Vous êtes sur la page 1sur 4

RESEARCH METHODS

& REPORTING
Three techniques for integrating data in
mixed methods studies
Alicia O’Cathain,1 Elizabeth Murphy,2 Jon Nicholl1

1
Medical Care Research Unit, Techniques designed to combine the results SUMMARY POINTS
School of Health and Related
Research, University of Sheffield, of qualitative and quantitative studies can Health researchers are increasingly using designs which
Sheffield S1 4DA, UK combine qualitative and quantitative methods
2
University of Leicester, Leicester,
provide researchers with more knowledge However, there is often lack of integration between methods
UK than separate analysis Three techniques are described that can help researchers
Correspondence to: A O’Cathain
to integrate data from different components of a study:
a.ocathain@sheffield.ac.uk
Health researchers are increasingly using designs that triangulation protocol, following a thread, and the mixed
Accepted: 8 June 2010
combine qualitative and quantitative methods, and this methods matrix
Cite this as: BMJ 2010;341:c4587 is often called mixed methods research.1 Integration— Use of these methods will allow researchers to learn more
doi: 10.1136/bmj.c4587 the interaction or conversation between the qualitative from the information they have collected
and quantitative components of a study—is an impor‑
tant aspect of mixed methods research, and, indeed, is
bmj.com archive essential to some definitions.2 Recent empirical studies of triangulation. The term triangulation can be confusing
Previous articles in this mixed methods research in health show, however, a lack because it has two meanings.10 It can be used to describe
series of integration between components,3  4 which limits the corroboration between two sets of findings or to describe
ЖЖAssessing equity amount of knowledge that these types of studies gener‑ a process of studying a problem using different methods
in systematic ate. Without integration, the knowledge yield is equiva‑ to gain a more complete picture. The latter meaning is
reviews: realising the lent to that from a qualitative study and a quantitative commonly used in mixed methods research and is the
recommendations of the study undertaken independently, rather than achieving meaning used here.
Commission on Social a “whole greater than the sum of the parts.”5 The process of triangulating findings from differ‑
Determinants of Health Barriers to integration have been identified in both ent methods takes place at the interpretation stage of
health and social research.6  7 One barrier is the absence of a study when both data sets have been analysed sepa‑
(BMJ 2010;341:c4739)
formal education in mixed methods research. Fortunately, rately (­figure). Several techniques have been described
ЖЖMeta-analysis of
literature is rapidly expanding to fill this educational gap, for ­triangulating findings. They require researchers to list
individual participant data:
including descriptions of how to integrate data and find‑ the findings from each component of a study on the same
rationale, conduct, and ings from qualitative and quantitative methods.8  9 In this page and consider where findings from each method
reporting article we outline three techniques that may help health agree (convergence), offer complementary information on
(BMJ 2010;340:c221) researchers to integrate data or findings in their mixed the same issue (complementarity), or appear to contra‑
ЖЖEvaluating policy and methods studies and show how these might enhance dict each other (discrepancy or ­dissonance).11‑13 ­Explicitly
service interventions: knowledge generated from this approach. looking for disagreements between findings from differ‑
framework to ent methods is an important part of this ­process. Disa‑
guide selection and Triangulation protocol greement is not a sign that something is wrong with a
interpretation of study Researchers will often use qualitative and quantita‑ study. Exploration of any apparent ­“inter-method dis‑
end points tive methods to examine different aspects of an overall
(BMJ 2010;341:c4413) research question. For example, they might use a ran‑
Design
ЖЖImportance of domised controlled trial to assess the effectiveness of a
accurately identifying healthcare intervention and semistructured interviews
Data collection
disease in studies using with patients and health professionals to consider the
way in which the intervention was used in the real world.
electronic health records Mixed methods
Analysis Following a thread
Alternatively, they might use a survey of service users to matrix
(BMJ 2010;341:c4226)
measure satisfaction with a service and focus groups to
ЖЖTaking healthcare Triangulation
explore views of care in more depth. Data are collected protocol
Interpretation
interventions from trial to and analysed separately for each component to produce
practice two sets of findings. Researchers will then attempt to Point of application for three techniques for integrating data in
(BMJ 2010;341:c3852) combine these findings, sometimes calling this process mixed methods research

