International Emergency Nursing (2008) 16, 48–52

www.elsevierhealth.com/journals/aaen

Clinical research 4: Qualitative data collectionand analysis q

Ruth Endacott RN, DipN(Lond), MA, PhD (Professor of Clinical Nursing) *

University of Plymouth, Faculty of Health and Social Work, Earl Richards Road North, Exeter EX2 6AS,United KingdomLa Trobe University, PO Box 199, Bendigo, Victoria 3552, Australia

Summary This six-part research series is aimed at clinicians who wish to developresearch skills, or who have a particular clinical problem that they think could beaddressed through research. The series aims to provide insight into the decisionsthat researchers make in the course of their work, and to also provide a foundationfor decisions that nurses may make in applying the findings of a study to practice intheir own Unit or Department. The series emphasise the practical issues encoun-tered when undertaking research in critical care settings; readers are encouragedto source research methodology textbooks for more detailed guidance on specificaspects of the research process.

�c 2006 Published by Elsevier Ltd.

KEYWORDSMethodology;Observation;Qualitative research;Data analysis

0d

Cp

Introduction

Qualitative research is often referred to as ‘realworld’ research as it commonly involves undertak-ing observation and interviews in the location ofthe participants. There are two dominant para-digms in qualitative research: the interpretive par-adigm (ethnography, phenomenology, grounded

965-2302/$ - see front matter �c 2006 Published by Elsevier Ltd.oi:10.1016/j.aaen.2006.12.002

q This article was originally published in Intensive and Criticalare Nursing 2005 21(2) 123–127. The article is republished withermission from Intensive and Critical Care Nursing.* Tel.: +44 01392 475155; fax: +44 01392 475151.E-mail address: ruth.endacott@plymouth.ac.uk.

theory) and the critical paradigm (feminist or ac-tion research). The key features of these five areprovided in Table 1. Qualitative description is asixth approach that is becoming increasingly popu-lar when none of these five is suited to the researchstudy (Sandelowski, 2000); an example of this ap-proach is seen in Europe and Tyni-Lenne (2004)study of the male experience of heart failure. Itis increasingly common for qualitative and quanti-tative approaches to be used in the same study;useful examples are found in Evangelista et al.(2003) and Fitzsimmons et al. (2003). This paper fo-cuses on the common elements of qualitative re-search and issues to be addressed in clinicalqualitative research.

Clinical research 4: Qualitative data collection and analysis 49

There are three central methods used in qualita-tive research: interview, observation and documen-tary analysis. Interviews are commonly semi-structured– open questions that are predeterminedbut allow the participant or researcher to diverge topursue an idea inmore detail – or in-depth/unstruc-tured – one or two global questions allowing theinterview to progress in the direction dictated bythe participant. With this latter approach, subse-quent questions tend to probe and seek clarificationon details. Another approach for interviewing is thecritical incident technique; this is used when the re-searcher seeks to focus on the participant’s memo-ries of one (or more) specific incident.

Observation has been described as the ‘closestto a gold standard’ in qualitative research (Murphyand Dingwall, 1998) as it enables comparison be-tween stated and actual actions and often identi-fies issues that neither the researcher or the‘researched’ were overtly aware of. Observationcan be carried out as exploratory work to identifythemes for further exploration in a fuller study;alternatively observation and interviews can beundertaken simultaneously or in a sequential man-ner (for example Currey et al., 2003).

With the exception of historical research, docu-mentary analysis is rarely used as the sole methodof data collection. It will more commonly be usedto add complementary data (e.g. the analysis ofnursing records to identify differences between ob-served care and recorded care). The documentbeing analysed will not have been recorded as re-search data hence three factors have to be ex-plored: the context of the documentation, the

Table 1 Key features of qualitative research methodolog

Focus

Ethnography Understanding cultural rulesObserver role includes some particObservation is a key data collectio

Phenomenology Exploring a phenomenon in depthMay include ‘lived experience’

Grounded theory Developing theory inductively fromRelies on iterative process of dataGenerating hypotheses which are

Action research Attempts to bring about change inAttempts to influence the real wo

Feminist research Non-threatening and non-hierarchiTwo-fold goal:� To raise consciousness of wome� To empower women as a result

purpose for which it was written and the circum-stances under which it was written (for example,what is recorded? what is omitted? what is takenfor granted?).

