt, 2001).

On the other hand, we need to bear in mind that the ‘Aha-erlebenis’, the moment where one makes meaning beyond the facts, does not just happen out of the blue (Hunter et al., 2002). No themes, categories, concepts or theories will ‘emerge’ without the researcher who must ‘make it so’ (Sandelowski, 1995, p. 371). This requires expertise in reading, thinking, imagining, conceiving, conceptualizing, connecting, condensing, categorizing and thereby creating a new storyline (Jennings, 2007). This implies the development of ‘intellectual craftmanship’ (Mills, 1995/1978, p. 195) without which no valuable qualitative work can be produced (Sandelowski, 1995). Extensive preparation is required to open the researcher’s mind to multiple meanings and perspectives and to lay the groundwork for one to be creative (Hunter et al., 2002). In qualitative research it is essential that we ask which techniques or methods can be used to guide and support researchers in this challenging intellectual process (Jen- nings, 2007; Hunter et al., 2002).

1 The following examples are inspired by our studies about nurses’

volvement in euthanasia in Flanders, Belgium (Denier et al., 2009,

010a,b; Dierckx de Casterlé et al., 2010).

2.

an m ex six

2.1

co qu Th vo of fa m th co

an on w on co so no sit ge ‘ah ho se ex th tiv

2.2

an by at or ha th ki co an tiv in str te sto

2.3

w th Ja

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371362

Problem statement

The process of qualitative data analysis is an extensive d challenging activity, confronting the researcher with any problems. Based on the literature and on our own periences with qualitative data analysis, we can discern

major problems researchers are often struggling with.

. Over-reliance on qualitative software packages

Figuring out what to do with the data once they are llected is one of ‘the most paralyzing moments’ in alitative analyses (Jennings, 2007; Sandelowski, 1995). e data generated with qualitative methods are often luminous, and researchers are faced with the challenge

grasping a sense of the whole, extracting significant cts, distinguishing relevant themes, discovering the eaning beyond the facts and ultimately reconstructing e story of the respondents on a general, overarching and nceptual level. The problem of figuring out how to start the process of

alysis frequently results in researchers relying too heavily qualitative software packages (Jennings, 2007). Over-

helmed by all the narrative material that they must work , researchers often focus too quickly and exclusively on ding the data and entering the codes into qualitative ftware packages (Jennings, 2007). Researchers often do t take the necessary time to read and reread the material,

back and reflect on what one has read, trying to grasp the neral themes and storylines and coming to the necessary a-erlebnis’ (Hunter et al., 2002). Software cannot decide w to segment data or what codes to attach to these gments, nor what data means (Sandelowski, 1995). An tensive preparation of the coding work is required to open e researcher’s mind to multiple meanings and perspec- es (Hunter et al., 2002).

. Word overload due to line-by-line approaches

Another problem that often occurs in qualitative alysis, is word overload, which is produced by line- -line approaches to coding. In such cases, the researcher taches labels to lines of data without a sense of the whole

of analytic direction. Consequently, these lines either ve no meaning by themselves or have more meanings an can be grasped by one label (Sandelowski, 1995). This nd of coding is meaningless. It is analytically and ntextually empty and produces nothing but fatigue d frustration. The generalizations developed in qualita- e analyses are embedded in the contextual richness of

dividual experience. Qualitative data management ategies that depend solely on coding and sorting of

xts into units of like meaning will give up much of the ry’s contextual richness (Ayres et al., 2003).

. Coding using a preconceived framework

Further, many researchers struggle with the dilemma of hether or not to perform pure inductive coding or to code e data with the help of preconceived notions (Bailey and ckson, 2003). Using a preconceived framework runs the

risk of prematurely excluding alternative ways of organiz- ing the data that may be more illuminating. As such, one runs the risk of premature analytic closure, resulting from a persistent (but often unconscious and unrecognized) commitment to some a priori view of the subject under investigation (Sandelowski, 1995).

2.4. Difficulty of retaining the integrity of each respondent’s

story

The feeling of losing the uniqueness of each of the individual interviews is another problem in the analysis of qualitative data (Bailey and Jackson, 2003). This is characteristic for the analytical process, which does not always respect the interviewees’ particular portrayal of their stories. The analytical method segments the data, thus limiting the researchers’ understanding of the interviewee’s perspective. As such, it prevents them from understanding and describing a participant’s experience in its richness (Bailey and Jackson, 2003; Riessman, 1990). The content of each interview is unique, differing from the other interviews qua experiences, tone, emotional invol- vement, physical involvement, etc. How to retain the integrity of each respondent’s responses constitutes one of the most important challenges that qualitative researchers are faced with (Bailey and Jackson, 2003).