BMJ | 27 NOVEMBER 2010 | VOLUME 341 1147


RESEARCH METHODS & REPORTING

AN EXAMPLE OF FOLLOWING A THREAD19


Adamson and colleagues explored the effect of patient series of symptoms presented in standardised vignettes as
views on the appropriate use of services and help seeking were respondents who thought that services were not used
using a survey of people registered at a general practice inappropriately. The researchers then followed the thread
and semistructured interviews. The qualitative (22 back to the interview data to help interpret this finding.
interviews) and quantitative components (survey with 911 After further analysis of the interview data the researchers
respondents) took place concurrently. understood that people considered the help seeking of
The researchers describe what they call an iterative or other people to be inappropriate, rather than their own.
cyclical approach to analysis. Firstly, the preliminary findings They also noted that feeling anxious about symptoms was
from the interviews generated a hypothesis for testing in considered to be a good justification for seeking care. The
the survey data. A key theme from the interviews concerned researchers followed this thread back into the survey data
the self rationing of services as a responsible way of using and tested whether anxiety levels about the symptoms in the
scarce health care. This theme was then explored in the standardised vignettes predicted help seeking behaviour.
survey data by testing the hypothesis that people’s views This second hypothesis was supported by the survey data.
of the appropriate use of services would explain their help Following a thread led the researchers to conclude that
seeking behaviour. However, there was no support for this patients who seek health care for seemingly minor problems
hypothesis in the quantitative analysis because the half of have exceeded their thresholds for the trade-off between not
survey respondents who felt that health services were used using services inappropriately and any anxiety caused by their
inappropriately were as likely to report help seeking for a symptoms.

crepancy” may lead to a better understanding of the using the new service rated the convenience of visiting
research question,14 and a range of approaches have the hospital as poor than those using usual care. The
been used within health services research to explore interviews supported this concern about the new serv‑
inter-method discrepancy.15 ice, but also identified that the weight carers gave to this
The most detailed description of how to carry out trian‑ concern was low in the context of their family member’s
gulation is the triangulation protocol,11 which although life being saved.
developed for multiple qualitative methods, is relevant Morgan describes this move as the “third effort”
to mixed methods studies. This technique involves pro‑ because it occurs after analysis of the qualitative and the
ducing a “convergence coding matrix” to display find‑ quantitative components.17 It requires time and energy
ings emerging from each component of a study on the that must be planned into the study timetable. It is also
same page. This is followed by consideration of where useful to consider who will carry out the integration
there is agreement, partial agreement, silence, or disso‑ process. Farmer and colleagues require two researchers
nance between findings from different components. This to work together during triangulation, which can be par‑
technique for triangulation is the only one to include ticularly important in mixed methods studies if different
silence—where a theme or finding arises from one data researchers take responsibility for the qualitative and
set and not another. Silence might be expected because quantitative components.11
of the strengths of different methods to examine differ‑
ent aspects of a phenomenon, but surprise silences might Following a thread
also arise that help to increase understanding or lead to Moran-Ellis and colleagues describe a different technique
further investigations. for integrating the findings from the qualitative and quan‑
The triangulation protocol moves researchers from titative components of a study, called following a thread.18
thinking about the findings related to each method, to They state that this takes place at the analysis stage of the
what Farmer and colleagues call meta-themes that cut research process (figure). It begins with an initial analysis
across the findings from different methods.11 They show of each component to identify key themes and questions
a worked example of triangulation protocol, but we could requiring further exploration. Then the researchers select
find no other published example. However, similar prin‑ a question or theme from one component and follow it
ciples were used in an iterative mixed methods study to across the other components—they call this the thread.
understand patient and carer satisfaction with a new The authors do not specify steps in this technique but
primary angioplasty service.16 Researchers conducted offer a visual model for working between datasets. An
semistructured interviews with 16 users and carers to approach similar to this has been undertaken in health
explore their experiences and views of the new service. services research, although the researchers did not label
These were used to develop a questionnaire for a survey of it as such, probably because the technique has not been
595 patients (and 418 of their carers) receiving either the used frequently in the literature (box).
new service or usual care. Finally, 17 of the patients who
expressed dissatisfaction with aftercare and rehabilita‑ Mixed methods matrix
tion were followed up to explore this further in semistruc‑ A unique aspect of some mixed methods studies is the
tured interviews. A shift of thinking to meta-themes led availability of both qualitative and quantitative data on
the researchers away from reporting the findings from the the same cases. Data from the qualitative and quantita‑
interviews, survey, and follow-up interviews sequentially tive components can be integrated at the analysis stage
to consider the meta-themes of speed and efficiency, con‑ of a mixed methods study (figure). For example, in-depth
venience of care, and discharge and after care. The survey interviews might be carried out with a sample of survey
identified that a higher percentage of carers of patients respondents, creating a subset of cases for which there is