Data collection activities

There are a number of key activities common toqualitative data collection (Creswell, 1998, p.110). Key issues in relation to critical care settingsare addressed below.

Gaining access/achieving entree

Prior to formal processes for access via the hospitalR&D committee and ethics committees, permissionmust be obtained from the nurse manager. In addi-tion, for observation studies, permission needs tobe refreshed for each shift of observation, withthe shift leader. A helpful tool has been developedby Scholes (1996) to assist in this process. Keyinformation to be provided to the manager in seek-ing access is detailed at Table 2.

Three ethical principles underpin data collec-tion: autonomy, anonymity/confidentiality and in-formed consent (Endacott, 2004). In critical caresettings the latter poses particular dilemmas forresearchers and ethics committees (Grap and Mun-ro, 2003); for a lively debate about the impact ofrequiring patient consent on recruitment for a sepsisstudy, see Annane et al. (2004) and Lemaire (2004).A particular issue to address with documentary

ies (after Creswell, 1998; Hek et al., 2002)

ipation in the situationn method, with informal and formal interviewing

the datacollection and data analysis

then tested through data collection

practice during the researchrld through a spiral process of change and evaluation

cal relationship between researcher and participants

n’s issuesof the research

Table 2 Information to be provided to the managerwhen seeking access for observation studies

� Why was the site chosen?� What will be done at the site?� Will the researcher’s presence be disruptive?� How will the results be published?� Will the site gain from the study?

50 R. Endacott

analysis is the steps necessary to anonymise re-cords, for example removal of staff names/signa-tures. Access to photocopying facilities is apractical consideration, particularly when the doc-ument is part of patient records for the currentcare episode. Ethics committees commonly requireboth of these processes to be made explicit.

Identifying the sample

Sampling in qualitative studies is based on theoryand decisions about who to sample will either bemade prior to data collection (purposive sam-pling) or during data collection, as the theoryemerges (theoretical sampling). A useful exampleof theoretical sampling is found in Ball and Cox(2003). Sample size in qualitative research isbased on saturation, or ceasing data collectionwhen data categories have been exhausted. The-oretical sampling and sampling based on satura-tion can raise issues for ethics committees. Anindication of the ‘population’ of participantswho will be sampled (e.g. junior or senior nursingstaff) with likely numbers should be given. If theemerging theory requires a researcher to ap-proach a different group of participants, forexample relatives, the ethics approval will needto reflect this change.

Theoretical (or theory-based) sampling is arequirement for grounded theory studies, wherethe goal is to find a homogenous sample. Phenom-enology requires the seeking out of individuals whohave experienced the ‘phenomenon’ (criterionbased sampling) and ethnography requires the cul-tural group to be accessed (representative basedsampling) (Creswell, 1998).

Collecting and recording data

Four key decisions underpin data collection:

1. degree of structure for the interviews and/orobservation (is the agenda being driven by theresearcher or the participants?);

2. timing (are you wishing to observe, or interviewfollowing, key events such as weaning fromventilation?);

3. number of participants (e.g. individual or groupinterviews);

4. location (whose territory?).

All decisions must be made in the context of theaims of the study. Interview data can be collectedthrough individual or focus group interviews. Theuse of focus group interviews has to be driven bythe additional synergy (data!) to be gained throughthe group process, and this synergy should be dem-onstrated in the presentation of data. The use ofgroup interviews predominantly for the research-er’s convenience is considered bad practice andcan result in poor quality data.

Recording of observation data can be achievedusing video or audio taping, in addition to fieldnotes. Observation data also commonly involvesrecording a ‘map’ of the setting in narrative or gra-phic form, for example ICU layout, staffing andworkload during the observation period. This is par-ticularly important for ethnographic studies.