2.5. Full potential of data is not exploited

Next, the analysis does not always go beyond a mere descriptive account. It does not always offer a thorough interpretation or theoretical development, although the use of a Grounded Theory Approach is reported. It happens that explanation is oversimplified and the complexities of the research phenomena are ignored, so that the ambi- guities and diversities of the participants’ experiences are not reflected in the final description (Froggatt, 2001). In such cases, we meet research reports that present only lists of themes and subthemes, but stop short of interpretation. Here, the full potential of the data is not exploited. The analysis does not offer a thorough interpretation of the interviewee’s world, which clearly undermines the cred- ibility of the results.

This type of merely descriptive presentation happens, for instance, when the analysis is separated out as a discrete activity without analogously undertaking an iterative dialogue with the interview data. It also occurs when deductive rather than inductive analysis is under- taken or when too much emphasis is being placed upon allowing the data to speak for themselves. In such cases, we see papers that successively present large fragments from interviews with little explanation or interpretation, with no attempt to identify commonalities within the data, and without clarification of the purpose of the quotes (Froggatt, 2001).

2.6. Data analysis as individual process

Finally, conceiving the qualitative data analysis as an individual process rather than a team process is also a common problem among qualitative researchers, leading

to H d s m e in p to r

d a C a c fi a b m d o

3

Q th r b w li m S th

4

b e p p s p

th a g o li a s 2 fi p

c th d S

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371 363

personal frustration and little depth in the analysis. unter et al. (2002) underscores the importance of viewing ata from several perspectives facilitating multidimen- ional thinking and offering different ways of making eaning of the interview data. A team approach will

nhance the possibility to gain creative and thoughtful sight in the research phenomenon. Jennings (2007) also

oints to the importance of mentors, rather than manuals, guide the researcher in grasping the essence of the

esearch findings. As qualitative researchers we experienced similar

ifficulties in analyzing qualitative interview data within Grounded Theory Approach (Glaser and Strauss, 1999; orbin and Strauss, 2008). The process of analysis, as well s the guidance of young researchers in this process onstitutes a real challenge. As such, we were triggered to nd a method that could support researchers in the nalytical process without imposing a rigid, detailed step- y-step plan. We searched for a supporting guide that akes researchers able to understand the meaning of the

ata in a consistent and scientific way, sufficiently based n the use of intuition, imagination and creativity.

. Aim

The purpose of this article is to propose and discuss the ualitative Analysis Guide of Leuven (QUAGOL), a guide at we developed in order to be able to truly capture the

ich insights of qualitative interview data. The QUAGOL is ased on our own experiences with qualitative research as ell as on that of other researchers (as described in the

terature) and is inspired by the constant comparative ethod of the Grounded Theory Approach (Corbin and

trauss, 2008). QUAGOL is proposed as a guide to facilitate e process of qualitative data analysis.

. The Qualitative Analysis Guide of Leuven (QUAGOL)

The proposed method is comprehensive and systematic ut not rigid; it offers space that stimulates the research- r’s intuition and creativity as maximal and optimal as ossible. The method gets the researcher out of his isolated osition as the analysis process is predominantly con- idered as a team activity rather than a purely individual rocess.

The process of analysis consists of two parts: (1) a orough preparation of the coding process and (2) the

ctual coding process using a qualitative software pro- ram. Both parts consist of 5 stages which, for the purpose f this article, are summarized artificially as discrete and near stages. However, in reality, our method is char- cterized by iterative processes of digging deeper, con- tantly moving between the different stages (Froggatt, 001). The process of analysis immediately starts after the rst interview has been conducted and continues till the oint of data saturation has been reached.

The first part consists of a thorough preparation of the oding process, implying only paper and pencil work. In

is part, the researcher and his team explicitly and eliberately postpone the process of actual coding. As andelowski (1995, p. 371) reports, ‘first look at your data

in order to see what you should look for in your data’. This preparatory work is crucially important to develop a useful and empirically based framework for the actual coding process.

While the first part happens by paper and pencil work, the stages of the second part require the use of qualitative software, as we start with the actual coding process. Based on the conceptual insights resulting from the previous stages, a list of contextually and analytically meaningful concepts is drawn up. It serves as a coding list for the actual coding process allowing a systematic analysis of the concepts based on empirical data. This part ends with an empirically based description of the results. Fig. 1 offers a schematic overview of the 10 stages in the process of data analysis.

As the collection and analysis of data occurs simulta- neously, both parts cannot be strictly separated. Newly collected data, even at the end of the study, require that the researchers go through the previous stages again, inevi- tably resulting in partial overlap and interaction between both parts of the process of analysis.

4.1. Preparation of the coding process

4.1.1. Stage 1: thorough (re)reading of the interviews

Every interview is meticulously transcribed verbatim immediately by the interviewing researcher, including the non-verbal signals. Additionally, a short report about the interviewee’s and contextual characteristics of the inter- view is made, helping the researcher to comprehend the interview within its particular context. The transcript is thoroughly read different times in order to familiarize with the data and getting a sense of the interview as a whole. What is this interview about? What does this participant tell me that is relevant for the research question? As the analysis is considered as a team process, the transcript is also read by the other members of the research team. Each interview is read as many times as necessary to apprehend its essential features, without feeling pressured to move forward analytically. During this reading process, the researcher will underline key phrases, simply because they make some, though yet embryonic, impression on him/her. The meaning of some words or passages, as interpreted tentatively by the researcher, thoughts or reflections evoked by some passages are noted in the margins next to the text. It is clear that a rudimentary kind of analysis begins in this stage. Fig. 2 offers an example of the results of the (re)reading process.