1148 BMJ | 27 NOVEMBER 2010 | VOLUME 341


RESEARCH METHODS & REPORTING

Example of a mixed methods matrix for a study exploring the relationship between types of teams and integration between qualitative and quantitative components
of studies*22
Evidence of Types of publications
Study integration in report† emerging‡ Qualitative expertise on the team Team working Respect for team members
1 1 5 Yes Close and friendly Yes
2 1 5 Yes Single researcher Yes
3 1 4 No senior qualitative expertise on Integrated team. The qualitative and Yes
team but project researcher worked quantitative researchers were in same
hard at it department
4 2 4 Yes. There was also expertise Integrated team. The lead researcher Initially some team members did not
developing in mixed methods worked closely with qualitative and respect the qualitative research but
research quantitative researchers learnt to as the study progressed
5 2 4 Reported as no problem even though The junior researcher delivered both Lead researcher did not respect the
junior staff had no expertise the qualitative and the quantitative qualitative research but other senior
components. The team was team members did
geographically close
6 2 4 Yes, including mixed methods Worked well together. Lead researcher Yes
expertise worked closely with qualitative and
quantitative researchers
*Shows the first six cases and a selection of themes from the full matrix. The content of some of the original cells has been changed to increase comprehension and protect confidentiality.
† 1=yes, 2=yes but more possible, 3=no.
‡1=none, 2=only qualitative, 3=only quantitative, 4=both published separately, 5=mixed methods article.

both a completed questionnaire and a transcript. Cases ordering of rows helped us to see patterns across rows.
may be individuals, groups, organisations, or geographi‑ The next columns were themes from the qualitative
cal areas.9 All the data collected on a single case can be interview with a researcher from that project. For exam‑
studied together, focusing attention on cases, rather ple, the first theme was about the expertise in qualitative
than variables or themes, within a study. The data can research within the team and whether the interviewee
be examined in detail for each case—for example, com‑ reported this as adequate for the study. The matrix was
paring people’s responses to a questionnaire with their then used in the context of the qualitative analysis to
interview transcript. Alternatively, data on each case can explore the issues that affected integration. In particu‑
be summarised and displayed in a matrix8  9  20 along the lar, it helped to identify negative cases (when someone in
lines of Miles and Huberman’s meta-matrix.21 Within a the analysis doesn’t fit with the conclusions the analysis
mixed methods matrix, the rows represent the cases for is coming to) within the qualitative analysis to facilitate
which there is both qualitative and quantitative data, and understanding. Interviewees reported the need for expe‑
the columns display different data collected on each case. rienced qualitative researchers on mixed methods studies
This allows researchers to pay attention to surprises and to ensure that the qualitative component was published,
paradoxes between types of data on a single case and yet two cases showed that this was neither necessary nor
then look for patterns across all cases20 in a qualitative sufficient. This pushed us to explore other factors in a
cross case analysis.21 research team that helped generate outputs, and inte‑
We used a mixed methods matrix to study the rela‑ grated outputs, from a mixed methods study.
tion between types of team working and the extent of Themes from a qualitative study can be summarised
integration in mixed methods studies in health services to the point where they are coded into quantitative data.
research (table).22 Quantitative data were extracted from In the matrix (table), the interviewee’s perception of
the proposals, reports, and peer reviewed publications the adequacy of qualitative expertise on the team could
of 75 mixed methods studies, and these were analysed to have been coded as adequate=1 or not=2. This is called
describe the proportion of studies with integrated outputs “quantitising” of qualitative data23; coded data can then
such as mixed methods journal articles. Two key variables be analysed with data from the quantitative component.
in the quantitative component were whether the study This technique has been used to great effect in health‑
was assessed as attempting to integrate qualitative or care research to identify the discrepancy between health
quantitative data or findings and the type of publications improvement assessed using quantitative measures
produced. We conducted qualitative interviews with 20 and with in-depth interviews in a randomised controlled
researchers who had worked on some of these studies trial.24
to explore how mixed methods research was practised,
including how the team worked together. Conclusion
The shared cases between the qualitative and quantita‑ We have presented three techniques for integration
tive components were 21 mixed methods studies (because in mixed methods research in the hope that they will
one interviewee had worked on two studies in the quan‑ inspire researchers to explore what can be learnt from
titative component). A matrix was formed with each of bringing together data from the qualitative and quantita‑
the 21 studies as a row. The first column of the matrix tive components of their studies. Using these techniques
contained the study identification, the second ­column may give the process of integration credibility rather than
indicated whether integration had occurred in that leaving researchers feeling that they have “made things
project, and the third column the score for integration up.” It may also encourage researchers to describe their
of publications emerging from the study. The rows were approaches to integration, allowing them to be transpar‑
then ordered to show the most integrated cases first. This ent and helping them to develop, critique, and improve