Resolving field issues

It is essential to outline from the start the role tobe taken by the researcher. This is of particularimportance in observation studies where the roleis typically a blend of observation and participa-tion. Qualitative researchers also have to be sen-sitive (reflexive) to the ways in which theresearcher and the research process have shapedthe data, including the role of prior assumptionsand experience (Mays and Pope, 2000). Curreyet al. (2003) provide a useful discussion of the is-sues involved in observation research in intensivecare settings.

Validity and reliability

In general, the rules applied to quantitative studiesin order to achieve validity and reliability are notappropriate for qualitative research. Guba and Lin-coln’s (1989) three criteria for qualitative studiesare most commonly applied:

1. credibility – return data to the subjects for ver-ification (e.g. return transcripts or analysis toindividual participants, or discuss early analysisat focus group interviews);

2. transferability – of the theory, rather than‘sample to population’ generalisability;

Clinical research 4: Qualitative data collection and analysis 51

3. dependability – auditability, use of a ‘decisiontrail’. Researchers often use a narrative diaryto record theoretical, methodological and ana-lytical choices.

Inter-coder reliability is also used to establishdependability of the data analysis. The most com-mon approach requires two researchers to codethe same transcript and then compare coding. In-ter-coder reliability will not be established if aconsensus approach is taken as this may suppress,rather than illuminate, the inter-coder differ-ences. Another pitfall to avoid is the use of onlythose data on which there is inter-coder agree-ment; this results in data being selected becausethey suit the research method, rather than be-cause they represent the phenomenon/situationbeing studied.

The following strategies can also help to im-prove/verify the trustworthiness of the data:

– triangulation, using multiple data sources,– using credible informants,– prolonged and persistent observation (can

increase credibility of the researcher),– continuous data analysis, enabling any con-

tradictions to be clarified ‘in the field’,– searching for conflicting evidence (negative

cases),– observing at different times of the day,– acknowledging and documenting the impact of

the researcher on the situation.

Data analysis

A central tenet of qualitative research is early dataanalysis whilst data collection continues. This over-lapping of the two stages allows the analysis toguide subsequent data collection, either throughtheoretical sampling (essential for a grounded the-ory study) or through amending interview/observa-tion instruments to ensure emerging areas areexplored. For example, during an interview withrelatives exploring stressful events in ICU a specificsituation may be raised by the relatives. Subse-quent data collection may target participants whowould have witnessed a similar event (theoreticalsampling) or explore such a scenario with all partic-ipants. This interplay between data collection andanalysis is also essential to identify the point atwhich data saturation occurs.

Qualitative data analysis is choreographed (Milesand Huberman, 1994) not lifted off the shelf. Somemethodologies have specific processes for dataanalysis (see Ball and Cox, 2003 for an example of

grounded theory data analysis and Arslanian-Engo-ren and Scott, 2003 for one approach used in phe-nomenology). Common to all approaches is theprocess of developing codes and categories. Gen-eral categories (or themes) may be developed fromthe data, which are then broken down into moreexplicit codes. Alternatively, line-by-line analysisof transcripts is used to develop codes, which arethen built up into categories/themes. A third ap-proach is to use a pre-determined framework forcodes, for example, Carper’s (1978) four ways ofknowing or Benner’s (1984) novice to expert frame-work. A number of software packages are availableto assist in data analysis (e.g. N-Vivo QSR Interna-tional Pty Ltd, Melbourne: Victoria); these all havea specific application but commonly work on a’node and tree’ approach. It is essential to use acoding system for reporting data (e.g. ParticipantA/Interview 1/date). For qualitative studies thisis the equivalent of identifying statistical signifi-cance as it enables the reader to judge whetherthe data have been used selectively to pursue a lineof thinking. If 20 participants were involved in astudy but the paper only reports data from 8, thisdemonstrates a level of selection bias.

The findings from ethnographic research arepresented as a narrative (an ethnography), aug-mented by tables, figures or sketches; in ICU thismight include a floor map of the layout or organisa-tional map to identify how many staff are at differ-ent levels/in different roles.