4.1.2. Stage 2: narrative interview report

Stage 1 results in a holistic understanding of the respondent’s experience. In the second stage, the researcher tries to phrase (articulate) this understanding.

The interview is read again and put aside. Then, the researcher tries to articulate the essence of the inter- viewee’s story in answer to the research question. The writing of the narrative report is guided by the question: ‘What are the essential characteristics of the interviewee’s story that may contribute to a better insight in the research topic?’ The answer is described in a narrative way, using the specific story of the interviewee. The narrative report

1. Thorough (re)reading of the interviews A holis�c understanding of the respondent’s experience

2. Narra�ve interview report A brief abstract of the key storylines of the interview

3. From narra�ve interview report to conceptual interview scheme Concrete experiences replaced by concepts

4. Fi�ng-test of the conceptual interview scheme Tes�ng the appropriateness of schema�c card in dialogue

5. Constant comparison process Forward-backwards movement between within -case and across -case analysis

ACTUAL CODING PROCESS (using qualita�ve so�ware)

PREPARATION OF CODING PROCES (paper and pencil work)

6. Draw up a list of concepts A common list of concepts as preliminary codes

7. Coding process – back to the ‘ground’ Linking all relevant fragments to the appropriate codes

8. Analysis of concepts Descrip�on of concepts, their meaning, dimensions & characteris�cs

9. Extrac�on of the essen�al structure Conceptual framework or story -line

10. Descrip�on of the results Description of the essential findings

T

E

A

M

P

R

O

C

E

S

S

Fig. 1. Stages of the Qualitative Analysis Guide of Leuven (QUAGOL).

B .

D ie

rck x

d e

C a

ste rlé

e t

a l.

/ In

te rn

a tio

n a

l Jo

u rn

a l

o f

N u

rsin g

S tu

d ie

s 4

9 (2

0 1

2 )

3 6

0 –

3 7

1 3

6 4

c m n v b im

in ti a v

to m n th

4

s

n c a s le C a th in c p b

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371 365

an include brief paraphrasing that stays close to the data, ore abstract renderings of the data, or comments on the

arrative structure or interactional features of the inter- iew event (Sandelowski, 1995). This stage results in a rief abstract of the key storylines including a summary

pression of the characteristics of the interview. It is suggested to start the second stage after some

terviews have been conducted and to select (in consulta- on with the other team members) the interview that ppears to provide the most ‘rich’ information, i.e. the most aluable information to contribute to the research aim.

Focusing on the real essence of the story, it is suggested limit the narrative report to one page. Analogously, all embers of the team read the interviews and make

arrative interview reports, which are discussed during e meetings of the research team.

.1.3. Stage 3: from narrative report to conceptual interview

cheme

While the narrative interview report provides a general, arrative view of the essence of the interview, the onceptual interview scheme provides concepts that ppear relevant to get insight into the research topic. As uch, the researcher makes a first move from the concrete vel of experience to the conceptual level of the story. oncrete experiences are being replaced by concepts rising from these experiences. What has been told during e interview and (narratively) described in the narrative terview report is being brought to a more abstract and

onceptual level. The researcher distances from the articularity of the interview and the narrative report, y filtering the most important data and clustering them in

concepts. Which concepts grasp the essence of the interview in response to the research question? All- embracing concepts must be avoided in this stage as one looks for manageable concepts that will guide the coding process. The key concepts – those considered as most characteristic for the interview – are highlighted; they can help find the essential structure of the research answer (see stage 9). The concepts are represented in a scheme and, where necessary, clarified with respect to their content (see Fig. 3).

The translation of the narrative report into a conceptual interview scheme is a crucial preparatory stage for the actual analysis of the data with the qualitative software as this scheme will facilitate the transition from raw data to manageable concepts. The concepts will be further developed and refined as the researcher gets more insight into the research phenomenon. We experienced these schemes as an important analytic instrument to retain the integrity of each respondent’s story. It also helps in keeping track of the data as a whole, since every interview will have its own conceptual interview scheme. After having analyzed 20 interviews, one can easily go back and grasp the essence of the first interview by looking into the conceptual interview scheme of this interview. Further- more, these schemes are also an important instrument of communication within the research team, for they provide the researchers with a strategy to support the trustworthi- ness of the process of analysis.

We have observed that more experienced researchers sometimes skip the second stage and immediately start with the formulation of the conceptual interview scheme after having read the interview.

Fig. 2. Example of the results of the (re)reading process.