BMJ | 27 NOVEMBER 2010 | VOLUME 341 1149


RESEARCH METHODS & REPORTING

on these techniques. Most importantly, we believe it may 6 O’Cathain A, Nicholl J, Murphy E. Structural issues affecting mixed
methods studies in health research: a qualitative study. BMC Med Res
help researchers to generate further understanding from Methodol 2009;9:82.
their research. 7 Bryman A. Barriers to integrating quantitative and qualitative
We have presented integration as unproblematic, but it research. J Mix Methods Res 2007;1:8-22.
8 Creswell JW, Plano-Clark V. Designing and conducting mixed methods
is not. It may be easier for single researchers to use these research. Sage, 2007.
techniques than a large research team. Large teams will 9 Bazeley P. Analysing mixed methods data. In: Andrew S, Halcomb EJ,
eds. Mixed methods research for nursing and the health sciences.
need to pay attention to team dynamics, considering who Wiley-Blackwell, 2009:84-118.
will take responsibility for integration and who will be 10 Sandelowski M. Triangles and crystals: on the geometry of qualitative
research. Res Nurs Health 1995;18:569-74.
taking part in the process. In addition, we have taken a 11 Farmer T, Robinson K, Elliott SJ, Eyles J. Developing and implementing
technical stance here rather than paying attention to dif‑ a triangulation protocol for qualitative health research. Qual Health
ferent philosophical beliefs that may shape approaches Res 2006;16:377-94.
12 Foster RL. Addressing the epistemologic and practical issues in
to integration. We consider that these techniques would multimethod research: a procedure for conceptual triangulation. Adv
work in the context of a pragmatic or subtle realist stance Nurs Sci 1997;20:1-12.
13 Erzerberger C, Prein G. Triangulation: validity and empirically based
adopted by some mixed methods researchers.25 Finally, it hypothesis construction. Qual Quant 1997;31:141-54.
is important to remember that these techniques are aids 14 Fielding NG, Fielding JL. Linking data. Sage, 1986.
to integration and are helpful only when applied with 15 Moffatt S, White M, Mackintosh J, Howel D. Using quantitative and
qualitative data in health services research—what happens when
expertise. mixed method findings conflict? BMC Health Serv Res 2006;6:28.
Funding: Medical Research Council grant reference G106/1116. 16 Sampson FC, O’Cathain A, Goodacre S. Is primary angioplasty an
Competing interests: All authors have completed the unified competing acceptable alternative to thrombolysis? Quantitative and qualitative
study of patient and carer satisfaction. Health Expectations
interest form at www.icmje.org/coi_disclosure.pdf (available on
(forthcoming).
request from the corresponding author) and declare financial support 17 Morgan DL. Practical strategies for combining qualitative and
for the submitted work from the Medical Research Council; no financial quantitative methods: applications to health research. Qual Health
relationships with commercial entities that might have an interest in the Res 1998;8:362-76.
submitted work; no spouses, partners, or children with relationships with 18 Moran-Ellis J, Alexander VD, Cronin A, Dickinson M, Fielding J,
commercial entities that might have an interest in the submitted work; and Sleney J, et al. Triangulation and integration: processes, claims and
no non-financial interests that may be relevant to the submitted work. implications. Qualitative Research 2006;6:45-59.
19 Adamson J, Ben-Shlomo Y, Chaturvedi N, Donovan J. Exploring the
Contributors: AOC wrote the paper. JN and EM contributed to drafts and all impact of patient views on ‘appropriate’ use of services and help
authors agreed the final version. AOC is guarantor. seeking: a mixed method study. Br J Gen Pract 2009;59:496-502.
Provenance and peer review: Not commissioned; externally peer reviewed. 20 Wendler MC. Triangulation using a meta-matrix. J Adv Nurs
1 Lingard L, Albert M, Levinson W. Grounded theory, mixed methods 2001;35:521-5.
and action research. BMJ 2008;337:a567. 21 Miles M, Huberman A. Qualitative data analysis: an expanded
2 Creswell JW, Fetters MD, Ivankova NV. Designing a mixed methods sourcebook. Sage, 1994.
study in primary care. Ann Fam Med 2004;2:7-12. 22 O’Cathain A, Murphy E, Nicholl J. Multidisciplinary, interdisciplinary or
dysfunctional? Team working in mixed methods research. Qual Health
3 Lewin S, Glenton C, Oxman AD. Use of qualitative methods alongside
Res 2008;18:1574-85.
randomised controlled trials of complex healthcare interventions:
23 Sandelowski M. Combining qualitative and quantitative sampling,
methodological study. BMJ 2009;339:b3496. data collection, and analysis techniques in mixed-method studies.
4 O’Cathain A, Murphy E, Nicholl J. Integration and publications as Res Nurs Health 2000;23:246-55.
indicators of ‘yield’ from mixed methods studies. J Mix Methods Res 24 Campbell R, Quilty B, Dieppe P. Discrepancies between patients’
2007;1:147-63. assessments of outcome: qualitative study nested within a
5 Barbour RS. The case for combining qualitative and quantitative randomised controlled trial. BMJ 2003;326:252-3.
approaches in health services research. J Health Serv Res Policy 25 Mays N, Pope C. Assessing quality in qualitative research. BMJ
1999;4:39-43. 2000;320:50-2.