Phenomenological research is presented as anarration of the essence of the experience or phe-nomenon and grounded theory results in a visualmodel or theory. For a fuller description, see Cre-swell (1998).

Conclusions

A number of different approaches can be taken inqualitative research and the roles of researcherand participant are much less prescribed than inquantitative studies. The choices made must beappropriate for the skills of the researcher andthe timeframe and resources available for the re-search. However, the usefulness of qualitativestudies can only be judged if such decisions aremade explicit (for example, what ‘theory’ under-pinned the theoretical sampling? Why was observa-tion only undertaken during the day?). Qualitativestudies have much to offer both in terms of guidingpractice (through providing rich insights into theexperiences of patients, relatives and staff in crit-ical care) and identifying areas/variables to be ex-plored in future studies.

52 R. Endacott

References

Annane, D., Outin, H., Fisch, C., Bellissant, E., 2004. The effectof waiving consent on enrolment in a sepsis trial. Intens. CareMed. 30 (2), 321–324.

Arslanian-Engoren, C., Scott, L.D., 2003. The lived experienceof survivors of prolonged mechanical ventilation. Heart Lung32 (5), 328–334.

Ball, C., Cox, C.I., 2003. Part one: restoring patients to health-outcomes and indicators of advanced nursing practice inadult critical care. Int. J. Nurs. Pract. 9, 356–367.

Benner, P., 1984. FromNovice to Expert: Excellence and Power inClinicalNursingPractice.Addison-Wesley,RedwoodCity,USA.

Carper, B.A., 1978. Fundamental patterns of knowing in nursing.Adv. Nurs. Sci. 1, 13–23.

Creswell, J.W., 1998. Qualitative Inquiry and Research Design:Choosing Among Five Traditions. Sage, Thousand Oaks.

Currey, J., Botti, M., Browne, J., 2003. Hemodynamic teamdecision-making in the cardiac surgical intensive care con-text. Heart Lung 32 (3), 181–189.

Endacott, R., 2004. Clinical research 2: legal and ethical issuesin research. Intens. Crit. Care Nurs. 20 (5), 313–315.

Europe, E., Tyni-Lenne, R., 2004. Qualitative analysis of themaleexperience of heart failure. Heart Lung 33 (4), 227–234.

Evangelista, L.S., Doering, L., Dracup, K., 2003. Meaning and lifepurpose: the perspectives of post-transplant women. HeartLung 32 (4), 250–257.

Fitzsimmons, D., Parahoo, K., Richardson, S.G., Stringer, M.,2003. Patient anxiety while on a waiting list for coronaryartery bypass surgery: a qualitative and quantitative analysis.Heart Lung 32 (1), 23–31.

Grap, M.J., Munro, C.L., 2003. Subject recruitment in criticalcare research: a complex task in a complex environment.Heart Lung 32 (3), 162–168.

Guba, Y.S., Lincoln, E.G., 1989. Fourth Generation Evaluation.Sage, California.

Hek, G., Judd, M., Moule, P., 2002. Making Sense of Research:an Introduction for Health and Social Care Practitioners.Continuum, London.

Lemaire, F., 2004. A waiver of consent for intensive careresearch. Intens. Care Med. 30 (2), 177–179.

Mays, N., Pope, C., 2000. Assessing quality in qualitativeresearch. Br. Med. J. 320, 50–52.

Miles, M., Huberman, A., 1994. Early Steps in Analysis. In:Qualitative Data Analysis, second ed. Sage Publications,California, pp. 50–89.

Murphy, E., Dingwall, R., 1998. Qualitative methods in healthservices research. In: Black, N. et al. (Eds.), Health ServicesResearch Methods: A Guide to Best Practice. BMJ Publishing,London.

Sandelowski, M., 2000. What happened to qualitative descrip-tion. Res. Nurs. Health 23 (4), 334–340.

Scholes, J., 1996. Therapeutic use of self: how the critical carenurse uses self to the patient’s therapeutic benefit. Nurs.Crit. Care 1 (2), 60–66.