4.1

sc

in w w qu th im Ar

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371366

.4. Stage 4: Fitting-test of the conceptual interview

hemes

In stage 4, the appropriateness of the conceptual terview schemes is being verified by iterative dialogue ith the interview data. The researcher reread the interview ith the conceptual interview scheme in mind. Two estions need to be answered: (1) Does the content of e conceptual interview scheme actually reflect the most portant concepts in answer to the research question? e there any other important concepts the researcher

overlooks? (2) Can the concepts of the conceptual interview scheme be linked to the interview data? Through scrapping, completion or reformulation, the conceptual interview schemes are adapted, completed or refined.

Characteristic for this stage is that it represents the first forward–backward movement. In stage 1 till 3 we went forwards, starting from the interview data, and then formulating the narrative interview reports, followed by translation into conceptual interview schemes. Stage 4 stimulates the researcher to go back to the interview data.

Fig. 3. Example of a conceptual interview scheme.

v o

4

m w c 1 a th c c c b r c d v m th m fu u o e p h

s in c

4

4

e in s im w a T r m in T b r b s n y L

4

E a

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371 367

A comparison and discussion of the conceptual inter- iew schemes within the research team will help to further ptimize the schemes.

.1.5. Stage 5: Constant comparison process

Stage 5 is characterized by a forward–backward move- ent between within-case and across-case analysis which ill facilitate the identification of common themes, con-

epts or hypotheses (Swanson-Kauffman and Schonwarld, 988). The concepts of the conceptual interview schemes re further tested and developed through comparison with e schemes and data of the other interviews. New themes,

oncepts or hypotheses discovered in new interviews are hecked for their presence in the previous interviews. The onceptual interview schemes are adapted to and refined ased on these new insights. This gradually allows the esearcher to find the essential and common themes and oncepts throughout the interviews which are consequently escribed in a common and overarching conceptual inter- iew scheme. These constant forward–backward move- ents with reflection on and adjustments of the common emes, concepts or hypotheses is carefully reported (using emos) and will guide the researcher during the process of rther data collection and analysis. This information can be

seful in allowing the researcher to chart the development f ideas throughout the analytical process and help provide vidence of why particular decisions were made during the rocess (Froggatt, 2001) and to demonstrate how a concept as been developed.

This stage will end in an increasing conceptual under- tanding of the research data as a whole, retaining the tegrity of each individual case but taking into account the

haracteristics of other cases.

.2. The actual coding process

.2.1. Stage 6: draw up a list of concepts

By now, we have a well thought-out conceptual view of ach particular interview, as well as of all the available terviews together. Based on the conceptual interview

chemes, a common list of concepts is drawn up without posing an hierarchical order (see Table 1). All concepts

e have used so far in the conceptual interview schemes re listed and may represent different levels of abstraction. he list of concepts is evaluated and discussed within the esearch team; overlap or vagueness are remedied by

utual consensus. The resulting list of concepts is troduced as preliminary codes in the software program.

he researcher is not yet allowed to categorize the codes ecause a premature hierarchical organization of the codes isks imposing a structure on the data that is not supported y them, thus preventing the development of other tructures and insights. In this stage, the concepts are ot yet filled in with concrete interview data. They are not et empirically supported, described and explained. inking concepts is not yet recommended in this stage.

.2.2. Stage 7: coding process – back to the ‘ground’

The actual coding process starts in the seventh stage. ach interview is read again with the list of concepts

A critical use of the list is of crucial importance. Does this list help me to reconstruct the story-line? To which extent do the concepts help me to identify and classify the significant passages in the interviews? Each significant passage of the interview is linked to one of the concepts of the list. If no concept is found to be linked to in a particular interview, the list may need to be adapted. Every new concept is verified in the light of the other interviews. Does the missing concept also appear as an essential concept in other interviews? Can we explain why the concept is present in some and not in other interviews? Can we link other interview fragments to this missing concept?

Analogously, the researcher examines the ‘quality’ of the concepts of the list. Are the concepts sufficiently defined and well-delineated to capture all significant ideas, messages or hypotheses in a differentiated way? Codes that are too abstract (embracing large parts of interviews) as well as codes that are too concrete (broadly overlapping with concrete interview data) will prevent an efficient coding process. Questions or comments regard- ing concepts, their meaning or name, are reported (in memos) and discussed within the research team. This critical use and adaptation of the concepts will help to optimize the coding list.

4.2.3. Stage 8: analysis and description of concepts

After having linked all the relevant fragments of the available interviews to the appropriate codes (stage 7), the researcher proceeds on the across-case analysis of the concepts. Every code is analyzed through a careful exploration and study of all citations associated with the code. This analysis is guided by the following questions: Does every citation fit with the concept? Is there one

Table 1

Example of nonhierarchical list of concepts.