Calculate the carbon footprint of your hospital


The NHS is a big polluter, and this shouldn’t be forgotten. However, 5. Information on other sources of emissions relevant to your hospital. This
the priority of environmental issues in the public services fluctuates. is more difficult but relates to indirect causes of emissions such as from
Calculating the carbon footprint of your organisation is relatively your supply chain, waste disposal, water and sewerage, etc.
straightforward, and the Carbon Trust has produced an online calculator When the calculation is completed (it is roughly 20 000 tonnes of CO2 per
to enable you to do so (www.carbontrust.co.uk/cut-carbon-reduce-costs/ year in Brighton) the data need to be presented to the hospital management
calculate/carbon-footprinting/Pages/carbon-footprinting.aspx). board to signal the importance of the issue and to help deliver a strategy for
You need to gather the following information, which should be readily change. Clinical leadership has a big part to play, and nationally nephrologists
accessible (the main problem is knowing who is the right person to ask are taking a key role
within a hospital trust): (www.greenerhealthcare.org/green-nephrology-programme).
1. Annual electricity use in kWh (available from the Estates & Facilities Examples of practical improvements to hospital infrastructure include
department). The Royal Sussex County Hospital in Brighton spends reducing solid waste, increasing recycling, installing a fuel cell, upgrading
£2.2m pounds a year on electricity, for example the lighting, and making laundry, heating, and cooling systems more
2. Vehicle use as average annual mileage environmentally friendly and efficient. How about a solar powered dialysis
(contact the Transport department) unit? Hospitals should be at the forefront of helping to keep our environment
3. Your energy use on site, such as gas, oil, steam healthy as opposed to the current toxic situation.
(Estates again) Paul Grant clinical leadership fellow, Royal Sussex County Hospital, Brighton drpaul.grant@
doctors.org.uk
4. Employee travel by road, rail, and plane
(Human Resources should be able to provide the total annual average Sarah Lou Bailey clinical lecturer in global health, University of Sussex, Brighton
and the costs incurred) Cite this as: BMJ 2010;341:c2366

1150 BMJ | 27 NOVEMBER 2010 | VOLUME 341

Vous aimerez peut-être aussi