Farewell

Physician-Patient

Autonomy patient

Understanding

Experiencing

Respect

Contradictories in the care assignment

Discussibility

Assisting in care

Broad and extensive guidance

Coordination

Gratefulness

Sorrow

Delegate

Delicate

Irrevocable

Showing emotions

Emotional preparation

Ethics

Existential

No Protocol

One chance

Evolution

Dynamics

Impact

Powerlessness

Human involvement

. . .

ommon message describing the essence of the concept or

t hand. c

ca th su co

th A w ap in su re m em re w

4.2

th sta co re vi es re he m all all sto

4.2

st th th th ha ca th fin w de co an th

m de es re ac th an ca if di pe ex re

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371368

n we discern more than one message? Can we maintain e concept as such, or do we have to split it into several bconcepts? Or, reversely, do the empirical data suggest ngregating various concepts into one? Next, the researcher tries to understand and articulate

e specific meaning of the concepts in his/her own words. deeper analysis of the concepts allows to find out when, here, why, and in which circumstances the concepts pear. In this way, the tentative concepts (as formulated

the coding list) are cleaned up, delimited and defined. As ch, a thorough analysis of the empirical data allows the searcher to give a clear description of concepts, their eaning, dimensions and characteristics, grounded in the

pirical data. Such in-depth analytical work can only sult from an intensive and collaborative effort of the hole research team.

.4. Stage 9: extraction of the essential structure

Stage 8 results in a list of rather isolated concepts and eir meaning, dimensions and characteristics. The aim of ge 9 is to integrate all these concepts in a meaningful

nceptual framework or story-line in response to the search question. Inspired through the conceptual inter- ew schemes of all available interviews (referring to the sential structure of each interview separately), the searcher tries to formulate a conceptual framework lping to organize and structure all concepts in a eaningful way. This framework, again, is verified against

interviews and interview schemes. Does this framework ow us to describe and explicate all individual interview ries?

.5. Stage 10: description of the results

At this stage, the researcher is able to reconstruct the ory of the respondents, this time on a conceptual, eoretical level, grounded in the interview data. Based on e conceptual framework (stage 9) on the one hand and e in-depth analysis of concepts (stage 8) on the other nd, the researcher is able to systematically and refully describe the essential findings in answer to e research question. The description starts with the core dings (the core category and related concepts) after

hich the researcher systematically and carefully scribes and explicates the concepts and their inter- nnection. Significant quotes are added where necessary d relevant to fully grasp the essence of the concepts and eir relation. Even in this final stage, the constant comparison

ethod is used to continually check, discuss and further velop the theoretical insights. After having described the sential research findings, the researcher will again read all the interviews for a final evaluation of the curacy and comprehensiveness of the storyline. Does the eory fit with all interviews? Are there missing concepts d if there are, are they essential? Are there negative ses (cases that appear to disconfirm earlier findings) and there are, can the researcher explain these differences or screpancies? Next, the results are checked by a formal er debriefing, during which an interdisciplinary panel of ternal experts discuss the results in answer to the search question.

Ideally, this stage gives rise to a theory or theoretical model in answer to the research question. However, due to methodological and practical limitations, the results are often limited to the development of theoretical concepts and their mutual relationships, allowing to describe and explain the phenomenon under investigation.

5. Discussion

5.1. Strengths of the method

The method described in this article is presented as a guiding tool in the analysis of qualitative interview data. According to our experiences, this guide can serve as a valuable aid in the qualitative analysis process. The strengths of the guide lie in the underlying principles on which the guide is built, most of which have been supported by other authors: a case-oriented approach characterized by a continual balancing between within- case and cross-case analysis (e.g. Ayres et al., 2003; Sandelowski, 1995, 1996); a forward–backward dynamics using the constant comparative method (e.g. Froggatt, 2001; Glaser and Strauss, 1999; Sandelowski, 1995, 1996); the combination of analytical approaches (e.g. Coffey and Atkinson, 1996; Hunter et al., 2002; Savage, 2000; Simons and Lathlean, 2008; Sandelowski, 1996); use of data- generated sensitizing concepts (Sandelowski, 1995); its focus on peopleware rather than software (e.g. Jennings, 2007; Hunter et al., 2002; Sandelowski, 1995) and interdisciplinary team approach.

5.1.1. A case-oriented approach

A case-oriented approach focuses on the understand- ing of ‘a particular in the all-together’ (Sandelowski, 1995, 1996). The appropriate initial approach to quali- tative data analysis is to understand and treat each sampling unit as one case. The researcher must, first and foremost, make sense of the data collected for each individual sampling unit. Looking at and through each case is, according to Sandelowski (1996), the basis from which researchers may make idiographic generalizations, syntheses or interpretations of data. Next, the researcher proceeds to a cross-case analysis looking for common- alities and differences across cases (Ayres et al., 2003; Sandelowski, 1996).

The combination of within-case and across-case analysis techniques produces contextually grounded findings, retaining the integrity of each interview and taking into account the context of other interviews. Generalizations are, as reported by Ayres et al. (2003), embedded in the contextual richness of individual experiences.

However, the sequence is important. According to Ayres et al. (2003), information must first have explanatory force in one case. Ideas or insights from one case sensitize the researcher to look for similar information in other cases. Only when an idea occurs repeatedly in multiple contexts, it can be instantiated as a theme. When a theme has explanatory force in individual cases, as well as across several cases, it will most likely also apply beyond one sample (Ayres et al., 2003).

5

c

w d r d p th H h th e fo r r r b

p id a b u T e 2

5

a m a h o s 1 th th

fo r a to tr (S c ‘‘ a c a K p o c 2 lo

d a 1

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371 369

.1.2. A forward–backward dynamics using the constant

omparative method

From the start till the end of the process, the analytical ork is characterized by iterative processes of analysis in

ialogue with the data digging deeper and deeper in the esearch phenomenon (Froggatt, 2001). Data analysis is istinguished but not isolated from description and inter- retation (Sandelowski, 1995; Wolcott, 1994). (1) During e description stage, the data are to speak for themselves. ere, we have to answer the question: ‘‘What is going on ere?’’ (2) Within the process of analysis, we have to leave e purely descriptive account by inquiring into key

lements and the relationships between them. Here, we cus on the question ‘‘How do things work? How are they

elated?’’ (3) During the interpretation stage, we attempt to each an understanding about meaning, particularly in elation to context by focusing on the question ‘‘What is to e made of it all?’’ (Wolcott, 1994, p. 12).

The interplay between description, analysis and inter- retation and the continuous verification of developing eas, themes, hypotheses and concepts against available

nd newly collected data, allows the researcher to go eyond a descriptive account and to reach a deep nderstanding about meaning in relationship to context. his forward–backward move permits the researcher to xploit the full potential of the qualitative data (Froggatt, 001; Sandelowski, 1995).

.1.3. Combination of analytical approaches

The guide combines a traditional and creative analytical pproach permitting to view the data from different ethodological perspectives and preventing a line-by-line

pproach to coding. The combination of two approaches elps the researcher to find out alternative interpretations f the data and to elucidate different layers of under- tanding represented in the data (Coffey and Atkinson, 996; Savage, 2000). The use of two methods of analysis us offers greater complexity and depth in understanding e research phenomenon (Savage, 2000). The process starts with a creative and holistic approach,

cusing on the intuition, imagination and creativity of the esearcher. Starting with a case-oriented, narrative pproach, the researcher treats the case as a whole, trying

comprehend its essence; the features of the cases are eated as a whole rather than as disaggregated variables andelowski, 1996). The combination of within-case and

ross-case analyses contributes to the understanding of the particular in the all-together’’ (Sandelowski, 1996) nd facilitates the process of intuiting. Intuiting is the ritical reflection on and identification of themes as they re discovered in the stories of the respondents (Swanson- auffman and Schonwarld, 1988). This approach makes it ossible to develop themes in a way that it takes advantage f the richness of the data and does justice to the omplexity of the respondent’s experiences (Ayres et al., 003). However, in this approach we do not systematically ok for empirical support.

After getting a first sense of the whole, a more isciplined approach is used; an approach that is system- tically and consistently applied to all data (Sandelowski, 995). In this approach we explicitly look for empirical

support for our findings. More concretely, the holistic, narrative approach is followed by a more traditional process of thematic analysis, derived from and taking forward the findings of the first approach. The preliminary narrative approach prevents the researchers of getting lost in the details of the actual coding process in the thematic approach.

5.1.4. The use of data-generated sensitizing concepts as

coding framework

The guide proposes a compromise between a strictly inductive and a theory-driven coding system. The guide prescribes a thorough and extensive preparation of the coding process instead of a ‘line-by-line coding’, inviting the researcher to ‘first look at own data in order to see what he/she should look for in the data’ (Sandelowski, 1995). This preparatory work produces analytically and contex- tually meaningful concepts or codes that may help the researcher to grasp the essence of the research phenom- enon. These concepts are data grounded in the reality and should be considered as points of departure from which to study the data (Charmaz, 2000). These sensitizing, data- generated concepts offer ways of seeing, organizing and understanding experiences of respondents in a way that they make sense.

5.1.5. Focus on peopleware and not software

By focusing on a thorough preparation of the coding work, this method prevents the researcher from relying too quickly and too heavily on qualitative software packages, thereby getting lost in a meaningless mass of codes. The focus on an extensive and thorough preparation of the coding work and the combination of different analytical approaches allow the researcher to view the data from several perspectives and open his/her perception to multiple meanings and perspectives (Hunter et al., 2002). The researcher’s skills in thinking, imagining, conceiving, conceptualizing, connecting and creating are continuously helping him/her in finding meaning beyond the facts.

5.1.6. Interdisciplinary team approach

Last but not least, the interdisciplinary team approach constitutes one of the most important strengths of QUAGOL. Based on years of experience in qualitative research and in guiding young qualitative researchers, we discovered the essential value of teamwork in the process of qualitative data analysis. From the beginning, the process of analysis is predominantly considered as a team activity rather than an individual process. A team approach enhances the possibility to grasp the essence of the interview data, to correct misinterpretations and to obtain rich, well-considered and creative insight in the research phenomenon. The interdisciplinary composition of the team will contribute to the quality of the discussion and so to the trustworthiness of the findings.

5.2. Pitfalls

Our experiences with the QUAGOL guide of analysis are positive. It appears as a useful and helpful guiding tool both

fo an er th in th

5.2

in Fo th ar ca re str sto m co ex sta gr oc an

5.2

co di re gu be as to na su tri co

5.2

in

co sm co m fo es w sc He co ‘la co ca

5.2

gu sh un

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371370

r the process of qualitative data analysis and for teaching d supervision of less experienced qualitative research- s. However, there are potential pitfalls originating from e requirements associated with the method. Good sight in these potential pitfalls and strategies to prevent em, may enhance the usefulness of the guide.

.1. Distinguishing relevant from less relevant information

Distinguishing relevant information from less relevant formation constitutes a real challenge for the researcher. r fear of leaving out relevant information, especially in e beginning of the process of analysis, many researchers e tempted to select too much information. This choice n lead to an overload of information hindering the searcher to find meaning in the data. It is therefore ongly suggested to focus, first, on the essence of the ries rather than on the completeness of the stories’

essages in the preliminary stages of analysis. The nceptual interview schemes may be helpful in this ercise. We recommend the researcher to deliberately rt with a restricted selection of data. In case of not

asping important information in these stages, it will cur in a later stage when iterative, sequential methods of alysis are being used.

.2. Narrative report: key storylines

Writing down the narrative interview report also nstitutes a challenge. The difficulty lies in trying to scover the key storylines that are an answer to the search question. Hence, the research question must ide the analysis processes explicitly, and from the ginning onwards. We are not interested in the storylines

such, but in those messages in the story that contribute better insight in the research phenomenon. The rrative interview report, therefore, is more than a mere mmary of the content of the story. It is suggested to find cks (e.g. to stick the research question on the wall) to ntinuously call the research question to mind.

.3. From narrative interview report to conceptual

terview schemes

The translation of the narrative interview report in nceptual interview schemes does not always proceed oothly. Often, the researcher is inclined to add too much

ncrete information to the scheme in order to make them ore clear and complete. In this case, there is a risk of cusing too much on details, thus losing sight on the sence of the story. It is important to constantly question hether the information of the conceptual interview hemes is essential to respond to the research question. re, it is useful to carry out the development of the nceptual interview scheme in 2–4 stages, starting with a rge’ version (maximally two pages), and gradually ming to the ‘small’ version (on one page) by selecting refully the most essential information.

.4. Initial within-case analysis

Another important pitfall associated with the use of this ide, lies in the initial within-case analysis. Every case ould be considered and analyzed as a separated data it. We observed among researchers the tendency to

analyze one case, while being biased by the insights developed during the analysis of other cases. This results in conceptual interview schemes, which mainly focus on common insights rather than reflecting the essence and uniqueness of each particular case. Avoiding this pitfall requires strong analytical skills from the researcher as well the use of bracketing strategies. Breaking down the analytical work in three stages (narrative interview report, translation in conceptual interview schemes and validity testing) explicitly aims to prevent this problem. It is therefore suggested not to skip stages, especially not for less experienced qualitative researchers.

5.2.5. Choice and formulation of concepts

The choice and formulation of concepts is one of the most challenging activities of the qualitative researcher. The focus on intuition and creativity is a strength as well as a pitfall of the QUAGOL guide. It is the researcher who gives meaning to the data and does the abstract thinking, resulting in a framework of concepts. The suggested stages can only facilitate and optimize the quality of this abstract thinking (Jennings, 2007). Concepts need to be clear and unambiguous; they must fit with the data and contribute to the knowledge development in the research phenomenon. It is our experience that many concepts are too vague, all-embracing or abstract, making the actual coding work almost impossible. The transition from concrete to abstract data should be considered as a stepwise process activity. The more the process of analysis progresses, the more the researcher will be able to conceptualize. It is therefore suggested to start the analysis with the search for the most obvious messages, themes and ideas. It is our experience that interdisci- plinary teamwork optimizes the process of conceptuali- zation. Looking at the data from different perspectives actually contributes to a deeper and more nuanced understanding of the data.

6. Conclusion

The QUAGOL guide is a theory- and practice-based guide that supports and facilitates the process of analysis of qualitative interview data. Although the method as described above, can facilitate the process of analysis of qualitative data, it cannot guarantee automatic quality of analysis. The method is proposed as a guiding tool rather than as a strict procedure or technique that has to be implemented correctly step by step. The skills of the researcher and the common quality of the research team remain the most crucial components of a successful process of analysis. It is absolutely essential to consider the process of data analysis as a team activity. Finally, the importance of constantly moving between the stages throughout the research process cannot be overstated.

Conflict of interest. No conflicts of interest.

Funding. No funding.

Ethical approval. Not applicable.

R

A

B

C

C

C

D

D

D

D

F

G

H

Je

B. Dierckx de Casterlé et al. / International Journal of Nursing Studies 49 (2012) 360–371 371

eferences

yres, L., Kavanaugh, K., Knafl, K.A., 2003. Within-case and across-case approaches to qualitative data analysis. Qualitative Health Research 13 (6), 871–883.

ailey, D.M., Jackson, J.M., 2003. Qualitative data analysis: challenges and dilemmas related to theory and method. The American Journal of Occupational Therapy 57 (1), 57–65.

harmaz, K., 2000. Grounded theory: objectivist and constructivist meth- ods. In: Denzin, N.K., Lincoln, Y.S. (Eds.), Handbook of qualitative research. Sage, Thousand Oaks, California, pp. 509–535.

offey, A., Atkinson, P., 1996. Making Sense of Qualitative Data: Com- plementary Research Strategies. Sage, Thousand Oaks, California.

orbin, J., Strauss, A., 2008. Basics of Qualitative Research. Techniques and Procedures for Developing Grounded Theory, third edition. Sage Publication, Thousand Oaks, CA.

enier, Y., Dierckx de Casterlé, B., De Bal, N., Gastmans, C., 2009. Involve- ment of nurses in the euthanasia care process in Flanders (Belgium): an exploration of two perspectives. Journal of Palliative Care 25 (4), 264–274.

enier, Y., Dierckx de Casterlé, B., De Bal, N., Gastmans, C., 2010a. ‘‘It’s intense, you know.’’ Nurses’ experiences in caring for patients request- ing euthanasia. Medicine, Health Care and Philosophy 1 (13), 41–48.

enier, Y., Gastmans, C., De Bal, N., Dierckx de Casterlé, B., 2010b. Communication in nursing care for patients requesting euthanasia: a qualitative study. Journal of Clinical Nursing 19 (23–24), 3372– 3380.

ierckx de Casterlé, B., Denier, Y., De Bal, N., Gastmans, C., 2010. Nursing care for patients requesting euthanasia in general hospitals in Flan- ders, Belgium. Journal of Advanced Nursing 66 (11), 2410–2420.

roggatt, K.A., 2001. The analysis of qualitative data: processes and pit- falls. Palliative Medicine 15, 433–438.

laser, B.G., Strauss, A.L., 1999. The Discovery of Grounded Theory: Strategies for Qualitative Research. Aldine Transaction, New Jersey.

unter, A., Lusardi, P., Zucker, D., Jacelon, C., Chandler, G., 2002. Making meaning: the creative component in qualitative research. Qualitative Health Research 12 (3), 388–398.

nnings, B.M., 2007. Qualitative analysis: a case of software of ‘peopleware?’. Research in Nursing Health 30, 483–484.

Letts, L., Wilkons, S., Law, M., Stewart, D., Bosch, J., Westmorland, M., 2007. Critical Review Form – Qualitative Studies (Version 2.0) Retrieved September 9, 2011 from http://www.srs-mcmaster.ca/Portals/20/pdf/ ebp/qualreview_version2.0.pdf.

Malterud, K., 2001. Qualitative research: standards, challenges, and guidelines. The Lancet 358, 483–488.

McCance, T.V., McKenna, H.P., Boore, J.R.P., 2001. Exploring caring using narrative methodology: an analysis of the approach. Journal of Advanced Nursing 33, 350–356.

Mills, C.W., 1995/1978. The Sociological Imagination. Oxford University Press, London.

Public Health Resource Unit, NHS, England, 2006. Critical Appraisal Skills programme (CASP) Making Sense of Evidence. , In: http://www. sph.nhs.uk/sph-files/casp-appraisal-tools/Qualitative%20Appraisal%20 Tool.pdf.

Riessman, C.K., 1990. Strategic uses of narrative in the presentation of self and illness: a research note. Social Science and Medicine 30, 1195–1200.

Sandelowski, M., 1995. Focus on qualitative methods. Qualitative ana- lysis: what it is and how to begin. Research in Nursing, Health 18, 371–375.

Sandelowski, M., 1996. Focus on qualitative methods. One is the liveliest number: the case orientation of qualitative research. Research in Nursing Health 19, 525–529.

Savage, J., 2000. One voice, different tunes: issues raised by dual analysis of a segment of qualitative data. Journal of Advanced Nursing 31 (6), 1493–1500.

Simons, L., Lathlean, J., 2008. Shifting the focus: sequential methods of analysis with qualitative data. Qualitative health Research 18 (1), 120–132.

Swanson-Kauffman, K.M, Schonwarld, E., 1988. Phenomenology. In: Sar- ter, B. (Ed.), Paths to Knowledge: Innovative Research Methods for Nursing. National League for Nursing, New York, pp. 97–105.

Tong, A., Sainsbury, P., Craig, J., 2007. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. International Journal for Quality in Health Care 19 (6), 349–357.

Wolcott, H., 1994. Transforming Qualitative Data: Description, Analysis and Interpretation. Sage, Thousand Oaks, CA.